Max Tegmark: AI and Physics
AI 与机器学习生物与进化音乐与艺术心理与人性技术与编程
📋 章节目录
暂无章节信息
🔑 关键词
donlearningmachinegoinghumanbetterphysicshumanssaidintelligencewholefutureneuralcoursedoingabledidnsciencestuffsystems
💬 精彩语录
暂无语录
🎙️ 完整对话（3986 条）
Lex Fridman (00:00.000)
The following is a conversation with Max Tegmark,
以下是与 Max Tegmark 的对话，
Lex Fridman (00:02.840)
his second time on the podcast.
他第二次参加播客。
Lex Fridman (00:04.760)
In fact, the previous conversation
其实之前的对话
Lex Fridman (00:07.120)
was episode number one of this very podcast.
是这个播客的第一集。
Lex Fridman (00:10.960)
He is a physicist and artificial intelligence researcher
他是一位物理学家和人工智能研究员
Max Tegmark (00:14.800)
at MIT, cofounder of the Future of Life Institute,
麻省理工学院未来生命研究所联合创始人
Lex Fridman (00:18.840)
and author of Life 3.0,
《Life 3.0》的作者，
Max Tegmark (00:21.360)
Being Human in the Age of Artificial Intelligence.
人工智能时代的人类。
Lex Fridman (00:24.560)
He's also the head of a bunch of other huge,
他也是其他一些大公司的负责人
Max Tegmark (00:27.120)
fascinating projects and has written
有趣的项目并写过
Lex Fridman (00:29.240)
a lot of different things
很多不同的事情
Max Tegmark (00:30.560)
that you should definitely check out.
你一定要检查一下。
Lex Fridman (00:32.080)
He has been one of the key humans
他是关键人物之一
Max Tegmark (00:34.480)
who has been outspoken about longterm existential risks
谁直言不讳地谈论长期存在的风险
Lex Fridman (00:37.400)
of AI and also its exciting possibilities
人工智能及其令人兴奋的可能性
Lex Fridman (00:40.500)
and solutions to real world problems.
以及现实世界问题的解决方案。
Lex Fridman (00:42.900)
Most recently at the intersection of AI and physics,
最近在人工智能和物理学的交叉点，
Lex Fridman (00:46.440)
and also in reengineering the algorithms
以及重新设计算法
Lex Fridman (00:50.000)
that divide us by controlling the information we see
通过控制我们看到的信息来分裂我们
Lex Fridman (00:53.200)
and thereby creating bubbles and all other kinds
从而产生气泡和所有其他类型
Lex Fridman (00:56.160)
of complex social phenomena that we see today.
Max Tegmark (00:59.640)
In general, he's one of the most passionate
Lex Fridman (01:01.440)
and brilliant people I have the fortune of knowing.
Max Tegmark (01:04.340)
I hope to talk to him many more times
Lex Fridman (01:06.180)
on this podcast in the future.
Max Tegmark (01:08.280)
Quick mention of our sponsors,
Lex Fridman (01:10.000)
The Jordan Harbinger Show,
Max Tegmark (01:12.160)
Four Sigmatic Mushroom Coffee,
Lex Fridman (01:14.360)
BetterHelp Online Therapy, and ExpressVPN.
Lex Fridman (01:18.480)
So the choices, wisdom, caffeine, sanity, or privacy.
Lex Fridman (01:23.560)
Choose wisely, my friends, and if you wish,
Max Tegmark (01:25.880)
click the sponsor links below to get a discount
Lex Fridman (01:28.360)
and to support this podcast.
Max Tegmark (01:30.560)
As a side note, let me say that much of the researchers
Lex Fridman (01:33.900)
in the machine learning
Lex Fridman (01:35.400)
and artificial intelligence communities
Lex Fridman (01:37.760)
do not spend much time thinking deeply
Max Tegmark (01:40.400)
about existential risks of AI.
Lex Fridman (01:42.720)
Because our current algorithms are seen as useful but dumb,
Max Tegmark (01:46.160)
it's difficult to imagine how they may become destructive
Lex Fridman (01:49.240)
to the fabric of human civilization
Max Tegmark (01:51.240)
in the foreseeable future.
Lex Fridman (01:53.040)
I understand this mindset, but it's very troublesome.
Max Tegmark (01:56.120)
To me, this is both a dangerous and uninspiring perspective,
Lex Fridman (02:00.480)
reminiscent of a lobster sitting in a pot of lukewarm water
Max Tegmark (02:03.980)
that a minute ago was cold.
Lex Fridman (02:06.160)
I feel a kinship with this lobster.
Max Tegmark (02:08.640)
I believe that already the algorithms
Lex Fridman (02:10.560)
that drive our interaction on social media
Max Tegmark (02:12.960)
have an intelligence and power
Lex Fridman (02:14.980)
that far outstrip the intelligence and power
Max Tegmark (02:17.360)
of any one human being.
Lex Fridman (02:19.220)
Now really is the time to think about this,
Max Tegmark (02:21.640)
to define the trajectory of the interplay
Lex Fridman (02:24.140)
of technology and human beings in our society.
Max Tegmark (02:26.940)
I think that the future of human civilization
Lex Fridman (02:29.680)
very well may be at stake over this very question
Max Tegmark (02:32.820)
of the role of artificial intelligence in our society.
Lex Fridman (02:36.240)
If you enjoy this thing, subscribe on YouTube,
Max Tegmark (02:38.160)
review it on Apple Podcasts, follow on Spotify,
Lex Fridman (02:40.960)
support on Patreon, or connect with me on Twitter
Max Tegmark (02:43.840)
at Lex Friedman.
Lex Fridman (02:45.260)
And now, here's my conversation with Max Tegmark.
Lex Fridman (02:49.880)
So people might not know this,
Lex Fridman (02:51.440)
but you were actually episode number one of this podcast
Max Tegmark (02:55.280)
just a couple of years ago, and now we're back.
Lex Fridman (02:59.280)
And it so happens that a lot of exciting things happened
Max Tegmark (03:02.280)
in both physics and artificial intelligence,
Lex Fridman (03:05.600)
both fields that you're super passionate about.
Max Tegmark (03:08.480)
Can we try to catch up to some of the exciting things
Lex Fridman (03:11.640)
happening in artificial intelligence,
Max Tegmark (03:14.080)
especially in the context of the way it's cracking,
Lex Fridman (03:17.340)
open the different problems of the sciences?
Max Tegmark (03:20.020)
Yeah, I'd love to, especially now as we start 2021 here,
Lex Fridman (03:24.520)
it's a really fun time to think about
Lex Fridman (03:26.280)
what were the biggest breakthroughs in AI,
Lex Fridman (03:29.560)
not the ones necessarily that media wrote about,
Lex Fridman (03:31.800)
but that really matter, and what does that mean
Lex Fridman (03:35.200)
for our ability to do better science?
Lex Fridman (03:37.440)
What does it mean for our ability
Lex Fridman (03:39.920)
to help people around the world?
Lex Fridman (03:43.160)
And what does it mean for new problems
Lex Fridman (03:46.440)
that they could cause if we're not smart enough
Lex Fridman (03:48.440)
to avoid them, so what do we learn basically from this?
Lex Fridman (03:51.880)
Yes, absolutely.
Lex Fridman (03:52.720)
So one of the amazing things you're a part of
Lex Fridman (03:54.960)
is the AI Institute for Artificial Intelligence
Lex Fridman (03:57.680)
and Fundamental Interactions.
Lex Fridman (04:00.160)
What's up with this institute?
Lex Fridman (04:02.280)
What are you working on?
Lex Fridman (04:03.680)
What are you thinking about?
Max Tegmark (04:05.000)
The idea is something I'm very on fire with,
Lex Fridman (04:09.080)
which is basically AI meets physics.
Lex Fridman (04:11.920)
And it's been almost five years now
Lex Fridman (04:15.360)
since I shifted my own MIT research
Max Tegmark (04:18.360)
from physics to machine learning.
Lex Fridman (04:20.680)
And in the beginning, I noticed that a lot of my colleagues,
Max Tegmark (04:22.880)
even though they were polite about it,
Lex Fridman (04:24.280)
were like kind of, what is Max doing?
Lex Fridman (04:27.760)
What is this weird stuff?
Lex Fridman (04:29.040)
He's lost his mind.
Lex Fridman (04:30.880)
But then gradually, I, together with some colleagues,
Lex Fridman (04:35.080)
were able to persuade more and more of the other professors
Max Tegmark (04:40.080)
in our physics department to get interested in this.
Lex Fridman (04:42.520)
And now we've got this amazing NSF Center,
Lex Fridman (04:46.320)
so 20 million bucks for the next five years, MIT,
Lex Fridman (04:50.000)
and a bunch of neighboring universities here also.
Lex Fridman (04:53.200)
And I noticed now those colleagues
Lex Fridman (04:55.280)
who were looking at me funny have stopped
Max Tegmark (04:57.040)
asking what the point is of this,
Lex Fridman (05:00.320)
because it's becoming more clear.
Lex Fridman (05:02.400)
And I really believe that, of course,
Lex Fridman (05:05.560)
AI can help physics a lot to do better physics.
Lex Fridman (05:09.440)
But physics can also help AI a lot,
Lex Fridman (05:13.160)
both by building better hardware.
Max Tegmark (05:16.440)
My colleague, Marin Soljacic, for example,
Lex Fridman (05:18.840)
is working on an optical chip for much faster machine
Max Tegmark (05:23.000)
learning, where the computation is done
Lex Fridman (05:25.360)
not by moving electrons around, but by moving photons around,
Max Tegmark (05:30.240)
dramatically less energy use, faster, better.
Lex Fridman (05:34.240)
We can also help AI a lot, I think,
Max Tegmark (05:37.560)
by having a different set of tools
Lex Fridman (05:42.840)
and a different, maybe more audacious attitude.
Max Tegmark (05:46.440)
AI has, to a significant extent, been an engineering discipline
Lex Fridman (05:51.560)
where you're just trying to make things that work
Lex Fridman (05:54.240)
and being more interested in maybe selling them
Lex Fridman (05:56.560)
than in figuring out exactly how they work
Lex Fridman (06:00.280)
and proving theorems about that they will always work.
Lex Fridman (06:03.680)
Contrast that with physics.
Max Tegmark (06:05.240)
When Elon Musk sends a rocket to the International Space
Lex Fridman (06:08.920)
Station, they didn't just train with machine learning.
Max Tegmark (06:12.080)
Oh, let's fire it a little bit more to the left,
Lex Fridman (06:14.080)
a bit more to the right.
Max Tegmark (06:14.920)
Oh, that also missed.
Lex Fridman (06:15.800)
Let's try here.
Max Tegmark (06:16.920)
No, we figured out Newton's laws of gravitation and other things
Lex Fridman (06:23.400)
and got a really deep fundamental understanding.
Lex Fridman (06:26.800)
And that's what gives us such confidence in rockets.
Lex Fridman (06:30.840)
And my vision is that in the future,
Max Tegmark (06:36.040)
all machine learning systems that actually have impact
Lex Fridman (06:38.840)
on people's lives will be understood
Max Tegmark (06:40.960)
at a really, really deep level.
Lex Fridman (06:43.120)
So we trust them, not because some sales rep told us to,
Lex Fridman (06:46.960)
but because they've earned our trust.
Lex Fridman (06:50.360)
And really safety critical things
Max Tegmark (06:51.800)
even prove that they will always do what we expect them to do.
Lex Fridman (06:55.600)
That's very much the physics mindset.
Lex Fridman (06:57.040)
So it's interesting, if you look at big breakthroughs
Lex Fridman (07:00.160)
that have happened in machine learning this year,
Max Tegmark (07:03.680)
from dancing robots, it's pretty fantastic.
Lex Fridman (07:08.200)
Not just because it's cool, but if you just
Max Tegmark (07:10.280)
think about not that many years ago,
Lex Fridman (07:12.880)
this YouTube video at this DARPA challenge with the MIT robot
Max Tegmark (07:16.680)
comes out of the car and face plants.
Lex Fridman (07:20.680)
How far we've come in just a few years.
Max Tegmark (07:23.840)
Similarly, Alpha Fold 2, crushing the protein folding
Lex Fridman (07:30.360)
problem.
Max Tegmark (07:31.160)
We can talk more about implications
Lex Fridman (07:33.080)
for medical research and stuff.
Lex Fridman (07:34.400)
But hey, that's huge progress.
Lex Fridman (07:39.240)
You can look at the GPT3 that can spout off
Max Tegmark (07:44.120)
English text, which sometimes really, really blows you away.
Lex Fridman (07:48.840)
You can look at DeepMind's MuZero,
Max Tegmark (07:52.920)
which doesn't just kick our butt in Go and Chess and Shogi,
Lex Fridman (07:57.920)
but also in all these Atari games.
Lex Fridman (07:59.760)
And you don't even have to teach it the rules now.
Lex Fridman (08:02.960)
What all of those have in common is, besides being powerful,
Max Tegmark (08:06.920)
is we don't fully understand how they work.
Lex Fridman (08:10.520)
And that's fine if it's just some dancing robots.
Lex Fridman (08:13.160)
And the worst thing that can happen is they face plant.
Lex Fridman (08:16.440)
Or if they're playing Go, and the worst thing that can happen
Max Tegmark (08:19.120)
is that they make a bad move and lose the game.
Lex Fridman (08:22.240)
It's less fine if that's what's controlling
Max Tegmark (08:25.400)
your self driving car or your nuclear power plant.
Lex Fridman (08:29.120)
And we've seen already that even though Hollywood
Max Tegmark (08:33.600)
had all these movies where they try
Lex Fridman (08:35.040)
to make us worry about the wrong things,
Max Tegmark (08:37.000)
like machines turning evil, the actual bad things that
Lex Fridman (08:41.080)
have happened with automation have not
Max Tegmark (08:43.480)
been machines turning evil.
Lex Fridman (08:45.440)
They've been caused by overtrust in things
Max Tegmark (08:48.840)
we didn't understand as well as we thought we did.
Lex Fridman (08:51.440)
Even very simple automated systems
Max Tegmark (08:54.320)
like what Boeing put into the 737 MAX killed a lot of people.
Lex Fridman (09:00.440)
Was it that that little simple system was evil?
Max Tegmark (09:02.960)
Of course not.
Lex Fridman (09:03.920)
But we didn't understand it as well as we should have.
Lex Fridman (09:07.400)
And we trusted without understanding.
Lex Fridman (09:10.640)
Exactly.
Max Tegmark (09:11.440)
That's the overtrust.
Lex Fridman (09:12.440)
We didn't even understand that we didn't understand.
Max Tegmark (09:15.720)
The humility is really at the core of being a scientist.
Lex Fridman (09:19.880)
I think step one, if you want to be a scientist,
Max Tegmark (09:21.880)
is don't ever fool yourself into thinking you understand things
Lex Fridman (09:25.000)
when you actually don't.
Max Tegmark (09:27.080)
That's probably good advice for humans in general.
Lex Fridman (09:29.480)
I think humility in general can do us good.
Lex Fridman (09:31.320)
But in science, it's so spectacular.
Lex Fridman (09:33.240)
Why did we have the wrong theory of gravity
Lex Fridman (09:35.880)
ever from Aristotle onward until Galileo's time?
Lex Fridman (09:40.520)
Why would we believe something so dumb as that if I throw
Max Tegmark (09:43.680)
this water bottle, it's going to go up with constant speed
Lex Fridman (09:47.280)
until it realizes that its natural motion is down?
Max Tegmark (09:49.680)
It changes its mind.
Lex Fridman (09:51.040)
Because people just kind of assumed Aristotle was right.
Max Tegmark (09:55.320)
He's an authority.
Lex Fridman (09:56.120)
We understand that.
Lex Fridman (09:57.720)
Why did we believe things like that the sun is
Lex Fridman (09:59.920)
going around the Earth?
Lex Fridman (10:01.880)
Why did we believe that time flows
Lex Fridman (10:04.000)
at the same rate for everyone until Einstein?
Max Tegmark (10:06.440)
Same exact mistake over and over again.
Lex Fridman (10:08.560)
We just weren't humble enough to acknowledge that we actually
Max Tegmark (10:12.320)
didn't know for sure.
Lex Fridman (10:13.920)
We assumed we knew.
Lex Fridman (10:15.720)
So we didn't discover the truth because we
Lex Fridman (10:17.760)
assumed there was nothing there to be discovered, right?
Max Tegmark (10:20.560)
There was something to be discovered about the 737 Max.
Lex Fridman (10:24.400)
And if you had been a bit more suspicious
Lex Fridman (10:26.480)
and tested it better, we would have found it.
Lex Fridman (10:28.680)
And it's the same thing with most harm
Max Tegmark (10:30.600)
that's been done by automation so far, I would say.
Lex Fridman (10:33.760)
So I don't know if you heard here of a company called
Lex Fridman (10:35.720)
Knight Capital?
Lex Fridman (10:38.000)
So good.
Max Tegmark (10:38.760)
That means you didn't invest in them earlier.
Lex Fridman (10:42.080)
They deployed this automated trading system,
Max Tegmark (10:45.560)
all nice and shiny.
Lex Fridman (10:47.000)
They didn't understand it as well as they thought.
Lex Fridman (10:49.480)
And it went about losing $10 million
Lex Fridman (10:51.320)
per minute for 44 minutes straight
Max Tegmark (10:55.520)
until someone presumably was like, oh, no, shut this up.
Lex Fridman (10:59.520)
Was it evil?
Max Tegmark (11:00.480)
No.
Lex Fridman (11:01.040)
It was, again, misplaced trust, something they didn't fully
Lex Fridman (11:04.400)
understand, right?
Lex Fridman (11:05.240)
And there have been so many, even when people
Max Tegmark (11:09.040)
have been killed by robots, which is quite rare still,
Lex Fridman (11:12.640)
but in factory accidents, it's in every single case
Max Tegmark (11:15.680)
been not malice, just that the robot didn't understand
Lex Fridman (11:19.080)
that a human is different from an auto part or whatever.
Lex Fridman (11:24.400)
So this is why I think there's so much opportunity
Lex Fridman (11:28.000)
for a physics approach, where you just aim for a higher
Max Tegmark (11:32.040)
level of understanding.
Lex Fridman (11:33.600)
And if you look at all these systems
Max Tegmark (11:36.200)
that we talked about from reinforcement learning
Lex Fridman (11:40.680)
systems and dancing robots to all these neural networks
Max Tegmark (11:44.240)
that power GPT3 and go playing software and stuff,
Lex Fridman (11:49.600)
they're all basically black boxes,
Max Tegmark (11:53.480)
not so different from if you teach a human something,
Lex Fridman (11:55.920)
you have no idea how their brain works, right?
Max Tegmark (11:58.120)
Except the human brain, at least,
Lex Fridman (11:59.960)
has been error corrected during many, many centuries
Max Tegmark (12:03.800)
of evolution in a way that some of these systems have not,
Lex Fridman (12:06.560)
right?
Lex Fridman (12:07.560)
And my MIT research is entirely focused
Lex Fridman (12:10.640)
on demystifying this black box, intelligible intelligence
Max Tegmark (12:14.440)
is my slogan.
Lex Fridman (12:15.960)
That's a good line, intelligible intelligence.
Max Tegmark (12:18.440)
Yeah, that we shouldn't settle for something
Lex Fridman (12:20.360)
that seems intelligent, but it should
Max Tegmark (12:22.160)
be intelligible so that we actually trust it
Lex Fridman (12:24.280)
because we understand it, right?
Max Tegmark (12:26.640)
Like, again, Elon trusts his rockets
Lex Fridman (12:28.880)
because he understands Newton's laws and thrust
Lex Fridman (12:31.600)
and how everything works.
Lex Fridman (12:33.800)
And can I tell you why I'm optimistic about this?
Max Tegmark (12:36.880)
Yes.
Lex Fridman (12:37.520)
I think we've made a bit of a mistake
Max Tegmark (12:41.280)
where some people still think that somehow we're never going
Lex Fridman (12:44.880)
to understand neural networks.
Max Tegmark (12:47.320)
We're just going to have to learn to live with this.
Lex Fridman (12:49.520)
It's this very powerful black box.
Max Tegmark (12:52.240)
Basically, for those who haven't spent time
Lex Fridman (12:55.840)
building their own, it's super simple what happens inside.
Max Tegmark (12:59.000)
You send in a long list of numbers,
Lex Fridman (13:01.280)
and then you do a bunch of operations on them,
Max Tegmark (13:04.880)
multiply by matrices, et cetera, et cetera,
Lex Fridman (13:06.880)
and some other numbers come out that's output of it.
Lex Fridman (13:09.840)
And then there are a bunch of knobs you can tune.
Lex Fridman (13:13.520)
And when you change them, it affects the computation,
Max Tegmark (13:16.680)
the input output relation.
Lex Fridman (13:18.080)
And then you just give the computer
Max Tegmark (13:19.560)
some definition of good, and it keeps optimizing these knobs
Lex Fridman (13:22.680)
until it performs as good as possible.
Lex Fridman (13:24.760)
And often, you go like, wow, that's really good.
Lex Fridman (13:27.160)
This robot can dance, or this machine
Max Tegmark (13:29.480)
is beating me at chess now.
Lex Fridman (13:31.960)
And in the end, you have something
Max Tegmark (13:33.400)
which, even though you can look inside it,
Lex Fridman (13:35.240)
you have very little idea of how it works.
Max Tegmark (13:38.680)
You can print out tables of all the millions of parameters
Lex Fridman (13:42.040)
in there.
Lex Fridman (13:43.240)
Is it crystal clear now how it's working?
Lex Fridman (13:45.000)
No, of course not.
Max Tegmark (13:46.840)
Many of my colleagues seem willing to settle for that.
Lex Fridman (13:49.080)
And I'm like, no, that's like the halfway point.
Max Tegmark (13:54.360)
Some have even gone as far as sort of guessing
Lex Fridman (13:57.560)
that the mistrutability of this is
Max Tegmark (14:00.800)
where some of the power comes from,
Lex Fridman (14:02.760)
and some sort of mysticism.
Max Tegmark (14:05.120)
I think that's total nonsense.
Lex Fridman (14:06.840)
I think the real power of neural networks
Max Tegmark (14:10.240)
comes not from inscrutability, but from differentiability.
Lex Fridman (14:15.040)
And what I mean by that is simply
Max Tegmark (14:17.640)
that the output changes only smoothly if you tweak your knobs.
Lex Fridman (14:23.880)
And then you can use all these powerful methods
Max Tegmark (14:26.640)
we have for optimization in science.
Lex Fridman (14:28.320)
We can just tweak them a little bit and see,
Lex Fridman (14:30.160)
did that get better or worse?
Lex Fridman (14:31.680)
That's the fundamental idea of machine learning,
Max Tegmark (14:33.920)
that the machine itself can keep optimizing
Lex Fridman (14:36.080)
until it gets better.
Max Tegmark (14:37.240)
Suppose you wrote this algorithm instead in Python
Lex Fridman (14:41.920)
or some other programming language,
Lex Fridman (14:43.720)
and then what the knobs did was they just changed
Lex Fridman (14:46.280)
random letters in your code.
Max Tegmark (14:49.920)
Now it would just epically fail.
Lex Fridman (14:51.440)
You change one thing, and instead of saying print,
Max Tegmark (14:53.560)
it says, synth, syntax error.
Lex Fridman (14:56.840)
You don't even know, was that for the better
Lex Fridman (14:58.720)
or for the worse, right?
Lex Fridman (14:59.920)
This, to me, is what I believe is
Max Tegmark (15:02.720)
the fundamental power of neural networks.
Lex Fridman (15:05.240)
And just to clarify, the changing
Max Tegmark (15:06.640)
of the different letters in a program
Lex Fridman (15:08.400)
would not be a differentiable process.
Max Tegmark (15:10.600)
It would make it an invalid program, typically.
Lex Fridman (15:13.760)
And then you wouldn't even know if you changed more letters
Lex Fridman (15:16.800)
if it would make it work again, right?
Lex Fridman (15:18.560)
So that's the magic of neural networks, the inscrutability.
Max Tegmark (15:23.360)
The differentiability, that every setting of the parameters
Lex Fridman (15:26.600)
is a program, and you can tell is it better or worse, right?
Lex Fridman (15:29.040)
And so.
Lex Fridman (15:31.040)
So you don't like the poetry of the mystery of neural networks
Lex Fridman (15:33.680)
as the source of its power?
Lex Fridman (15:35.120)
I generally like poetry, but.
Max Tegmark (15:37.880)
Not in this case.
Lex Fridman (15:39.200)
It's so misleading.
Lex Fridman (15:40.440)
And above all, it shortchanges us.
Lex Fridman (15:42.880)
It makes us underestimate the good things
Max Tegmark (15:46.440)
we can accomplish.
Lex Fridman (15:47.880)
So what we've been doing in my group
Max Tegmark (15:49.400)
is basically step one, train the mysterious neural network
Lex Fridman (15:53.000)
to do something well.
Lex Fridman (15:54.920)
And then step two, do some additional AI techniques
Lex Fridman (15:59.560)
to see if we can now transform this black box into something
Max Tegmark (16:03.280)
equally intelligent that you can actually understand.
Lex Fridman (16:07.120)
So for example, I'll give you one example, this AI Feynman
Lex Fridman (16:09.800)
project that we just published, right?
Lex Fridman (16:11.560)
So we took the 100 most famous or complicated equations
Max Tegmark (16:18.080)
from one of my favorite physics textbooks,
Lex Fridman (16:20.880)
in fact, the one that got me into physics
Max Tegmark (16:22.560)
in the first place, the Feynman lectures on physics.
Lex Fridman (16:25.760)
And so you have a formula.
Max Tegmark (16:28.520)
Maybe it has what goes into the formula
Lex Fridman (16:31.680)
as six different variables, and then what comes out as one.
Lex Fridman (16:35.960)
So then you can make a giant Excel spreadsheet
Lex Fridman (16:38.000)
with seven columns.
Max Tegmark (16:39.600)
You put in just random numbers for the six columns
Lex Fridman (16:41.680)
for those six input variables, and then you
Max Tegmark (16:43.880)
calculate with a formula the seventh column, the output.
Lex Fridman (16:46.880)
So maybe it's like the force equals in the last column
Max Tegmark (16:50.440)
some function of the other.
Lex Fridman (16:51.720)
And now the task is, OK, if I don't tell you
Lex Fridman (16:53.840)
what the formula was, can you figure that out
Lex Fridman (16:57.320)
from looking at my spreadsheet I gave you?
Max Tegmark (17:00.080)
This problem is called symbolic regression.
Lex Fridman (17:04.400)
If I tell you that the formula is
Lex Fridman (17:05.800)
what we call a linear formula, so it's just
Lex Fridman (17:08.160)
that the output is sum of all the things, input, the times,
Max Tegmark (17:14.760)
some constants, that's the famous easy problem
Lex Fridman (17:17.440)
we can solve.
Max Tegmark (17:18.680)
We do it all the time in science and engineering.
Lex Fridman (17:21.360)
But the general one, if it's more complicated functions
Max Tegmark (17:24.480)
with logarithms or cosines or other math,
Lex Fridman (17:27.920)
it's a very, very hard one and probably impossible
Max Tegmark (17:30.560)
to do fast in general, just because the number of formulas
Lex Fridman (17:34.560)
with n symbols just grows exponentially,
Max Tegmark (17:37.360)
just like the number of passwords
Lex Fridman (17:38.760)
you can make grow dramatically with length.
Lex Fridman (17:43.320)
But we had this idea that if you first
Lex Fridman (17:46.160)
have a neural network that can actually approximate
Max Tegmark (17:48.480)
the formula, you just trained it,
Lex Fridman (17:49.880)
even if you don't understand how it works,
Max Tegmark (17:51.960)
that can be the first step towards actually understanding
Lex Fridman (17:56.560)
how it works.
Lex Fridman (17:58.280)
So that's what we do first.
Lex Fridman (18:00.600)
And then we study that neural network now
Lex Fridman (18:03.240)
and put in all sorts of other data
Lex Fridman (18:04.880)
that wasn't in the original training data
Lex Fridman (18:06.720)
and use that to discover simplifying
Lex Fridman (18:09.400)
properties of the formula.
Lex Fridman (18:11.460)
And that lets us break it apart, often
Lex Fridman (18:13.160)
into many simpler pieces in a kind of divide
Lex Fridman (18:15.560)
and conquer approach.
Lex Fridman (18:17.480)
So we were able to solve all of those 100 formulas,
Max Tegmark (18:20.120)
discover them automatically, plus a whole bunch
Lex Fridman (18:22.160)
of other ones.
Lex Fridman (18:22.720)
And it's actually kind of humbling
Lex Fridman (18:26.320)
to see that this code, which anyone who wants now
Max Tegmark (18:29.480)
is listening to this, can type pip install AI Feynman
Lex Fridman (18:33.200)
on the computer and run it.
Max Tegmark (18:34.560)
It can actually do what Johannes Kepler spent four years doing
Lex Fridman (18:38.360)
when he stared at Mars data until he was like,
Max Tegmark (18:40.800)
finally, Eureka, this is an ellipse.
Lex Fridman (18:44.520)
This will do it automatically for you in one hour.
Max Tegmark (18:46.960)
Or Max Planck, he was looking at how much radiation comes out
Lex Fridman (18:51.600)
from different wavelengths from a hot object
Lex Fridman (18:54.160)
and discovered the famous blackbody formula.
Lex Fridman (18:57.400)
This discovers it automatically.
Max Tegmark (19:00.400)
I'm actually excited about seeing
Lex Fridman (19:05.120)
if we can discover not just old formulas again,
Lex Fridman (19:08.640)
but new formulas that no one has seen before.
Lex Fridman (19:12.000)
I do like this process of using kind of a neural network
Max Tegmark (19:14.680)
to find some basic insights and then dissecting
Lex Fridman (19:18.440)
the neural network to then gain the final.
Lex Fridman (19:21.680)
So in that way, you've forcing the explainability issue,
Lex Fridman (19:30.680)
really trying to analyze the neural network for the things
Max Tegmark (19:34.880)
it knows in order to come up with the final beautiful,
Lex Fridman (19:38.360)
simple theory underlying the initial system
Max Tegmark (19:42.240)
that you were looking at.
Lex Fridman (19:43.080)
I love that.
Lex Fridman (19:44.280)
And the reason I'm so optimistic that it
Lex Fridman (19:47.440)
can be generalized to so much more
Max Tegmark (19:49.040)
is because that's exactly what we do as human scientists.
Lex Fridman (19:53.480)
Think of Galileo, whom we mentioned, right?
Max Tegmark (19:55.680)
I bet when he was a little kid, if his dad threw him an apple,
Lex Fridman (19:58.760)
he would catch it.
Max Tegmark (1:00:01.560)
to figure out how to really play this new game
Lex Fridman (1:00:04.080)
and become very, very influential.
Lex Fridman (1:00:05.960)
But I think Donald Trump was as simple.
Lex Fridman (1:00:09.600)
He took advantage of it.
Max Tegmark (1:00:11.120)
He didn't create the fundamental conditions
Lex Fridman (1:00:14.520)
were created by machine learning taking over the news media.
Lex Fridman (1:00:19.040)
So this is what motivated my little COVID project here.
Lex Fridman (1:00:22.920)
So I said before, machine learning and tech in general
Max Tegmark (1:00:27.120)
is not evil, but it's also not good.
Lex Fridman (1:00:29.040)
It's just a tool that you can use
Max Tegmark (1:00:31.400)
for good things or bad things.
Lex Fridman (1:00:32.680)
And as it happens, machine learning and news
Max Tegmark (1:00:36.000)
was mainly used by the big players, big tech,
Lex Fridman (1:00:39.680)
to manipulate people and to watch as many ads as possible,
Max Tegmark (1:00:43.240)
which had this unintended consequence of really screwing
Lex Fridman (1:00:45.720)
up our democracy and fragmenting it into filter bubbles.
Lex Fridman (1:00:50.440)
So I thought, well, machine learning algorithms
Lex Fridman (1:00:53.200)
are basically free.
Max Tegmark (1:00:54.400)
They can run on your smartphone for free also
Lex Fridman (1:00:56.200)
if someone gives them away to you, right?
Max Tegmark (1:00:57.840)
There's no reason why they only have to help the big guy
Lex Fridman (1:01:01.840)
to manipulate the little guy.
Max Tegmark (1:01:02.960)
They can just as well help the little guy
Lex Fridman (1:01:05.280)
to see through all the manipulation attempts
Max Tegmark (1:01:07.880)
from the big guy.
Lex Fridman (1:01:08.720)
So this project is called,
Max Tegmark (1:01:10.600)
you can go to improvethenews.org.
Lex Fridman (1:01:12.800)
The first thing we've built is this little news aggregator.
Max Tegmark (1:01:16.600)
Looks a bit like Google News,
Lex Fridman (1:01:17.880)
except it has these sliders on it to help you break out
Max Tegmark (1:01:20.200)
of your filter bubble.
Lex Fridman (1:01:21.760)
So if you're reading, you can click, click
Lex Fridman (1:01:24.440)
and go to your favorite topic.
Lex Fridman (1:01:27.080)
And then if you just slide the left, right slider
Max Tegmark (1:01:31.120)
away all the way over to the left.
Lex Fridman (1:01:32.560)
There's two sliders, right?
Max Tegmark (1:01:33.720)
Yeah, there's the one, the most obvious one
Lex Fridman (1:01:36.280)
is the one that has left, right labeled on it.
Max Tegmark (1:01:38.800)
You go to the left, you get one set of articles,
Lex Fridman (1:01:40.560)
you go to the right, you see a very different truth
Max Tegmark (1:01:43.360)
appearing.
Lex Fridman (1:01:44.200)
Oh, that's literally left and right on the political spectrum.
Max Tegmark (1:01:47.640)
On the political spectrum.
Lex Fridman (1:01:48.480)
So if you're reading about immigration, for example,
Max Tegmark (1:01:52.720)
it's very, very noticeable.
Lex Fridman (1:01:55.560)
And I think step one always,
Max Tegmark (1:01:57.080)
if you wanna not get manipulated is just to be able
Lex Fridman (1:02:00.960)
to recognize the techniques people use.
Lex Fridman (1:02:02.960)
So it's very helpful to just see how they spin things
Lex Fridman (1:02:05.880)
on the two sides.
Max Tegmark (1:02:08.160)
I think many people are under the misconception
Lex Fridman (1:02:11.240)
that the main problem is fake news.
Max Tegmark (1:02:14.080)
It's not.
Lex Fridman (1:02:14.920)
I had an amazing team of MIT students
Max Tegmark (1:02:17.520)
where we did an academic project to use machine learning
Lex Fridman (1:02:20.360)
to detect the main kinds of bias over the summer.
Lex Fridman (1:02:23.080)
And yes, of course, sometimes there's fake news
Lex Fridman (1:02:25.640)
where someone just claims something that's false, right?
Max Tegmark (1:02:30.000)
Like, oh, Hillary Clinton just got divorced or something.
Lex Fridman (1:02:32.920)
But what we see much more of is actually just omissions.
Max Tegmark (1:02:37.800)
If you go to, there's some stories which just won't be
Lex Fridman (1:02:41.920)
mentioned by the left or the right, because it doesn't suit
Max Tegmark (1:02:45.520)
their agenda.
Lex Fridman (1:02:46.360)
And then they'll mention other ones very, very, very much.
Lex Fridman (1:02:49.680)
So for example, we've had a number of stories
Lex Fridman (1:02:54.680)
about the Trump family's financial dealings.
Lex Fridman (1:02:59.600)
And then there's been a bunch of stories
Lex Fridman (1:03:01.560)
about the Biden family's, Hunter Biden's financial dealings.
Max Tegmark (1:03:05.280)
Surprise, surprise, they don't get equal coverage
Lex Fridman (1:03:07.520)
on the left and the right.
Max Tegmark (1:03:08.920)
One side loves to cover the Biden, Hunter Biden's stuff,
Lex Fridman (1:03:13.320)
and one side loves to cover the Trump.
Lex Fridman (1:03:15.360)
You can never guess which is which, right?
Lex Fridman (1:03:17.320)
But the great news is if you're a normal American citizen
Lex Fridman (1:03:21.560)
and you dislike corruption in all its forms,
Lex Fridman (1:03:24.960)
then slide, slide, you can just look at both sides
Lex Fridman (1:03:28.560)
and you'll see all those political corruption stories.
Lex Fridman (1:03:32.520)
It's really liberating to just take in the both sides,
Max Tegmark (1:03:37.520)
the spin on both sides.
Lex Fridman (1:03:39.440)
It somehow unlocks your mind to think on your own,
Max Tegmark (1:03:42.720)
to realize that, I don't know, it's the same thing
Lex Fridman (1:03:47.040)
that was useful, right, in the Soviet Union times
Max Tegmark (1:03:49.840)
for when everybody was much more aware
Lex Fridman (1:03:54.360)
that they're surrounded by propaganda, right?
Max Tegmark (1:03:57.200)
That is so interesting what you're saying, actually.
Lex Fridman (1:04:00.600)
So Noam Chomsky, used to be our MIT colleague,
Max Tegmark (1:04:04.000)
once said that propaganda is to democracy
Lex Fridman (1:04:07.640)
what violence is to totalitarianism.
Lex Fridman (1:04:11.960)
And what he means by that is if you have
Lex Fridman (1:04:15.080)
a really totalitarian government,
Max Tegmark (1:04:16.680)
you don't need propaganda.
Lex Fridman (1:04:19.680)
People will do what you want them to do anyway,
Lex Fridman (1:04:22.880)
but out of fear, right?
Lex Fridman (1:04:24.320)
But otherwise, you need propaganda.
Lex Fridman (1:04:28.080)
So I would say actually that the propaganda
Lex Fridman (1:04:29.880)
is much higher quality in democracies,
Max Tegmark (1:04:32.560)
much more believable.
Lex Fridman (1:04:34.240)
And it's really, it's really striking.
Max Tegmark (1:04:36.960)
When I talk to colleagues, science colleagues
Lex Fridman (1:04:39.520)
like from Russia and China and so on,
Max Tegmark (1:04:42.200)
I notice they are actually much more aware
Lex Fridman (1:04:45.120)
of the propaganda in their own media
Max Tegmark (1:04:47.200)
than many of my American colleagues are
Lex Fridman (1:04:48.840)
about the propaganda in Western media.
Max Tegmark (1:04:51.120)
That's brilliant.
Lex Fridman (1:04:51.960)
That means the propaganda in the Western media
Max Tegmark (1:04:53.880)
is just better.
Lex Fridman (1:04:54.720)
Yes.
Max Tegmark (1:04:55.560)
That's so brilliant.
Lex Fridman (1:04:56.400)
Everything's better in the West, even the propaganda.
Lex Fridman (1:04:58.200)
But once you realize that,
Lex Fridman (1:05:07.360)
you realize there's also something very optimistic there
Lex Fridman (1:05:09.280)
that you can do about it, right?
Lex Fridman (1:05:10.480)
Because first of all, omissions,
Max Tegmark (1:05:14.040)
as long as there's no outright censorship,
Lex Fridman (1:05:16.920)
you can just look at both sides
Lex Fridman (1:05:19.840)
and pretty quickly piece together
Lex Fridman (1:05:22.760)
a much more accurate idea of what's actually going on, right?
Lex Fridman (1:05:26.120)
And develop a natural skepticism too.
Lex Fridman (1:05:28.040)
Yeah.
Max Tegmark (1:05:28.880)
Just an analytical scientific mind
Lex Fridman (1:05:31.600)
about the way you're taking the information.
Max Tegmark (1:05:32.880)
Yeah.
Lex Fridman (1:05:33.720)
And I think, I have to say,
Max Tegmark (1:05:35.480)
sometimes I feel that some of us in the academic bubble
Lex Fridman (1:05:38.480)
are too arrogant about this and somehow think,
Max Tegmark (1:05:41.440)
oh, it's just people who aren't as educated
Lex Fridman (1:05:44.560)
as the dots are pulled.
Max Tegmark (1:05:45.800)
When we are often just as gullible also,
Lex Fridman (1:05:48.240)
we read only our media and don't see through things.
Max Tegmark (1:05:52.080)
Anyone who looks at both sides like this
Lex Fridman (1:05:53.960)
and compares a little will immediately start noticing
Max Tegmark (1:05:56.320)
the shenanigans being pulled.
Lex Fridman (1:05:58.080)
And I think what I tried to do with this app
Max Tegmark (1:06:01.840)
is that the big tech has to some extent
Lex Fridman (1:06:05.760)
tried to blame the individual for being manipulated,
Max Tegmark (1:06:08.960)
much like big tobacco tried to blame the individuals
Lex Fridman (1:06:12.320)
entirely for smoking.
Lex Fridman (1:06:13.680)
And then later on, our government stepped up and say,
Lex Fridman (1:06:16.880)
actually, you can't just blame little kids
Max Tegmark (1:06:19.560)
for starting to smoke.
Lex Fridman (1:06:20.400)
We have to have more responsible advertising
Lex Fridman (1:06:22.400)
and this and that.
Lex Fridman (1:06:23.480)
I think it's a bit the same here.
Max Tegmark (1:06:24.600)
It's very convenient for a big tech to blame.
Lex Fridman (1:06:27.600)
So it's just people who are so dumb and get fooled.
Max Tegmark (1:06:32.160)
The blame usually comes in saying,
Lex Fridman (1:06:34.160)
oh, it's just human psychology.
Max Tegmark (1:06:36.000)
People just wanna hear what they already believe.
Lex Fridman (1:06:38.360)
But professor David Rand at MIT actually partly debunked that
Max Tegmark (1:06:43.160)
with a really nice study showing that people
Lex Fridman (1:06:45.280)
tend to be interested in hearing things
Max Tegmark (1:06:47.640)
that go against what they believe,
Lex Fridman (1:06:49.880)
if it's presented in a respectful way.
Max Tegmark (1:06:52.680)
Suppose, for example, that you have a company
Lex Fridman (1:06:57.560)
and you're just about to launch this project
Lex Fridman (1:06:59.120)
and you're convinced it's gonna work.
Lex Fridman (1:07:00.280)
And someone says, you know, Lex,
Max Tegmark (1:07:03.520)
I hate to tell you this, but this is gonna fail.
Lex Fridman (1:07:05.640)
And here's why.
Max Tegmark (1:07:06.640)
Would you be like, shut up, I don't wanna hear it.
Lex Fridman (1:07:08.920)
La, la, la, la, la, la, la, la, la.
Lex Fridman (1:07:10.640)
Would you?
Lex Fridman (1:07:11.480)
You would be interested, right?
Lex Fridman (1:07:13.000)
And also if you're on an airplane,
Lex Fridman (1:07:16.360)
back in the pre COVID times,
Lex Fridman (1:07:19.000)
and the guy next to you
Lex Fridman (1:07:20.240)
is clearly from the opposite side of the political spectrum,
Lex Fridman (1:07:24.160)
but is very respectful and polite to you.
Lex Fridman (1:07:26.720)
Wouldn't you be kind of interested to hear a bit about
Lex Fridman (1:07:28.840)
how he or she thinks about things?
Lex Fridman (1:07:31.960)
Of course.
Lex Fridman (1:07:32.800)
But it's not so easy to find out
Lex Fridman (1:07:35.360)
respectful disagreement now,
Max Tegmark (1:07:36.760)
because like, for example, if you are a Democrat
Lex Fridman (1:07:40.440)
and you're like, oh, I wanna see something
Max Tegmark (1:07:41.920)
on the other side,
Lex Fridman (1:07:42.760)
so you just go Breitbart.com.
Lex Fridman (1:07:45.080)
And then after the first 10 seconds,
Lex Fridman (1:07:46.960)
you feel deeply insulted by something.
Lex Fridman (1:07:49.400)
And they, it's not gonna work.
Lex Fridman (1:07:52.480)
Or if you take someone who votes Republican
Lex Fridman (1:07:55.640)
and they go to something on the left,
Lex Fridman (1:07:57.400)
then they just get very offended very quickly
Max Tegmark (1:08:00.120)
by them having put a deliberately ugly picture
Lex Fridman (1:08:02.200)
of Donald Trump on the front page or something.
Max Tegmark (1:08:04.320)
It doesn't really work.
Lex Fridman (1:08:05.640)
So this news aggregator also has this nuance slider,
Max Tegmark (1:08:09.800)
which you can pull to the right
Lex Fridman (1:08:11.440)
and then sort of make it easier to get exposed
Max Tegmark (1:08:13.960)
to actually more sort of academic style
Lex Fridman (1:08:16.120)
or more respectful,
Max Tegmark (1:08:17.200)
portrayals of different views.
Lex Fridman (1:08:19.480)
And finally, the one kind of bias
Max Tegmark (1:08:22.080)
I think people are mostly aware of is the left, right,
Lex Fridman (1:08:25.440)
because it's so obvious,
Lex Fridman (1:08:26.280)
because both left and right are very powerful here, right?
Lex Fridman (1:08:30.600)
Both of them have well funded TV stations and newspapers,
Lex Fridman (1:08:33.920)
and it's kind of hard to miss.
Lex Fridman (1:08:35.520)
But there's another one, the establishment slider,
Max Tegmark (1:08:39.000)
which is also really fun.
Lex Fridman (1:08:41.320)
I love to play with it.
Lex Fridman (1:08:42.840)
And that's more about corruption.
Lex Fridman (1:08:44.360)
Yeah, yeah.
Max Tegmark (1:08:45.200)
I love that one. Yes.
Lex Fridman (1:08:47.600)
Because if you have a society
Max Tegmark (1:08:53.240)
where almost all the powerful entities
Lex Fridman (1:08:57.320)
want you to believe a certain thing,
Max Tegmark (1:08:59.480)
that's what you're gonna read in both the big media,
Lex Fridman (1:09:01.840)
mainstream media on the left and on the right, of course.
Lex Fridman (1:09:04.640)
And the powerful companies can push back very hard,
Lex Fridman (1:09:08.200)
like tobacco companies push back very hard
Max Tegmark (1:09:10.160)
back in the day when some newspapers
Lex Fridman (1:09:12.160)
started writing articles about tobacco being dangerous,
Lex Fridman (1:09:15.400)
so that it was hard to get a lot of coverage
Lex Fridman (1:09:17.000)
about it initially.
Lex Fridman (1:09:18.480)
And also if you look geopolitically, right,
Lex Fridman (1:09:20.880)
of course, in any country, when you read their media,
Max Tegmark (1:09:23.120)
you're mainly gonna be reading a lot of articles
Lex Fridman (1:09:24.880)
about how our country is the good guy
Lex Fridman (1:09:27.360)
and the other countries are the bad guys, right?
Lex Fridman (1:09:30.400)
So if you wanna have a really more nuanced understanding,
Max Tegmark (1:09:33.360)
like the Germans used to be told that the British
Lex Fridman (1:09:37.040)
used to be told that the French were the bad guys
Lex Fridman (1:09:38.840)
and the French used to be told
Lex Fridman (1:09:39.880)
that the British were the bad guys.
Max Tegmark (1:09:41.880)
Now they visit each other's countries a lot
Lex Fridman (1:09:45.680)
and have a much more nuanced understanding.
Max Tegmark (1:09:47.360)
I don't think there's gonna be any more wars
Lex Fridman (1:09:48.840)
between France and Germany.
Lex Fridman (1:09:50.120)
But on the geopolitical scale,
Lex Fridman (1:09:53.000)
it's just as much as ever, you know,
Max Tegmark (1:09:54.520)
big Cold War, now US, China, and so on.
Lex Fridman (1:09:57.600)
And if you wanna get a more nuanced understanding
Max Tegmark (1:10:01.200)
of what's happening geopolitically,
Lex Fridman (1:10:03.520)
then it's really fun to look at this establishment slider
Max Tegmark (1:10:05.960)
because it turns out there are tons of little newspapers,
Lex Fridman (1:10:09.360)
both on the left and on the right,
Max Tegmark (1:10:11.360)
who sometimes challenge establishment and say,
Lex Fridman (1:10:14.480)
you know, maybe we shouldn't actually invade Iraq right now.
Max Tegmark (1:10:17.800)
Maybe this weapons of mass destruction thing is BS.
Lex Fridman (1:10:20.400)
If you look at the journalism research afterwards,
Max Tegmark (1:10:23.680)
you can actually see that quite clearly.
Lex Fridman (1:10:25.360)
Both CNN and Fox were very pro.
Max Tegmark (1:10:29.200)
Let's get rid of Saddam.
Lex Fridman (1:10:30.640)
There are weapons of mass destruction.
Max Tegmark (1:10:32.560)
Then there were a lot of smaller newspapers.
Lex Fridman (1:10:34.680)
They were like, wait a minute,
Max Tegmark (1:10:36.200)
this evidence seems a bit sketchy and maybe we...
Lex Fridman (1:10:40.240)
But of course they were so hard to find.
Lex Fridman (1:10:42.240)
Most people didn't even know they existed, right?
Lex Fridman (1:10:44.560)
Yet it would have been better for American national security
Max Tegmark (1:10:47.400)
if those voices had also come up.
Lex Fridman (1:10:50.160)
I think it harmed America's national security actually
Max Tegmark (1:10:52.560)
that we invaded Iraq.
Lex Fridman (1:10:53.800)
And arguably there's a lot more interest
Max Tegmark (1:10:55.560)
in that kind of thinking too, from those small sources.
Lex Fridman (1:11:00.480)
So like when you say big,
Max Tegmark (1:11:02.600)
it's more about kind of the reach of the broadcast,
Lex Fridman (1:11:07.600)
but it's not big in terms of the interest.
Max Tegmark (1:11:12.040)
I think there's a lot of interest
Lex Fridman (1:11:14.120)
in that kind of anti establishment
Max Tegmark (1:11:16.200)
or like skepticism towards, you know,
Lex Fridman (1:11:18.840)
out of the box thinking.
Max Tegmark (1:11:20.360)
There's a lot of interest in that kind of thing.
Lex Fridman (1:11:22.000)
Do you see this news project or something like it
Max Tegmark (1:11:26.920)
being basically taken over the world
Lex Fridman (1:11:30.600)
as the main way we consume information?
Lex Fridman (1:11:32.920)
Like how do we get there?
Lex Fridman (1:11:35.120)
Like how do we, you know?
Max Tegmark (1:11:37.320)
So, okay, the idea is brilliant.
Lex Fridman (1:11:39.000)
It's a, you're calling it your little project in 2020,
Lex Fridman (1:11:44.000)
but how does that become the new way we consume information?
Lex Fridman (1:11:48.480)
I hope, first of all, just to plant a little seed there
Max Tegmark (1:11:51.000)
because normally the big barrier of doing anything in media
Lex Fridman (1:11:55.920)
is you need a ton of money, but this costs no money at all.
Max Tegmark (1:11:59.280)
I've just been paying myself.
Lex Fridman (1:12:00.640)
You pay a tiny amount of money each month to Amazon
Max Tegmark (1:12:03.080)
to run the thing in their cloud.
Lex Fridman (1:12:04.640)
We're not, there never will never be any ads.
Max Tegmark (1:12:06.920)
The point is not to make any money off of it.
Lex Fridman (1:12:09.360)
And we just train machine learning algorithms
Max Tegmark (1:12:11.640)
to classify the articles and stuff.
Lex Fridman (1:12:13.160)
So it just kind of runs by itself.
Lex Fridman (1:12:14.840)
So if it actually gets good enough at some point
Lex Fridman (1:12:17.760)
that it starts catching on, it could scale.
Lex Fridman (1:12:20.720)
And if other people carbon copy it
Lex Fridman (1:12:23.120)
and make other versions that are better,
Max Tegmark (1:12:24.960)
that's the more the merrier.
Lex Fridman (1:12:28.200)
I think there's a real opportunity for machine learning
Max Tegmark (1:12:32.920)
to empower the individual against the powerful players.
Lex Fridman (1:12:39.880)
As I said in the beginning here, it's
Max Tegmark (1:12:41.600)
been mostly the other way around so far,
Lex Fridman (1:12:43.280)
that the big players have the AI and then they tell people,
Max Tegmark (1:12:46.960)
this is the truth, this is how it is.
Lex Fridman (1:12:49.600)
But it can just as well go the other way around.
Lex Fridman (1:12:52.200)
And when the internet was born, actually, a lot of people
Lex Fridman (1:12:54.320)
had this hope that maybe this will be
Max Tegmark (1:12:56.280)
a great thing for democracy, make it easier
Lex Fridman (1:12:58.120)
to find out about things.
Lex Fridman (1:12:59.480)
And maybe machine learning and things like this
Lex Fridman (1:13:02.320)
can actually help again.
Lex Fridman (1:13:03.720)
And I have to say, I think it's more important than ever now
Lex Fridman (1:13:07.080)
because this is very linked also to the whole future of life
Max Tegmark (1:13:12.160)
as we discussed earlier.
Lex Fridman (1:13:13.920)
We're getting this ever more powerful tech.
Max Tegmark (1:13:17.280)
Frank, it's pretty clear if you look
Lex Fridman (1:13:19.040)
on the one or two generation, three generation timescale
Max Tegmark (1:13:21.920)
that there are only two ways this can end geopolitically.
Lex Fridman (1:13:24.880)
Either it ends great for all humanity
Max Tegmark (1:13:27.640)
or it ends terribly for all of us.
Lex Fridman (1:13:31.640)
There's really no in between.
Lex Fridman (1:13:33.680)
And we're so stuck in that because technology
Lex Fridman (1:13:37.560)
knows no borders.
Lex Fridman (1:13:39.080)
And you can't have people fighting
Lex Fridman (1:13:42.200)
when the weapons just keep getting ever more
Max Tegmark (1:13:44.480)
powerful indefinitely.
Lex Fridman (1:13:47.040)
Eventually, the luck runs out.
Lex Fridman (1:13:50.280)
And right now we have, I love America,
Lex Fridman (1:13:55.480)
but the fact of the matter is what's good for America
Max Tegmark (1:13:59.840)
is not opposite in the long term to what's
Lex Fridman (1:14:02.000)
good for other countries.
Max Tegmark (1:14:04.600)
It would be if this was some sort of zero sum game
Lex Fridman (1:14:07.400)
like it was thousands of years ago when the only way one
Max Tegmark (1:14:10.960)
country could get more resources was
Lex Fridman (1:14:13.440)
to take land from other countries
Max Tegmark (1:14:14.960)
because that was basically the resource.
Lex Fridman (1:14:17.640)
Look at the map of Europe.
Max Tegmark (1:14:18.920)
Some countries kept getting bigger and smaller,
Lex Fridman (1:14:21.400)
endless wars.
Lex Fridman (1:14:23.280)
But then since 1945, there hasn't been any war
Lex Fridman (1:14:26.400)
in Western Europe.
Lex Fridman (1:14:27.160)
And they all got way richer because of tech.
Lex Fridman (1:14:29.920)
So the optimistic outcome is that the big winner
Max Tegmark (1:14:34.760)
in this century is going to be America and China and Russia
Lex Fridman (1:14:38.200)
and everybody else because technology just makes
Max Tegmark (1:14:40.200)
us all healthier and wealthier.
Lex Fridman (1:14:41.760)
And we just find some way of keeping the peace
Max Tegmark (1:14:44.680)
on this planet.
Lex Fridman (1:14:46.640)
But I think, unfortunately, there
Max Tegmark (1:14:48.760)
are some pretty powerful forces right now
Lex Fridman (1:14:50.440)
that are pushing in exactly the opposite direction
Lex Fridman (1:14:52.560)
and trying to demonize other countries, which just makes
Lex Fridman (1:14:55.920)
it more likely that this ever more powerful tech we're
Max Tegmark (1:14:58.360)
building is going to be used in disastrous ways.
Lex Fridman (1:15:02.200)
Yeah, for aggression versus cooperation,
Max Tegmark (1:15:04.400)
that kind of thing.
Lex Fridman (1:15:05.200)
Yeah, even look at just military AI now.
Max Tegmark (1:15:09.560)
It was so awesome to see these dancing robots.
Lex Fridman (1:15:12.160)
I loved it.
Lex Fridman (1:15:14.000)
But one of the biggest growth areas in robotics
Lex Fridman (1:15:17.080)
now is, of course, autonomous weapons.
Lex Fridman (1:15:19.480)
And 2020 was like the best marketing year
Lex Fridman (1:15:23.200)
ever for autonomous weapons.
Max Tegmark (1:15:24.400)
Because in both Libya, it's a civil war,
Lex Fridman (1:15:27.520)
and in Nagorno Karabakh, they made the decisive difference.
Lex Fridman (1:15:34.440)
And everybody else is watching this.
Lex Fridman (1:15:36.280)
Oh, yeah, we want to build autonomous weapons, too.
Max Tegmark (1:15:38.920)
In Libya, you had, on one hand, our ally,
Lex Fridman (1:15:45.080)
the United Arab Emirates that were flying
Max Tegmark (1:15:47.080)
their autonomous weapons that they bought from China,
Lex Fridman (1:15:50.640)
bombing Libyans.
Lex Fridman (1:15:51.880)
And on the other side, you had our other ally, Turkey,
Lex Fridman (1:15:54.280)
flying their drones.
Lex Fridman (1:15:57.200)
And they had no skin in the game,
Lex Fridman (1:16:00.480)
any of these other countries.
Lex Fridman (1:16:01.680)
And of course, it was the Libyans who really got screwed.
Lex Fridman (1:16:04.160)
In Nagorno Karabakh, you had actually, again,
Max Tegmark (1:16:09.280)
Turkey is sending drones built by this company that
Lex Fridman (1:16:12.400)
was actually founded by a guy who went to MIT AeroAstro.
Lex Fridman (1:16:17.080)
Do you know that?
Lex Fridman (1:16:17.800)
No.
Max Tegmark (1:16:18.280)
Bacratyar.
Lex Fridman (1:16:18.960)
Yeah.
Lex Fridman (1:16:19.520)
So MIT has a direct responsibility
Lex Fridman (1:16:21.480)
for ultimately this.
Lex Fridman (1:16:22.680)
And a lot of civilians were killed there.
Lex Fridman (1:16:25.680)
So because it was militarily so effective,
Max Tegmark (1:16:29.640)
now suddenly there's a huge push.
Lex Fridman (1:16:31.240)
Oh, yeah, yeah, let's go build ever more autonomy
Max Tegmark (1:16:35.680)
into these weapons, and it's going to be great.
Lex Fridman (1:16:39.440)
And I think, actually, people who
Max Tegmark (1:16:44.640)
are obsessed about some sort of future Terminator scenario
Lex Fridman (1:16:47.760)
right now should start focusing on the fact
Max Tegmark (1:16:51.640)
that we have two much more urgent threats happening
Lex Fridman (1:16:54.000)
from machine learning.
Max Tegmark (1:16:54.960)
One of them is the whole destruction of democracy
Lex Fridman (1:16:57.880)
that we've talked about now, where
Max Tegmark (1:17:01.600)
our flow of information is being manipulated
Lex Fridman (1:17:03.560)
by machine learning.
Lex Fridman (1:17:04.400)
And the other one is that right now,
Lex Fridman (1:17:06.960)
this is the year when the big arms race and out of control
Max Tegmark (1:17:10.440)
arms race in at least Thomas Weapons is going to start,
Lex Fridman (1:17:12.800)
or it's going to stop.
Lex Fridman (1:17:14.640)
So you have a sense that there is like 2020
Lex Fridman (1:17:18.480)
was an instrumental catalyst for the autonomous weapons race.
Max Tegmark (1:17:24.280)
Yeah, because it was the first year when they proved
Lex Fridman (1:17:26.560)
decisive in the battlefield.
Lex Fridman (1:17:28.360)
And these ones are still not fully autonomous, mostly.
Lex Fridman (1:17:31.400)
They're remote controlled, right?
Lex Fridman (1:17:32.640)
But we could very quickly make things
Lex Fridman (1:17:38.720)
about the size and cost of a smartphone, which you just put
Max Tegmark (1:17:43.280)
in the GPS coordinates or the face of the one
Lex Fridman (1:17:45.160)
you want to kill, a skin color or whatever,
Lex Fridman (1:17:47.000)
and it flies away and does it.
Lex Fridman (1:17:48.480)
And the real good reason why the US and all
Max Tegmark (1:17:53.920)
the other superpowers should put the kibosh on this
Lex Fridman (1:17:57.040)
is the same reason we decided to put the kibosh on bioweapons.
Lex Fridman (1:18:01.680)
So we gave the Future of Life Award
Lex Fridman (1:18:05.000)
that we can talk more about later to Matthew Messelson
Max Tegmark (1:18:07.200)
from Harvard before for convincing
Lex Fridman (1:18:08.680)
Nixon to ban bioweapons.
Lex Fridman (1:18:10.320)
And I asked him, how did you do it?
Lex Fridman (1:18:13.600)
And he was like, well, I just said, look,
Max Tegmark (1:18:16.560)
we don't want there to be a $500 weapon of mass destruction
Lex Fridman (1:18:20.520)
that all our enemies can afford, even nonstate actors.
Lex Fridman (1:18:26.560)
And Nixon was like, good point.
Lex Fridman (1:18:32.120)
It's in America's interest that the powerful weapons are all
Max Tegmark (1:18:34.520)
really expensive, so only we can afford them,
Lex Fridman (1:18:37.600)
or maybe some more stable adversaries, right?
Max Tegmark (1:18:41.080)
Nuclear weapons are like that.
Lex Fridman (1:18:42.960)
But bioweapons were not like that.
Max Tegmark (1:18:44.920)
That's why we banned them.
Lex Fridman (1:18:46.400)
And that's why you never hear about them now.
Max Tegmark (1:18:48.400)
That's why we love biology.
Lex Fridman (1:18:50.280)
So you have a sense that it's possible for the big power
Max Tegmark (1:18:55.440)
houses in terms of the big nations in the world
Lex Fridman (1:18:58.480)
to agree that autonomous weapons is not a race we want to be on,
Max Tegmark (1:19:02.360)
that it doesn't end well.
Lex Fridman (1:19:03.680)
Yeah, because we know it's just going
Max Tegmark (1:19:05.320)
to end in mass proliferation.
Lex Fridman (1:19:06.560)
And every terrorist everywhere is
Max Tegmark (1:19:08.560)
going to have these super cheap weapons
Lex Fridman (1:19:10.280)
that they will use against us.
Lex Fridman (1:19:13.440)
And our politicians have to constantly worry
Lex Fridman (1:19:15.960)
about being assassinated every time they go outdoors
Max Tegmark (1:19:18.240)
by some anonymous little mini drone.
Lex Fridman (1:19:21.040)
We don't want that.
Lex Fridman (1:19:21.840)
And even if the US and China and everyone else
Lex Fridman (1:19:25.920)
could just agree that you can only
Max Tegmark (1:19:27.840)
build these weapons if they cost at least $10 million,
Lex Fridman (1:19:31.560)
that would be a huge win for the superpowers
Max Tegmark (1:19:34.760)
and, frankly, for everybody.
Lex Fridman (1:19:38.800)
And people often push back and say, well, it's
Lex Fridman (1:19:41.000)
so hard to prevent cheating.
Lex Fridman (1:19:43.200)
But hey, you could say the same about bioweapons.
Max Tegmark (1:19:45.800)
Take any of your MIT colleagues in biology.
Lex Fridman (1:19:49.360)
Of course, they could build some nasty bioweapon
Max Tegmark (1:19:52.000)
if they really wanted to.
Lex Fridman (1:19:53.560)
But first of all, they don't want to
Max Tegmark (1:19:55.280)
because they think it's disgusting because of the stigma.
Lex Fridman (1:19:57.640)
And second, even if there's some sort of nutcase and want to,
Max Tegmark (1:20:02.120)
it's very likely that some of their grad students
Lex Fridman (1:20:04.160)
or someone would rat them out because everyone else thinks
Max Tegmark (1:20:06.560)
it's so disgusting.
Lex Fridman (1:20:08.000)
And in fact, we now know there was even a fair bit of cheating
Max Tegmark (1:20:11.480)
on the bioweapons ban.
Lex Fridman (1:20:13.480)
But no countries used them because it was so stigmatized
Max Tegmark (1:20:17.520)
that it just wasn't worth revealing that they had cheated.
Lex Fridman (1:20:22.400)
You talk about drones, but you kind of
Max Tegmark (1:20:24.840)
think that drones is a remote operation.
Lex Fridman (1:20:28.960)
Which they are, mostly, still.
Lex Fridman (1:20:30.680)
But you're not taking the next intellectual step
Lex Fridman (1:20:34.600)
of where does this go.
Max Tegmark (1:20:36.320)
You're kind of saying the problem with drones
Lex Fridman (1:20:38.760)
is that you're removing yourself from direct violence.
Max Tegmark (1:20:42.400)
Therefore, you're not able to sort of maintain
Lex Fridman (1:20:44.920)
the common humanity required to make
Max Tegmark (1:20:46.720)
the proper decisions strategically.
Lex Fridman (1:20:48.720)
But that's the criticism as opposed to like,
Max Tegmark (1:20:51.360)
if this is automated, and just exactly as you said,
Lex Fridman (1:20:55.520)
if you automate it and there's a race,
Max Tegmark (1:20:58.640)
then the technology's gonna get better and better and better
Lex Fridman (1:21:01.280)
which means getting cheaper and cheaper and cheaper.
Lex Fridman (1:21:03.720)
And unlike, perhaps, nuclear weapons
Lex Fridman (1:21:06.080)
which is connected to resources in a way,
Max Tegmark (1:21:10.240)
like it's hard to engineer, yeah.
Lex Fridman (1:21:13.760)
It feels like there's too much overlap
Max Tegmark (1:21:17.600)
between the tech industry and autonomous weapons
Lex Fridman (1:21:20.400)
to where you could have smartphone type of cheapness.
Max Tegmark (1:21:24.400)
If you look at drones, for $1,000,
Lex Fridman (1:21:29.280)
you can have an incredible system
Max Tegmark (1:21:30.800)
that's able to maintain flight autonomously for you
Lex Fridman (1:21:34.600)
and take pictures and stuff.
Max Tegmark (1:21:36.240)
You could see that going into the autonomous weapons space
Lex Fridman (1:21:39.440)
that's, but why is that not thought about
Lex Fridman (1:21:43.240)
or discussed enough in the public, do you think?
Lex Fridman (1:21:45.640)
You see those dancing Boston Dynamics robots
Lex Fridman (1:21:48.960)
and everybody has this kind of,
Lex Fridman (1:21:52.600)
as if this is like a far future.
Max Tegmark (1:21:55.360)
They have this fear like, oh, this'll be Terminator
Lex Fridman (1:21:58.640)
in like some, I don't know, unspecified 20, 30, 40 years.
Lex Fridman (1:22:03.080)
And they don't think about, well, this is like
Lex Fridman (1:22:05.640)
some much less dramatic version of that
Max Tegmark (1:22:09.120)
is actually happening now.
Lex Fridman (1:22:11.160)
It's not gonna be legged, it's not gonna be dancing,
Lex Fridman (1:22:14.840)
but it already has the capability
Lex Fridman (1:22:17.160)
to use artificial intelligence to kill humans.
Max Tegmark (1:22:20.240)
Yeah, the Boston Dynamics legged robots,
Lex Fridman (1:22:22.880)
I think the reason we imagine them holding guns
Lex Fridman (1:22:24.960)
is just because you've all seen Arnold Schwarzenegger, right?
Lex Fridman (1:22:28.440)
That's our reference point.
Max Tegmark (1:22:30.600)
That's pretty useless.
Lex Fridman (1:22:32.680)
That's not gonna be the main military use of them.
Max Tegmark (1:22:35.360)
They might be useful in law enforcement in the future
Lex Fridman (1:22:38.720)
and then there's a whole debate about,
Lex Fridman (1:22:40.280)
do you want robots showing up at your house with guns
Lex Fridman (1:22:42.640)
telling you who'll be perfectly obedient
Lex Fridman (1:22:45.440)
to whatever dictator controls them?
Lex Fridman (1:22:47.560)
But let's leave that aside for a moment
Lex Fridman (1:22:49.240)
and look at what's actually relevant now.
Lex Fridman (1:22:51.320)
So there's a spectrum of things you can do
Max Tegmark (1:22:54.760)
with AI in the military.
Lex Fridman (1:22:55.760)
And again, to put my card on the table,
Max Tegmark (1:22:57.560)
I'm not the pacifist, I think we should have good defense.
Lex Fridman (1:23:03.480)
So for example, a predator drone is basically
Lex Fridman (1:23:08.480)
a fancy little remote controlled airplane, right?
Lex Fridman (1:23:11.720)
There's a human piloting it and the decision ultimately
Max Tegmark (1:23:16.040)
about whether to kill somebody with it
Lex Fridman (1:23:17.280)
is made by a human still.
Lex Fridman (1:23:19.400)
And this is a line I think we should never cross.
Lex Fridman (1:23:23.880)
There's a current DOD policy.
Max Tegmark (1:23:25.880)
Again, you have to have a human in the loop.
Lex Fridman (1:23:27.920)
I think algorithms should never make life
Max Tegmark (1:23:30.680)
or death decisions, they should be left to humans.
Lex Fridman (1:23:34.120)
Now, why might we cross that line?
Lex Fridman (1:23:37.720)
Well, first of all, these are expensive, right?
Lex Fridman (1:23:40.520)
So for example, when Azerbaijan had all these drones
Lex Fridman (1:23:46.560)
and Armenia didn't have any, they start trying
Lex Fridman (1:23:48.280)
to jerry rig little cheap things, fly around.
Lex Fridman (1:23:51.760)
But then of course, the Armenians would jam them
Lex Fridman (1:23:54.040)
or the Azeris would jam them.
Lex Fridman (1:23:55.600)
And remote control things can be jammed,
Lex Fridman (1:23:58.320)
that makes them inferior.
Max Tegmark (1:24:00.040)
Also, there's a bit of a time delay between,
Lex Fridman (1:24:02.960)
if we're piloting something from far away,
Max Tegmark (1:24:05.400)
speed of light, and the human has a reaction time as well,
Lex Fridman (1:24:08.640)
it would be nice to eliminate that jamming possibility
Max Tegmark (1:24:11.560)
in the time that they by having it fully autonomous.
Lex Fridman (1:24:14.320)
But now you might be, so then if you do,
Lex Fridman (1:24:17.080)
but now you might be crossing that exact line.
Lex Fridman (1:24:19.400)
You might program it to just, oh yeah, the air drone,
Max Tegmark (1:24:22.360)
go hover over this country for a while
Lex Fridman (1:24:25.280)
and whenever you find someone who is a bad guy,
Max Tegmark (1:24:28.760)
kill them.
Lex Fridman (1:24:30.960)
Now the machine is making these sort of decisions
Lex Fridman (1:24:33.480)
and some people who defend this still say,
Lex Fridman (1:24:36.120)
well, that's morally fine because we are the good guys
Lex Fridman (1:24:39.960)
and we will tell it the definition of bad guy
Lex Fridman (1:24:43.000)
that we think is moral.
Lex Fridman (1:24:45.640)
But now it would be very naive to think
Lex Fridman (1:24:48.720)
that if ISIS buys that same drone,
Max Tegmark (1:24:51.480)
that they're gonna use our definition of bad guy.
Lex Fridman (1:24:54.040)
Maybe for them, bad guy is someone wearing
Max Tegmark (1:24:55.840)
a US army uniform or maybe there will be some,
Lex Fridman (1:25:00.840)
weird ethnic group who decides that someone
Lex Fridman (1:25:04.680)
of another ethnic group, they are the bad guys, right?
Lex Fridman (1:25:07.160)
The thing is human soldiers with all our faults,
Max Tegmark (1:25:11.320)
we still have some basic wiring in us.
Lex Fridman (1:25:14.080)
Like, no, it's not okay to kill kids and civilians.
Lex Fridman (1:25:20.040)
And Thomas Weprin has none of that.
Lex Fridman (1:25:21.760)
It's just gonna do whatever is programmed.
Max Tegmark (1:25:23.600)
It's like the perfect Adolf Eichmann on steroids.
Lex Fridman (1:25:27.720)
Like they told him, Adolf Eichmann, you know,
Max Tegmark (1:25:30.840)
he wanted to do this and this and this
Lex Fridman (1:25:32.240)
to make the Holocaust more efficient.
Lex Fridman (1:25:33.680)
And he was like, yeah, and off he went and did it, right?
Lex Fridman (1:25:37.840)
Do we really wanna make machines that are like that,
Max Tegmark (1:25:41.120)
like completely amoral and we'll take the user's definition
Lex Fridman (1:25:44.240)
of who is the bad guy?
Lex Fridman (1:25:45.720)
And do we then wanna make them so cheap
Lex Fridman (1:25:47.920)
that all our adversaries can have them?
Lex Fridman (1:25:49.640)
Like what could possibly go wrong?
Lex Fridman (1:25:52.720)
That's I think the big ordeal of the whole thing.
Max Tegmark (1:25:56.720)
I think the big argument for why we wanna,
Lex Fridman (1:26:00.200)
this year really put the kibosh on this.
Lex Fridman (1:26:03.520)
And I think you can tell there's a lot
Lex Fridman (1:26:06.360)
of very active debate even going on within the US military
Lex Fridman (1:26:10.120)
and undoubtedly in other militaries around the world also
Lex Fridman (1:26:13.080)
about whether we should have some sort
Max Tegmark (1:26:14.200)
of international agreement to at least require
Lex Fridman (1:26:16.760)
that these weapons have to be above a certain size
Lex Fridman (1:26:20.640)
and cost, you know, so that things just don't totally spiral
Lex Fridman (1:26:27.320)
out of control.
Lex Fridman (1:26:29.800)
And finally, just for your question,
Lex Fridman (1:26:31.600)
but is it possible to stop it?
Max Tegmark (1:26:33.560)
Because some people tell me, oh, just give up, you know.
Lex Fridman (1:26:37.000)
But again, so Matthew Messelsen again from Harvard, right,
Max Tegmark (1:26:41.560)
who the bioweapons hero, he had exactly this criticism
Lex Fridman (1:26:46.640)
also with bioweapons.
Max Tegmark (1:26:47.760)
People were like, how can you check for sure
Lex Fridman (1:26:49.920)
that the Russians aren't cheating?
Lex Fridman (1:26:52.960)
And he told me this, I think really ingenious insight.
Lex Fridman (1:26:58.560)
He said, you know, Max, some people
Max Tegmark (1:27:01.200)
think you have to have inspections and things
Lex Fridman (1:27:03.640)
and you have to make sure that you can catch the cheaters
Max Tegmark (1:27:06.760)
with 100% chance.
Lex Fridman (1:27:08.960)
You don't need 100%, he said.
Max Tegmark (1:27:10.800)
1% is usually enough.
Lex Fridman (1:27:14.080)
Because if it's another big state,
Max Tegmark (1:27:19.240)
suppose China and the US have signed the treaty drawing
Lex Fridman (1:27:23.480)
a certain line and saying, yeah, these kind of drones are OK,
Lex Fridman (1:27:26.200)
but these fully autonomous ones are not.
Lex Fridman (1:27:28.800)
Now suppose you are China and you have cheated and secretly
Max Tegmark (1:27:34.400)
developed some clandestine little thing
Lex Fridman (1:27:36.000)
or you're thinking about doing it.
Lex Fridman (1:27:37.560)
What's your calculation that you do?
Lex Fridman (1:27:39.200)
Well, you're like, OK, what's the probability
Lex Fridman (1:27:41.880)
that we're going to get caught?
Lex Fridman (1:27:44.920)
If the probability is 100%, of course, we're not going to do it.
Lex Fridman (1:27:49.120)
But if the probability is 5% that we're going to get caught,
Lex Fridman (1:27:52.720)
then it's going to be like a huge embarrassment for us.
Lex Fridman (1:27:55.560)
And we still have our nuclear weapons anyway,
Lex Fridman (1:28:00.120)
so it doesn't really make an enormous difference in terms
Max Tegmark (1:28:05.160)
of deterring the US.
Lex Fridman (1:28:07.520)
And that feeds the stigma that you kind of established,
Max Tegmark (1:28:11.640)
like this fabric, this universal stigma over the thing.
Lex Fridman (1:28:14.720)
Exactly.
Max Tegmark (1:28:15.520)
It's very reasonable for them to say, well, we probably
Lex Fridman (1:28:18.080)
get away with it.
Max Tegmark (1:28:18.800)
If we don't, then the US will know we cheated,
Lex Fridman (1:28:21.320)
and then they're going to go full tilt with their program
Lex Fridman (1:28:23.720)
and say, look, the Chinese are cheaters,
Lex Fridman (1:28:25.000)
and now we have all these weapons against us,
Lex Fridman (1:28:27.080)
and that's bad.
Lex Fridman (1:28:27.920)
So the stigma alone is very, very powerful.
Lex Fridman (1:28:32.160)
And again, look what happened with bioweapons.
Lex Fridman (1:28:34.520)
It's been 50 years now.
Lex Fridman (1:28:36.880)
When was the last time you read about a bioterrorism attack?
Lex Fridman (1:28:40.120)
The only deaths I really know about with bioweapons
Max Tegmark (1:28:42.680)
that have happened when we Americans managed
Lex Fridman (1:28:45.200)
to kill some of our own with anthrax,
Max Tegmark (1:28:47.200)
or the idiot who sent them to Tom Daschle and others
Lex Fridman (1:28:49.760)
in letters, right?
Lex Fridman (1:28:50.880)
And similarly in Sverdlovsk in the Soviet Union,
Lex Fridman (1:28:55.960)
they had some anthrax in some lab there.
Max Tegmark (1:28:57.960)
Maybe they were cheating or who knows,
Lex Fridman (1:29:00.000)
and it leaked out and killed a bunch of Russians.
Lex Fridman (1:29:02.520)
I'd say that's a pretty good success, right?
Lex Fridman (1:29:04.480)
50 years, just two own goals by the superpowers,
Lex Fridman (1:29:08.360)
and then nothing.
Lex Fridman (1:29:09.560)
And that's why whenever I ask anyone
Lex Fridman (1:29:12.120)
what they think about biology, they think it's great.
Lex Fridman (1:29:15.160)
They associate it with new cures, new diseases,
Max Tegmark (1:29:18.080)
maybe a good vaccine.
Lex Fridman (1:29:19.720)
This is how I want to think about AI in the future.
Lex Fridman (1:29:22.680)
And I want others to think about AI too,
Lex Fridman (1:29:24.840)
as a source of all these great solutions to our problems,
Max Tegmark (1:29:27.840)
not as, oh, AI, oh yeah, that's the reason
Lex Fridman (1:29:31.920)
I feel scared going outside these days.
Max Tegmark (1:29:34.600)
Yeah, it's kind of brilliant that bioweapons
Lex Fridman (1:29:37.920)
and nuclear weapons, we've figured out,
Max Tegmark (1:29:40.760)
I mean, of course there's still a huge source of danger,
Lex Fridman (1:29:43.320)
but we figured out some way of creating rules
Lex Fridman (1:29:47.760)
and social stigma over these weapons
Lex Fridman (1:29:51.440)
that then creates a stability to our,
Max Tegmark (1:29:54.600)
whatever that game theoretic stability that occurs.
Lex Fridman (1:29:57.640)
And we don't have that with AI,
Lex Fridman (1:29:59.200)
and you're kind of screaming from the top of the mountain
Lex Fridman (1:30:03.760)
about this, that we need to find that
Max Tegmark (1:30:05.520)
because it's very possible with the future of life,
Lex Fridman (1:30:10.520)
as you point out, Institute Awards pointed out
Max Tegmark (1:30:15.000)
that with nuclear weapons,
Lex Fridman (1:30:17.920)
we could have destroyed ourselves quite a few times.
Lex Fridman (1:30:21.040)
And it's a learning experience that is very costly.
Lex Fridman (1:30:28.520)
We gave this Future Life Award,
Max Tegmark (1:30:30.960)
we gave it the first time to this guy, Vasily Arkhipov.
Lex Fridman (1:30:34.640)
He was on, most people haven't even heard of him.
Lex Fridman (1:30:37.480)
Yeah, can you say who he is?
Lex Fridman (1:30:38.640)
Vasily Arkhipov, he has, in my opinion,
Max Tegmark (1:30:44.080)
made the greatest positive contribution to humanity
Lex Fridman (1:30:47.480)
of any human in modern history.
Lex Fridman (1:30:50.200)
And maybe it sounds like hyperbole here,
Lex Fridman (1:30:51.880)
like I'm just over the top,
Lex Fridman (1:30:53.320)
but let me tell you the story and I think maybe you'll agree.
Lex Fridman (1:30:56.080)
So during the Cuban Missile Crisis,
Max Tegmark (1:31:00.000)
we Americans first didn't know
Lex Fridman (1:31:01.800)
that the Russians had sent four submarines,
Lex Fridman (1:31:05.160)
but we caught two of them.
Lex Fridman (1:31:06.720)
And we didn't know that,
Lex Fridman (1:31:09.160)
so we dropped practice depth charges
Lex Fridman (1:31:11.040)
on the one that he was on,
Max Tegmark (1:31:12.360)
try to force it to the surface.
Lex Fridman (1:31:15.440)
But we didn't know that this nuclear submarine
Max Tegmark (1:31:17.680)
actually was a nuclear submarine with a nuclear torpedo.
Lex Fridman (1:31:20.560)
We also didn't know that they had authorization
Max Tegmark (1:31:22.640)
to launch it without clearance from Moscow.
Lex Fridman (1:31:25.120)
And we also didn't know
Max Tegmark (1:31:26.040)
that they were running out of electricity.
Lex Fridman (1:31:28.240)
Their batteries were almost dead.
Max Tegmark (1:31:29.880)
They were running out of oxygen.
Lex Fridman (1:31:31.840)
Sailors were fainting left and right.
Max Tegmark (1:31:34.280)
The temperature was about 110, 120 Fahrenheit on board.
Lex Fridman (1:31:39.120)
It was really hellish conditions,
Max Tegmark (1:31:40.920)
really just a kind of doomsday.
Lex Fridman (1:31:43.240)
And at that point,
Max Tegmark (1:31:44.520)
these giant explosions start happening
Lex Fridman (1:31:46.280)
from the Americans dropping these.
Max Tegmark (1:31:48.160)
The captain thought World War III had begun.
Lex Fridman (1:31:50.680)
They decided they were gonna launch the nuclear torpedo.
Lex Fridman (1:31:53.720)
And one of them shouted,
Lex Fridman (1:31:55.360)
we're all gonna die,
Lex Fridman (1:31:56.200)
but we're not gonna disgrace our Navy.
Lex Fridman (1:31:58.920)
We don't know what would have happened
Max Tegmark (1:32:00.120)
if there had been a giant mushroom cloud all of a sudden
Lex Fridman (1:32:03.400)
against the Americans.
Lex Fridman (1:32:04.760)
But since everybody had their hands on the triggers,
Lex Fridman (1:32:09.200)
you don't have to be too creative to think
Max Tegmark (1:32:10.800)
that it could have led to an all out nuclear war,
Lex Fridman (1:32:13.080)
in which case we wouldn't be having this conversation now.
Lex Fridman (1:32:15.680)
What actually took place was
Lex Fridman (1:32:17.600)
they needed three people to approve this.
Max Tegmark (1:32:21.040)
The captain had said yes.
Lex Fridman (1:32:22.200)
There was the Communist Party political officer.
Max Tegmark (1:32:24.120)
He also said, yes, let's do it.
Lex Fridman (1:32:26.000)
And the third man was this guy, Vasily Arkhipov,
Max Tegmark (1:32:29.040)
who said, no.
Lex Fridman (1:32:29.880)
For some reason, he was just more chill than the others
Lex Fridman (1:32:32.720)
and he was the right man at the right time.
Lex Fridman (1:32:34.240)
I don't want us as a species rely on the right person
Max Tegmark (1:32:38.120)
being there at the right time, you know.
Lex Fridman (1:32:40.720)
We tracked down his family
Max Tegmark (1:32:42.920)
living in relative poverty outside Moscow.
Lex Fridman (1:32:47.320)
When he flew his daughter,
Max Tegmark (1:32:48.800)
he had passed away and flew them to London.
Lex Fridman (1:32:52.720)
They had never been to the West even.
Max Tegmark (1:32:54.000)
It was incredibly moving to get to honor them for this.
Lex Fridman (1:32:57.160)
The next year we gave them a medal.
Max Tegmark (1:32:59.320)
The next year we gave this Future Life Award
Lex Fridman (1:33:01.800)
to Stanislav Petrov.
Lex Fridman (1:33:04.160)
Have you heard of him?
Lex Fridman (1:33:05.000)
Yes.
Lex Fridman (1:33:05.840)
So he was in charge of the Soviet early warning station,
Lex Fridman (1:33:10.000)
which was built with Soviet technology
Lex Fridman (1:33:12.880)
and honestly not that reliable.
Lex Fridman (1:33:14.760)
It said that there were five US missiles coming in.
Max Tegmark (1:33:18.280)
Again, if they had launched at that point,
Lex Fridman (1:33:21.440)
we probably wouldn't be having this conversation.
Max Tegmark (1:33:23.440)
He decided based on just mainly gut instinct
Lex Fridman (1:33:29.440)
to just not escalate this.
Lex Fridman (1:33:32.640)
And I'm very glad he wasn't replaced by an AI
Lex Fridman (1:33:35.200)
that was just automatically following orders.
Lex Fridman (1:33:37.600)
And then we gave the third one to Matthew Messelson.
Lex Fridman (1:33:39.840)
Last year, we gave this award to these guys
Max Tegmark (1:33:44.360)
who actually use technology for good,
Lex Fridman (1:33:46.760)
not avoiding something bad, but for something good.
Max Tegmark (1:33:50.120)
The guys who eliminated this disease,
Lex Fridman (1:33:52.120)
it was way worse than COVID that had killed
Max Tegmark (1:33:55.440)
half a billion people in its final century.
Lex Fridman (1:33:58.520)
Smallpox, right?
Lex Fridman (1:33:59.440)
So you mentioned it earlier.
Lex Fridman (1:34:01.200)
COVID on average kills less than 1% of people who get it.
Max Tegmark (1:34:05.320)
Smallpox, about 30%.
Lex Fridman (1:34:08.240)
And they just ultimately, Viktor Zhdanov and Bill Foege,
Max Tegmark (1:34:14.160)
most of my colleagues have never heard of either of them,
Lex Fridman (1:34:17.560)
one American, one Russian, they did this amazing effort
Max Tegmark (1:34:22.080)
not only was Zhdanov able to get the US and the Soviet Union
Lex Fridman (1:34:25.200)
to team up against smallpox during the Cold War,
Lex Fridman (1:34:27.920)
but Bill Foege came up with this ingenious strategy
Lex Fridman (1:34:30.320)
for making it actually go all the way
Max Tegmark (1:34:32.840)
to defeat the disease without funding
Lex Fridman (1:34:36.560)
for vaccinating everyone.
Lex Fridman (1:34:37.600)
And as a result, we haven't had any,
Lex Fridman (1:34:40.040)
we went from 15 million deaths the year
Max Tegmark (1:34:42.680)
I was born in smallpox.
Lex Fridman (1:34:44.280)
So what do we have in COVID now?
Lex Fridman (1:34:45.640)
A little bit short of 2 million, right?
Lex Fridman (1:34:47.240)
Yes.
Max Tegmark (1:34:48.120)
To zero deaths, of course, this year and forever.
Lex Fridman (1:34:52.040)
There have been 200 million people,
Max Tegmark (1:34:53.960)
we estimate, who would have died since then by smallpox
Lex Fridman (1:34:57.200)
had it not been for this.
Lex Fridman (1:34:58.080)
So isn't science awesome when you use it for good?
Lex Fridman (1:35:02.160)
The reason we wanna celebrate these sort of people
Max Tegmark (1:35:04.280)
is to remind them of this.
Lex Fridman (1:35:05.680)
Science is so awesome when you use it for good.
Lex Fridman (1:35:10.160)
And those awards actually, the variety there,
Lex Fridman (1:35:13.520)
it's a very interesting picture.
Lex Fridman (1:35:14.920)
So the first two are looking at,
Lex Fridman (1:35:19.360)
it's kind of exciting to think that these average humans
Max Tegmark (1:35:22.680)
in some sense, they're products of billions
Lex Fridman (1:35:26.200)
of other humans that came before them, evolution,
Lex Fridman (1:35:30.200)
and some little, you said gut,
Lex Fridman (1:35:33.360)
but there's something in there
Max Tegmark (1:35:35.320)
that stopped the annihilation of the human race.
Lex Fridman (1:35:41.080)
And that's a magical thing,
Lex Fridman (1:35:43.040)
but that's like this deeply human thing.
Lex Fridman (1:35:45.240)
And then there's the other aspect
Max Tegmark (1:35:47.400)
where that's also very human,
Lex Fridman (1:35:49.800)
which is to build solution
Max Tegmark (1:35:51.440)
to the existential crises that we're facing,
Lex Fridman (1:35:55.240)
like to build it, to take the responsibility
Lex Fridman (1:35:57.520)
and to come up with different technologies and so on.
Lex Fridman (1:36:00.600)
And both of those are deeply human,
Max Tegmark (1:36:04.080)
the gut and the mind, whatever that is that creates.
Lex Fridman (1:36:07.400)
The best is when they work together.
Max Tegmark (1:36:08.640)
Arkhipov, I wish I could have met him, of course,
Lex Fridman (1:36:11.400)
but he had passed away.
Max Tegmark (1:36:13.200)
He was really a fantastic military officer,
Lex Fridman (1:36:16.720)
combining all the best traits
Max Tegmark (1:36:18.680)
that we in America admire in our military.
Lex Fridman (1:36:21.000)
Because first of all, he was very loyal, of course.
Max Tegmark (1:36:23.160)
He never even told anyone about this during his whole life,
Lex Fridman (1:36:26.280)
even though you think he had some bragging rights, right?
Lex Fridman (1:36:28.440)
But he just was like, this is just business,
Lex Fridman (1:36:30.000)
just doing my job.
Max Tegmark (1:36:31.560)
It only came out later after his death.
Lex Fridman (1:36:34.320)
And second, the reason he did the right thing
Max Tegmark (1:36:37.120)
was not because he was some sort of liberal
Lex Fridman (1:36:39.240)
or some sort of, not because he was just,
Max Tegmark (1:36:43.960)
oh, peace and love.
Lex Fridman (1:36:47.360)
It was partly because he had been the captain
Max Tegmark (1:36:49.800)
on another submarine that had a nuclear reactor meltdown.
Lex Fridman (1:36:53.080)
And it was his heroism that helped contain this.
Max Tegmark (1:36:58.000)
That's why he died of cancer later also.
Lex Fridman (1:36:59.760)
But he had seen many of his crew members die.
Lex Fridman (1:37:01.480)
And I think for him, that gave him this gut feeling
Lex Fridman (1:37:04.160)
that if there's a nuclear war
Max Tegmark (1:37:06.200)
between the US and the Soviet Union,
Lex Fridman (1:37:08.400)
the whole world is gonna go through
Lex Fridman (1:37:11.080)
what I saw my dear crew members suffer through.
Lex Fridman (1:37:13.760)
It wasn't just an abstract thing for him.
Max Tegmark (1:37:15.840)
I think it was real.
Lex Fridman (1:37:17.680)
And second though, not just the gut, the mind, right?
Max Tegmark (1:37:20.640)
He was, for some reason, very levelheaded personality
Lex Fridman (1:37:23.960)
and very smart guy,
Max Tegmark (1:37:25.960)
which is exactly what we want our best fighter pilots
Lex Fridman (1:37:29.240)
to be also, right?
Max Tegmark (1:37:30.120)
I never forget Neil Armstrong when he's landing on the moon
Lex Fridman (1:37:32.880)
and almost running out of gas.
Lex Fridman (1:37:34.560)
And he doesn't even change when they say 30 seconds,
Lex Fridman (1:37:37.440)
he doesn't even change the tone of voice, just keeps going.
Max Tegmark (1:37:39.680)
Arkhipov, I think was just like that.
Lex Fridman (1:37:41.840)
So when the explosions start going off
Lex Fridman (1:37:43.480)
and his captain is screaming and we should nuke them
Lex Fridman (1:37:45.520)
and all, he's like,
Max Tegmark (1:37:50.960)
I don't think the Americans are trying to sink us.
Lex Fridman (1:37:54.280)
I think they're trying to send us a message.
Max Tegmark (1:37:58.080)
That's pretty bad ass.
Lex Fridman (1:37:59.200)
Yes.
Max Tegmark (1:38:00.040)
Coolness, because he said, if they wanted to sink us,
Lex Fridman (1:38:03.680)
and he said, listen, listen, it's alternating
Max Tegmark (1:38:06.920)
one loud explosion on the left, one on the right,
Lex Fridman (1:38:10.160)
one on the left, one on the right.
Max Tegmark (1:38:12.120)
He was the only one who noticed this pattern.
Lex Fridman (1:38:15.840)
And he's like, I think this is,
Max Tegmark (1:38:17.880)
I'm trying to send us a signal
Lex Fridman (1:38:20.640)
that they want it to surface
Lex Fridman (1:38:22.840)
and they're not gonna sink us.
Lex Fridman (1:38:25.800)
And somehow,
Max Tegmark (1:38:29.320)
this is how he then managed it ultimately
Lex Fridman (1:38:32.160)
with his combination of gut
Lex Fridman (1:38:34.640)
and also just cool analytical thinking,
Lex Fridman (1:38:37.960)
was able to deescalate the whole thing.
Lex Fridman (1:38:40.120)
And yeah, so this is some of the best in humanity.
Lex Fridman (1:38:44.240)
I guess coming back to what we talked about earlier,
Max Tegmark (1:38:45.880)
it's the combination of the neural network,
Lex Fridman (1:38:47.400)
the instinctive, with, I'm getting teary up here,
Max Tegmark (1:38:50.960)
getting emotional, but he was just,
Lex Fridman (1:38:53.240)
he is one of my superheroes,
Max Tegmark (1:38:56.120)
having both the heart and the mind combined.
Lex Fridman (1:39:00.440)
And especially in that time, there's something about the,
Max Tegmark (1:39:03.760)
I mean, this is a very, in America,
Lex Fridman (1:39:05.440)
people are used to this kind of idea
Max Tegmark (1:39:06.880)
of being the individual of like on your own thinking.
Lex Fridman (1:39:12.040)
I think under, in the Soviet Union under communism,
Max Tegmark (1:39:15.480)
it's actually much harder to do that.
Lex Fridman (1:39:17.600)
Oh yeah, he didn't even, he even got,
Max Tegmark (1:39:19.960)
he didn't get any accolades either
Lex Fridman (1:39:21.840)
when he came back for this, right?
Max Tegmark (1:39:24.240)
They just wanted to hush the whole thing up.
Lex Fridman (1:39:25.880)
Yeah, there's echoes of that with Chernobyl,
Max Tegmark (1:39:28.000)
there's all kinds of,
Lex Fridman (1:39:30.920)
that's one, that's a really hopeful thing
Max Tegmark (1:39:34.400)
that amidst big centralized powers,
Lex Fridman (1:39:37.520)
whether it's companies or states,
Max Tegmark (1:39:39.920)
there's still the power of the individual
Lex Fridman (1:39:42.480)
to think on their own, to act.
Lex Fridman (1:39:43.880)
But I think we need to think of people like this,
Lex Fridman (1:39:46.880)
not as a panacea we can always count on,
Lex Fridman (1:39:50.160)
but rather as a wake up call.
Lex Fridman (1:39:55.720)
So because of them, because of Arkhipov,
Max Tegmark (1:39:58.560)
we are alive to learn from this lesson,
Lex Fridman (1:40:01.320)
to learn from the fact that we shouldn't keep playing
Max Tegmark (1:40:03.120)
Russian roulette and almost have a nuclear war
Lex Fridman (1:40:04.840)
by mistake now and then,
Max Tegmark (1:40:06.600)
because relying on luck is not a good longterm strategy.
Lex Fridman (1:40:09.600)
If you keep playing Russian roulette over and over again,
Max Tegmark (1:40:11.360)
the probability of surviving just drops exponentially
Lex Fridman (1:40:13.560)
with time.
Max Tegmark (1:40:14.400)
Yeah.
Lex Fridman (1:40:15.240)
And if you have some probability
Max Tegmark (1:40:16.680)
of having an accidental nuke war every year,
Lex Fridman (1:40:18.640)
the probability of not having one also drops exponentially.
Max Tegmark (1:40:21.200)
I think we can do better than that.
Lex Fridman (1:40:22.840)
So I think the message is very clear,
Max Tegmark (1:40:26.000)
once in a while shit happens,
Lex Fridman (1:40:27.840)
and there's a lot of very concrete things we can do
Max Tegmark (1:40:31.320)
to reduce the risk of things like that happening
Lex Fridman (1:40:34.920)
in the first place.
Max Tegmark (1:40:36.520)
On the AI front, if we just link on that for a second.
Lex Fridman (1:40:39.600)
Yeah.
Lex Fridman (1:40:40.960)
So you're friends with, you often talk with Elon Musk
Lex Fridman (1:40:44.120)
throughout history, you've did a lot
Max Tegmark (1:40:46.680)
of interesting things together.
Lex Fridman (1:40:48.680)
He has a set of fears about the future
Max Tegmark (1:40:52.280)
of artificial intelligence, AGI.
Lex Fridman (1:40:55.840)
Do you have a sense, we've already talked about
Max Tegmark (1:40:59.720)
the things we should be worried about with AI,
Lex Fridman (1:41:01.560)
do you have a sense of the shape of his fears
Max Tegmark (1:41:04.040)
in particular about AI,
Lex Fridman (1:41:06.880)
of which subset of what we've talked about,
Max Tegmark (1:41:10.160)
whether it's creating, it's that direction
Lex Fridman (1:41:14.480)
of creating sort of these giant competition systems
Max Tegmark (1:41:17.520)
that are not explainable,
Lex Fridman (1:41:19.160)
they're not intelligible intelligence,
Max Tegmark (1:41:21.800)
or is it the...
Lex Fridman (1:41:26.720)
And then like as a branch of that,
Max Tegmark (1:41:28.840)
is it the manipulation by big corporations of that
Lex Fridman (1:41:31.840)
or individual evil people to use that for destruction
Lex Fridman (1:41:35.400)
or the unintentional consequences?
Lex Fridman (1:41:37.480)
Do you have a sense of where his thinking is on this?
Max Tegmark (1:41:40.280)
From my many conversations with Elon,
Lex Fridman (1:41:42.440)
yeah, I certainly have a model of how he thinks.
Max Tegmark (1:41:47.400)
It's actually very much like the way I think also,
Lex Fridman (1:41:49.880)
I'll elaborate on it a bit.
Max Tegmark (1:41:51.080)
I just wanna push back on when you said evil people,
Lex Fridman (1:41:54.680)
I don't think it's a very helpful concept.
Max Tegmark (1:41:58.520)
Evil people, sometimes people do very, very bad things,
Lex Fridman (1:42:02.320)
but they usually do it because they think it's a good thing
Max Tegmark (1:42:05.440)
because somehow other people had told them
Lex Fridman (1:42:07.760)
that that was a good thing
Lex Fridman (1:42:08.640)
or given them incorrect information or whatever, right?
Lex Fridman (1:42:15.440)
I believe in the fundamental goodness of humanity
Max Tegmark (1:42:18.400)
that if we educate people well
Lex Fridman (1:42:21.680)
and they find out how things really are,
Max Tegmark (1:42:24.240)
people generally wanna do good and be good.
Lex Fridman (1:42:27.240)
Hence the value alignment,
Max Tegmark (1:42:30.360)
as opposed to it's about information, about knowledge,
Lex Fridman (1:42:33.660)
and then once we have that,
Max Tegmark (1:42:35.320)
we'll likely be able to do good
Lex Fridman (1:42:39.960)
in the way that's aligned with everybody else
Max Tegmark (1:42:41.600)
who thinks differently.
Lex Fridman (1:42:42.440)
Yeah, and it's not just the individual people
Max Tegmark (1:42:44.000)
we have to align.
Lex Fridman (1:42:44.960)
So we don't just want people to be educated
Max Tegmark (1:42:49.600)
to know the way things actually are
Lex Fridman (1:42:51.200)
and to treat each other well,
Lex Fridman (1:42:53.200)
but we also need to align other nonhuman entities.
Lex Fridman (1:42:56.280)
We talked about corporations, there has to be institutions
Lex Fridman (1:42:58.560)
so that what they do is actually good
Lex Fridman (1:42:59.960)
for the country they're in
Lex Fridman (1:43:00.880)
and we should align, make sure that what countries do
Lex Fridman (1:43:03.480)
is actually good for the species as a whole, et cetera.
Max Tegmark (1:43:07.780)
Coming back to Elon,
Lex Fridman (1:43:08.680)
yeah, my understanding of how Elon sees this
Max Tegmark (1:43:13.600)
is really quite similar to my own,
Lex Fridman (1:43:15.240)
which is one of the reasons I like him so much
Lex Fridman (1:43:18.200)
and enjoy talking with him so much.
Lex Fridman (1:43:19.320)
I feel he's quite different from most people
Max Tegmark (1:43:22.960)
in that he thinks much more than most people
Lex Fridman (1:43:27.720)
about the really big picture,
Max Tegmark (1:43:29.840)
not just what's gonna happen in the next election cycle,
Lex Fridman (1:43:32.540)
but in millennia, millions and billions of years from now.
Lex Fridman (1:43:36.840)
And when you look in this more cosmic perspective,
Lex Fridman (1:43:39.280)
it's so obvious that we are gazing out into this universe
Max Tegmark (1:43:43.080)
that as far as we can tell is mostly dead
Lex Fridman (1:43:46.280)
with life being almost imperceptibly tiny perturbation,
Lex Fridman (1:43:49.800)
and he sees this enormous opportunity
Lex Fridman (1:43:52.640)
for our universe to come alive,
Max Tegmark (1:43:54.280)
first to become an interplanetary species.
Lex Fridman (1:43:56.480)
Mars is obviously just first stop on this cosmic journey.
Lex Fridman (1:44:02.120)
And precisely because he thinks more long term,
Lex Fridman (1:44:06.760)
it's much more clear to him than to most people
Max Tegmark (1:44:09.560)
that what we do with this Russian roulette thing
Lex Fridman (1:44:11.340)
we keep playing with our nukes is a really poor strategy,
Max Tegmark (1:44:15.320)
really reckless strategy.
Lex Fridman (1:44:16.720)
And also that we're just building
Max Tegmark (1:44:18.620)
these ever more powerful AI systems that we don't understand
Lex Fridman (1:44:21.640)
is also just a really reckless strategy.
Max Tegmark (1:44:23.840)
I feel Elon is very much a humanist
Lex Fridman (1:44:26.640)
in the sense that he wants an awesome future for humanity.
Max Tegmark (1:44:30.880)
He wants it to be us that control the machines
Lex Fridman (1:44:35.960)
rather than the machines that control us.
Lex Fridman (1:44:39.400)
And why shouldn't we insist on that?
Lex Fridman (1:44:42.080)
We're building them after all, right?
Lex Fridman (1:44:44.560)
Why should we build things that just make us
Lex Fridman (1:44:46.520)
into some little cog in the machinery
Lex Fridman (1:44:48.440)
that has no further say in the matter, right?
Lex Fridman (1:44:50.240)
That's not my idea of an inspiring future either.
Max Tegmark (1:44:54.560)
Yeah, if you think on the cosmic scale
Lex Fridman (1:44:57.880)
in terms of both time and space,
Lex Fridman (1:45:00.720)
so much is put into perspective.
Lex Fridman (1:45:02.600)
Yeah.
Max Tegmark (1:45:04.220)
Whenever I have a bad day, that's what I think about.
Lex Fridman (1:45:06.440)
It immediately makes me feel better.
Max Tegmark (1:45:09.200)
It makes me sad that for us individual humans,
Lex Fridman (1:45:13.520)
at least for now, the ride ends too quickly.
Max Tegmark (1:45:16.400)
That we don't get to experience the cosmic scale.
Lex Fridman (1:45:20.080)
Yeah, I mean, I think of our universe sometimes
Max Tegmark (1:45:22.280)
as an organism that has only begun to wake up a tiny bit,
Lex Fridman (1:45:26.080)
just like the very first little glimmers of consciousness
Max Tegmark (1:45:30.120)
you have in the morning when you start coming around.
Lex Fridman (1:45:32.120)
Before the coffee.
Max Tegmark (1:45:33.160)
Before the coffee, even before you get out of bed,
Lex Fridman (1:45:35.880)
before you even open your eyes.
Max Tegmark (1:45:37.280)
You start to wake up a little bit.
Lex Fridman (1:45:40.320)
There's something here.
Max Tegmark (1:45:43.440)
That's very much how I think of where we are.
Lex Fridman (1:45:47.120)
All those galaxies out there,
Max Tegmark (1:45:48.600)
I think they're really beautiful,
Lex Fridman (1:45:51.160)
but why are they beautiful?
Max Tegmark (1:45:52.840)
They're beautiful because conscious entities
Lex Fridman (1:45:55.040)
are actually observing them,
Max Tegmark (1:45:57.000)
experiencing them through our telescopes.
Lex Fridman (1:46:01.720)
I define consciousness as subjective experience,
Max Tegmark (1:46:05.880)
whether it be colors or emotions or sounds.
Lex Fridman (1:46:09.420)
So beauty is an experience.
Max Tegmark (1:46:12.340)
Meaning is an experience.
Lex Fridman (1:46:13.800)
Purpose is an experience.
Max Tegmark (1:46:15.880)
If there was no conscious experience,
Lex Fridman (1:46:18.000)
observing these galaxies, they wouldn't be beautiful.
Max Tegmark (1:46:20.320)
If we do something dumb with advanced AI in the future here
Lex Fridman (1:46:24.960)
and Earth originating, life goes extinct.
Lex Fridman (1:46:29.360)
And that was it for this.
Lex Fridman (1:46:30.480)
If there is nothing else with telescopes in our universe,
Max Tegmark (1:46:33.560)
then it's kind of game over for beauty
Lex Fridman (1:46:36.600)
and meaning and purpose in our whole universe.
Lex Fridman (1:46:38.120)
And I think that would be just such
Lex Fridman (1:46:39.880)
an opportunity lost, frankly.
Lex Fridman (1:46:41.800)
And I think when Elon points this out,
Lex Fridman (1:46:46.080)
he gets very unfairly maligned in the media
Max Tegmark (1:46:49.640)
for all the dumb media bias reasons we talked about.
Lex Fridman (1:46:52.440)
They want to print precisely the things about Elon
Max Tegmark (1:46:55.680)
out of context that are really click baity.
Lex Fridman (1:46:58.800)
He has gotten so much flack
Max Tegmark (1:47:00.440)
for this summoning the demon statement.
Lex Fridman (1:47:04.720)
I happen to know exactly the context
Max Tegmark (1:47:07.680)
because I was in the front row when he gave that talk.
Lex Fridman (1:47:09.720)
It was at MIT, you'll be pleased to know,
Max Tegmark (1:47:11.280)
it was the AeroAstro anniversary.
Lex Fridman (1:47:13.880)
They had Buzz Aldrin there from the moon landing,
Max Tegmark (1:47:16.800)
a whole house, a Kresge auditorium
Lex Fridman (1:47:19.000)
packed with MIT students.
Lex Fridman (1:47:20.840)
And he had this amazing Q&A, it might've gone for an hour.
Lex Fridman (1:47:23.920)
And they talked about rockets and Mars and everything.
Max Tegmark (1:47:27.160)
At the very end, this one student
Lex Fridman (1:47:29.600)
who has actually hit my class asked him, what about AI?
Max Tegmark (1:47:33.200)
Elon makes this one comment
Lex Fridman (1:47:35.240)
and they take this out of context, print it, goes viral.
Lex Fridman (1:47:39.440)
What is it like with AI,
Lex Fridman (1:47:40.600)
we're summoning the demons, something like that.
Lex Fridman (1:47:42.920)
And try to cast him as some sort of doom and gloom dude.
Lex Fridman (1:47:47.480)
You know Elon, he's not the doom and gloom dude.
Max Tegmark (1:47:51.960)
He is such a positive visionary.
Lex Fridman (1:47:54.000)
And the whole reason he warns about this
Max Tegmark (1:47:55.680)
is because he realizes more than most
Lex Fridman (1:47:57.720)
what the opportunity cost is of screwing up.
Max Tegmark (1:47:59.880)
That there is so much awesomeness in the future
Lex Fridman (1:48:02.360)
that we can and our descendants can enjoy
Lex Fridman (1:48:05.480)
if we don't screw up, right?
Lex Fridman (1:48:07.760)
I get so pissed off when people try to cast him
Max Tegmark (1:48:10.320)
as some sort of technophobic Luddite.
Lex Fridman (1:48:15.320)
And at this point, it's kind of ludicrous
Max Tegmark (1:48:18.480)
when I hear people say that people who worry about
Lex Fridman (1:48:21.640)
artificial general intelligence are Luddites
Max Tegmark (1:48:24.560)
because of course, if you look more closely,
Lex Fridman (1:48:27.000)
you have some of the most outspoken people making warnings
Max Tegmark (1:48:32.920)
are people like Professor Stuart Russell from Berkeley
Lex Fridman (1:48:35.640)
who's written the bestselling AI textbook, you know.
Lex Fridman (1:48:38.360)
So claiming that he's a Luddite who doesn't understand AI
Lex Fridman (1:48:43.360)
is the joke is really on the people who said it.
Lex Fridman (1:48:46.520)
But I think more broadly,
Lex Fridman (1:48:48.200)
this message is really not sunk in at all.
Lex Fridman (1:48:50.800)
What it is that people worry about,
Lex Fridman (1:48:52.640)
they think that Elon and Stuart Russell and others
Max Tegmark (1:48:56.680)
are worried about the dancing robots picking up an AR 15
Lex Fridman (1:49:02.280)
and going on a rampage, right?
Max Tegmark (1:49:04.360)
They think they're worried about robots turning evil.
Lex Fridman (1:49:08.440)
They're not, I'm not.
Max Tegmark (1:49:10.360)
The risk is not malice, it's competence.
Lex Fridman (1:49:15.880)
The risk is just that we build some systems
Max Tegmark (1:49:17.560)
that are incredibly competent,
Lex Fridman (1:49:18.760)
which means they're always gonna get
Max Tegmark (1:49:20.040)
their goals accomplished,
Lex Fridman (1:49:22.000)
even if they clash with our goals.
Max Tegmark (1:49:24.080)
That's the risk.
Lex Fridman (1:49:25.920)
Why did we humans drive the West African black rhino extinct?
Lex Fridman (1:49:30.920)
Is it because we're malicious, evil rhinoceros haters?
Lex Fridman (1:49:34.840)
No, it's just because our goals didn't align
Max Tegmark (1:49:38.000)
with the goals of those rhinos
Lex Fridman (1:49:39.240)
and tough luck for the rhinos, you know.
Lex Fridman (1:49:42.360)
So the point is just we don't wanna put ourselves
Lex Fridman (1:49:46.720)
in the position of those rhinos
Max Tegmark (1:49:48.120)
creating something more powerful than us
Lex Fridman (1:49:51.240)
if we haven't first figured out how to align the goals.
Lex Fridman (1:49:53.880)
And I am optimistic.
Lex Fridman (1:49:54.920)
I think we could do it if we worked really hard on it,
Max Tegmark (1:49:56.880)
because I spent a lot of time
Lex Fridman (1:49:59.200)
around intelligent entities that were more intelligent
Max Tegmark (1:50:01.800)
than me, my mom and my dad.
Lex Fridman (1:50:05.960)
And I was little and that was fine
Max Tegmark (1:50:07.560)
because their goals were actually aligned
Lex Fridman (1:50:09.160)
with mine quite well.
Lex Fridman (1:50:11.280)
But we've seen today many examples of where the goals
Lex Fridman (1:50:15.440)
of our powerful systems are not so aligned.
Lex Fridman (1:50:17.200)
So those click through optimization algorithms
Lex Fridman (1:50:22.960)
that are polarized social media, right?
Max Tegmark (1:50:24.560)
They were actually pretty poorly aligned
Lex Fridman (1:50:26.160)
with what was good for democracy, it turned out.
Lex Fridman (1:50:28.760)
And again, almost all problems we've had
Lex Fridman (1:50:31.520)
in the machine learning again came so far,
Max Tegmark (1:50:33.640)
not from malice, but from poor alignment.
Lex Fridman (1:50:35.520)
And that's exactly why that's why we should be concerned
Max Tegmark (1:50:38.240)
about it in the future.
Lex Fridman (1:50:39.320)
Do you think it's possible that with systems
Max Tegmark (1:50:43.240)
like Neuralink and brain computer interfaces,
Lex Fridman (1:50:47.320)
you know, again, thinking of the cosmic scale,
Max Tegmark (1:50:49.280)
Elon's talked about this, but others have as well
Lex Fridman (1:50:52.600)
throughout history of figuring out how the exact mechanism
Max Tegmark (1:50:57.240)
of how to achieve that kind of alignment.
Lex Fridman (1:51:00.000)
So one of them is having a symbiosis with AI,
Max Tegmark (1:51:03.160)
which is like coming up with clever ways
Lex Fridman (1:51:05.560)
where we're like stuck together in this weird relationship,
Max Tegmark (1:51:10.360)
whether it's biological or in some kind of other way.
Lex Fridman (1:51:14.200)
Do you think that's a possibility
Lex Fridman (1:51:17.240)
of having that kind of symbiosis?
Lex Fridman (1:51:19.200)
Or do we wanna instead kind of focus
Max Tegmark (1:51:20.960)
on this distinct entities of us humans talking
Lex Fridman (1:51:28.200)
to these intelligible, self doubting AIs,
Max Tegmark (1:51:31.720)
maybe like Stuart Russell thinks about it,
Lex Fridman (1:51:33.600)
like we're self doubting and full of uncertainty
Lex Fridman (1:51:37.640)
and our AI systems are full of uncertainty.
Lex Fridman (1:51:39.760)
We communicate back and forth
Lex Fridman (1:51:41.520)
and in that way achieve symbiosis.
Lex Fridman (1:51:44.680)
I honestly don't know.
Max Tegmark (1:51:46.200)
I would say that because we don't know for sure
Lex Fridman (1:51:48.600)
what if any of our, which of any of our ideas will work.
Lex Fridman (1:51:52.200)
But we do know that if we don't,
Lex Fridman (1:51:55.200)
I'm pretty convinced that if we don't get any
Max Tegmark (1:51:56.880)
of these things to work and just barge ahead,
Lex Fridman (1:51:59.840)
then our species is, you know,
Max Tegmark (1:52:01.440)
probably gonna go extinct this century.
Lex Fridman (1:52:03.720)
I think it's...
Max Tegmark (1:52:04.600)
This century, you think like,
Lex Fridman (1:52:06.320)
you think we're facing this crisis
Max Tegmark (1:52:09.720)
is a 21st century crisis.
Lex Fridman (1:52:11.320)
Like this century will be remembered.
Lex Fridman (1:52:13.520)
But on a hard drive and a hard drive somewhere
Lex Fridman (1:52:18.720)
or maybe by future generations is like,
Max Tegmark (1:52:22.280)
like there'll be future Future of Life Institute awards
Lex Fridman (1:52:26.240)
for people that have done something about AI.
Max Tegmark (1:52:30.640)
It could also end even worse,
Lex Fridman (1:52:31.880)
whether we're not superseded
Max Tegmark (1:52:33.720)
by leaving any AI behind either.
Lex Fridman (1:52:35.280)
We just totally wipe out, you know,
Max Tegmark (1:52:37.040)
like on Easter Island.
Lex Fridman (1:52:38.480)
Our century is long.
Lex Fridman (1:52:39.880)
You know, there are still 79 years left of it, right?
Lex Fridman (1:52:44.280)
Think about how far we've come just in the last 30 years.
Lex Fridman (1:52:47.680)
So we can talk more about what might go wrong,
Lex Fridman (1:52:53.080)
but you asked me this really good question
Max Tegmark (1:52:54.600)
about what's the best strategy.
Lex Fridman (1:52:55.800)
Is it Neuralink or Russell's approach or whatever?
Max Tegmark (1:52:59.800)
I think, you know, when we did the Manhattan project,
Lex Fridman (1:53:05.480)
we didn't know if any of our four ideas
Max Tegmark (1:53:08.480)
for enriching uranium and getting out the uranium 235
Lex Fridman (1:53:11.760)
were gonna work.
Lex Fridman (1:53:12.880)
But we felt this was really important
Lex Fridman (1:53:14.800)
to get it before Hitler did.
Lex Fridman (1:53:16.680)
So, you know what we did?
Lex Fridman (1:53:17.520)
We tried all four of them.
Max Tegmark (1:53:19.520)
Here, I think it's analogous
Lex Fridman (1:53:21.960)
where there's the greatest threat
Max Tegmark (1:53:24.360)
that's ever faced our species.
Lex Fridman (1:53:25.920)
And of course, US national security by implication.
Max Tegmark (1:53:29.240)
We don't know if we don't have any method
Lex Fridman (1:53:31.480)
that's guaranteed to work, but we have a lot of ideas.
Lex Fridman (1:53:34.680)
So we should invest pretty heavily
Lex Fridman (1:53:35.960)
in pursuing all of them with an open mind
Lex Fridman (1:53:38.040)
and hope that one of them at least works.
Lex Fridman (1:53:40.560)
These are, the good news is the century is long,
Lex Fridman (1:53:45.360)
and it might take decades
Lex Fridman (1:53:47.880)
until we have artificial general intelligence.
Lex Fridman (1:53:50.160)
So we have some time hopefully,
Lex Fridman (1:53:52.760)
but it takes a long time to solve
Max Tegmark (1:53:55.240)
these very, very difficult problems.
Lex Fridman (1:53:57.120)
It's gonna actually be the,
Max Tegmark (1:53:58.080)
it's the most difficult problem
Lex Fridman (1:53:59.160)
we were ever trying to solve as a species.
Lex Fridman (1:54:01.320)
So we have to start now.
Lex Fridman (1:54:03.400)
So we don't have, rather than begin thinking about it
Max Tegmark (1:54:05.840)
the night before some people who've had too much Red Bull
Lex Fridman (1:54:08.720)
switch it on.
Lex Fridman (1:54:09.560)
And we have to, coming back to your question,
Lex Fridman (1:54:11.840)
we have to pursue all of these different avenues and see.
Max Tegmark (1:54:14.240)
If you were my investment advisor
Lex Fridman (1:54:16.800)
and I was trying to invest in the future,
Lex Fridman (1:54:19.920)
how do you think the human species
Lex Fridman (1:54:23.040)
is most likely to destroy itself in the century?
Max Tegmark (1:54:29.440)
Yeah, so if the crises,
Lex Fridman (1:54:32.120)
many of the crises we're facing are really before us
Max Tegmark (1:54:34.680)
within the next hundred years,
Lex Fridman (1:54:37.160)
how do we make explicit,
Max Tegmark (1:54:42.320)
make known the unknowns and solve those problems
Lex Fridman (1:54:46.640)
to avoid the biggest,
Lex Fridman (1:54:49.560)
starting with the biggest existential crisis?
Lex Fridman (1:54:51.920)
So as your investment advisor,
Lex Fridman (1:54:53.160)
how are you planning to make money on us
Lex Fridman (1:54:55.680)
destroying ourselves?
Max Tegmark (1:54:56.640)
I have to ask.
Lex Fridman (1:54:57.480)
I don't know.
Max Tegmark (1:54:58.320)
It might be the Russian origins.
Lex Fridman (1:55:01.080)
Somehow it's involved.
Max Tegmark (1:55:02.840)
At the micro level of detailed strategies,
Lex Fridman (1:55:04.760)
of course, these are unsolved problems.
Max Tegmark (1:55:08.640)
For AI alignment,
Lex Fridman (1:55:09.680)
we can break it into three sub problems
Max Tegmark (1:55:12.240)
that are all unsolved.
Lex Fridman (1:55:13.480)
I think you want first to make machines
Max Tegmark (1:55:16.720)
understand our goals,
Lex Fridman (1:55:18.400)
then adopt our goals and then retain our goals.
Lex Fridman (1:55:23.600)
So to hit on all three real quickly.
Lex Fridman (1:55:27.400)
The problem when Andreas Lubitz told his autopilot
Max Tegmark (1:55:31.080)
to fly into the Alps was that the computer
Lex Fridman (1:55:34.320)
didn't even understand anything about his goals.
Max Tegmark (1:55:39.040)
It was too dumb.
Lex Fridman (1:55:40.520)
It could have understood actually,
Lex Fridman (1:55:42.720)
but you would have had to put some effort in
Lex Fridman (1:55:45.280)
as a systems designer to don't fly into mountains.
Lex Fridman (1:55:48.880)
So that's the first challenge.
Lex Fridman (1:55:49.920)
How do you program into computers human values,
Lex Fridman (1:55:54.480)
human goals?
Lex Fridman (1:55:56.240)
We can start rather than saying,
Max Tegmark (1:55:58.280)
oh, it's so hard.
Lex Fridman (1:55:59.120)
We should start with the simple stuff, as I said,
Max Tegmark (1:56:02.400)
self driving cars, airplanes,
Lex Fridman (1:56:04.120)
just put in all the goals that we all agree on already,
Lex Fridman (1:56:07.240)
and then have a habit of whenever machines get smarter
Lex Fridman (1:56:10.560)
so they can understand one level higher goals,
Max Tegmark (1:56:15.480)
put them into.
Lex Fridman (1:56:16.960)
The second challenge is getting them to adopt the goals.
Max Tegmark (1:56:20.840)
It's easy for situations like that
Lex Fridman (1:56:22.320)
where you just program it in,
Lex Fridman (1:56:23.280)
but when you have self learning systems like children,
Lex Fridman (1:56:26.040)
you know, any parent knows
Max Tegmark (1:56:29.320)
that there was a difference between getting our kids
Lex Fridman (1:56:33.440)
to understand what we want them to do
Lex Fridman (1:56:34.840)
and to actually adopt our goals, right?
Lex Fridman (1:56:37.600)
With humans, with children, fortunately,
Max Tegmark (1:56:40.040)
they go through this phase.
Lex Fridman (1:56:44.000)
First, they're too dumb to understand
Lex Fridman (1:56:45.480)
what we want our goals are.
Lex Fridman (1:56:46.800)
And then they have this period of some years
Max Tegmark (1:56:50.360)
when they're both smart enough to understand them
Lex Fridman (1:56:52.080)
and malleable enough that we have a chance
Max Tegmark (1:56:53.520)
to raise them well.
Lex Fridman (1:56:55.400)
And then they become teenagers kind of too late.
Lex Fridman (1:56:59.160)
But we have this window with machines,
Lex Fridman (1:57:01.360)
the challenges, the intelligence might grow so fast
Max Tegmark (1:57:04.120)
that that window is pretty short.
Lex Fridman (1:57:06.800)
So that's a research problem.
Max Tegmark (1:57:08.480)
The third one is how do you make sure they keep the goals
Lex Fridman (1:57:11.320)
if they keep learning more and getting smarter?
Max Tegmark (1:57:14.520)
Many sci fi movies are about how you have something
Lex Fridman (1:57:17.360)
in which initially was aligned,
Lex Fridman (1:57:18.520)
but then things kind of go off keel.
Lex Fridman (1:57:20.320)
And, you know, my kids were very, very excited
Max Tegmark (1:57:24.680)
about their Legos when they were little.
Lex Fridman (1:57:27.360)
Now they're just gathering dust in the basement.
Max Tegmark (1:57:29.800)
If we create machines that are really on board
Lex Fridman (1:57:32.560)
with the goal of taking care of humanity,
Max Tegmark (1:57:34.320)
we don't want them to get as bored with us
Lex Fridman (1:57:36.080)
as my kids got with Legos.
Lex Fridman (1:57:39.480)
So this is another research challenge.
Lex Fridman (1:57:41.920)
How can you make some sort of recursively
Lex Fridman (1:57:43.400)
self improving system retain certain basic goals?
Lex Fridman (1:57:47.440)
That said, a lot of adult people still play with Legos.
Lex Fridman (1:57:50.880)
So maybe we succeeded with the Legos.
Lex Fridman (1:57:52.720)
Maybe, I like your optimism.
Lex Fridman (1:57:55.320)
But above all.
Lex Fridman (1:57:56.160)
So not all AI systems have to maintain the goals, right?
Max Tegmark (1:57:59.120)
Just some fraction.
Lex Fridman (1:58:00.200)
Yeah, so there's a lot of talented AI researchers now
Max Tegmark (1:58:04.920)
who have heard of this and want to work on it.
Lex Fridman (1:58:07.280)
Not so much funding for it yet.
Max Tegmark (1:58:10.960)
Of the billions that go into building AI more powerful,
Lex Fridman (1:58:14.800)
it's only a minuscule fraction
Lex Fridman (1:58:16.240)
so far going into this safety research.
Lex Fridman (1:58:18.280)
My attitude is generally we should not try to slow down
Max Tegmark (1:58:20.880)
the technology, but we should greatly accelerate
Lex Fridman (1:58:22.840)
the investment in this sort of safety research.
Lex Fridman (1:58:25.880)
And also, this was very embarrassing last year,
Lex Fridman (1:58:29.320)
but the NSF decided to give out
Max Tegmark (1:58:31.840)
six of these big institutes.
Lex Fridman (1:58:33.680)
We got one of them for AI and science, you asked me about.
Max Tegmark (1:58:37.040)
Another one was supposed to be for AI safety research.
Lex Fridman (1:58:40.720)
And they gave it to people studying oceans
Lex Fridman (1:58:43.520)
and climate and stuff.
Lex Fridman (1:58:46.920)
So I'm all for studying oceans and climates,
Lex Fridman (1:58:49.320)
but we need to actually have some money
Lex Fridman (1:58:51.120)
that actually goes into AI safety research also
Lex Fridman (1:58:53.360)
and doesn't just get grabbed by whatever.
Lex Fridman (1:58:56.400)
That's a fantastic investment.
Lex Fridman (1:58:57.960)
And then at the higher level, you asked this question,
Lex Fridman (1:59:00.480)
okay, what can we do?
Lex Fridman (1:59:02.680)
What are the biggest risks?
Lex Fridman (1:59:05.240)
I think we cannot just consider this
Max Tegmark (1:59:08.760)
to be only a technical problem.
Lex Fridman (1:59:11.000)
Again, because if you solve only the technical problem,
Lex Fridman (1:59:13.640)
can I play with your robot?
Lex Fridman (1:59:14.680)
Yes, please.
Max Tegmark (1:59:15.520)
If we can get our machines to just blindly obey
Lex Fridman (1:59:20.560)
the orders we give them,
Lex Fridman (1:59:22.760)
so we can always trust that it will do what we want.
Lex Fridman (1:59:26.160)
That might be great for the owner of the robot.
Max Tegmark (1:59:28.440)
That might not be so great for the rest of humanity
Lex Fridman (1:59:31.400)
if that person is that least favorite world leader
Lex Fridman (1:59:34.080)
or whatever you imagine, right?
Lex Fridman (1:59:36.600)
So we have to also take a look at the,
Max Tegmark (1:59:39.200)
apply alignment, not just to machines,
Lex Fridman (1:59:41.960)
but to all the other powerful structures.
Max Tegmark (1:59:44.560)
That's why it's so important
Lex Fridman (1:59:45.720)
to strengthen our democracy again,
Max Tegmark (1:59:47.040)
as I said, to have institutions,
Lex Fridman (1:59:48.520)
make sure that the playing field is not rigged
Lex Fridman (1:59:51.440)
so that corporations are given the right incentives
Lex Fridman (1:59:54.800)
to do the things that both make profit
Lex Fridman (1:59:57.240)
and are good for people,
Lex Fridman (1:59:58.880)
to make sure that countries have incentives
Lex Fridman (20:01.080)
Why?
Lex Fridman (20:01.560)
Because he had a neural network in his brain
Max Tegmark (20:04.480)
that he had trained to predict the parabolic orbit of apples
Lex Fridman (20:07.960)
that are thrown under gravity.
Max Tegmark (20:09.960)
If you throw a tennis ball to a dog,
Lex Fridman (20:12.000)
it also has this same ability of deep learning
Max Tegmark (20:15.360)
to figure out how the ball is going to move and catch it.
Lex Fridman (20:18.160)
But Galileo went one step further when he got older.
Max Tegmark (20:21.960)
He went back and was like, wait a minute.
Lex Fridman (20:26.040)
I can write down a formula for this.
Max Tegmark (20:27.880)
Y equals x squared, a parabola.
Lex Fridman (20:31.560)
And he helped revolutionize physics as we know it, right?
Lex Fridman (20:36.520)
So there was a basic neural network
Lex Fridman (20:38.200)
in there from childhood that captured the experiences
Max Tegmark (20:43.360)
of observing different kinds of trajectories.
Lex Fridman (20:46.440)
And then he was able to go back in
Max Tegmark (20:48.240)
with another extra little neural network
Lex Fridman (20:51.000)
and analyze all those experiences and be like,
Max Tegmark (20:53.480)
wait a minute.
Lex Fridman (20:54.600)
There's a deeper rule here.
Max Tegmark (20:56.240)
Exactly.
Lex Fridman (20:56.960)
He was able to distill out in symbolic form
Lex Fridman (21:00.720)
what that complicated black box neural network was doing.
Lex Fridman (21:03.960)
Not only did the formula he got ultimately
Max Tegmark (21:07.320)
become more accurate, and similarly, this
Lex Fridman (21:09.840)
is how Newton got Newton's laws, which
Lex Fridman (21:12.000)
is why Elon can send rockets to the space station now, right?
Lex Fridman (21:15.600)
So it's not only more accurate, but it's also simpler,
Max Tegmark (21:19.480)
much simpler.
Lex Fridman (21:20.120)
And it's so simple that we can actually describe it
Lex Fridman (21:22.320)
to our friends and each other, right?
Lex Fridman (21:26.080)
We've talked about it just in the context of physics now.
Lex Fridman (21:28.800)
But hey, isn't this what we're doing when we're
Lex Fridman (21:31.560)
talking to each other also?
Max Tegmark (21:33.360)
We go around with our neural networks,
Lex Fridman (21:35.440)
just like dogs and cats and chipmunks and Blue Jays.
Lex Fridman (21:38.760)
And we experience things in the world.
Lex Fridman (21:41.920)
But then we humans do this additional step
Max Tegmark (21:43.840)
on top of that, where we then distill out
Lex Fridman (21:46.720)
certain high level knowledge that we've extracted from this
Max Tegmark (21:50.280)
in a way that we can communicate it
Lex Fridman (21:52.240)
to each other in a symbolic form in English in this case, right?
Lex Fridman (21:56.600)
So if we can do it and we believe
Lex Fridman (21:59.960)
that we are information processing entities,
Max Tegmark (22:02.880)
then we should be able to make machine learning that
Lex Fridman (22:04.960)
does it also.
Lex Fridman (22:07.160)
Well, do you think the entire thing could be learning?
Lex Fridman (22:10.200)
Because this dissection process, like for AI Feynman,
Max Tegmark (22:14.160)
the secondary stage feels like something like reasoning.
Lex Fridman (22:19.240)
And the initial step feels more like the more basic kind
Max Tegmark (22:23.400)
of differentiable learning.
Lex Fridman (22:25.280)
Do you think the whole thing could be differentiable
Lex Fridman (22:27.680)
learning?
Lex Fridman (22:28.720)
Do you think the whole thing could be basically neural
Lex Fridman (22:31.120)
networks on top of each other?
Lex Fridman (22:32.320)
It's like turtles all the way down.
Lex Fridman (22:33.800)
Could it be neural networks all the way down?
Lex Fridman (22:35.920)
I mean, that's a really interesting question.
Max Tegmark (22:37.920)
We know that in your case, it is neural networks all the way
Lex Fridman (22:41.040)
down because that's all you have in your skull
Lex Fridman (22:42.960)
is a bunch of neurons doing their thing, right?
Lex Fridman (22:45.880)
But if you ask the question more generally,
Lex Fridman (22:50.320)
what algorithms are being used in your brain,
Lex Fridman (22:54.120)
I think it's super interesting to compare.
Max Tegmark (22:56.160)
I think we've gone a little bit backwards historically
Lex Fridman (22:58.760)
because we humans first discovered good old fashioned
Max Tegmark (23:02.800)
AI, the logic based AI that we often call GoFi
Lex Fridman (23:06.880)
for good old fashioned AI.
Lex Fridman (23:09.080)
And then more recently, we did machine learning
Lex Fridman (23:12.600)
because it required bigger computers.
Lex Fridman (23:14.160)
So we had to discover it later.
Lex Fridman (23:15.960)
So we think of machine learning with neural networks
Max Tegmark (23:19.160)
as the modern thing and the logic based AI
Lex Fridman (23:21.840)
as the old fashioned thing.
Lex Fridman (23:24.280)
But if you look at evolution on Earth,
Lex Fridman (23:27.800)
it's actually been the other way around.
Max Tegmark (23:29.800)
I would say that, for example, an eagle
Lex Fridman (23:34.120)
has a better vision system than I have using.
Lex Fridman (23:38.680)
And dogs are just as good at casting tennis balls as I am.
Lex Fridman (23:42.360)
All this stuff which is done by training in neural network
Lex Fridman (23:45.920)
and not interpreting it in words is
Lex Fridman (23:49.920)
something so many of our animal friends can do,
Lex Fridman (23:51.880)
at least as well as us, right?
Lex Fridman (23:53.680)
What is it that we humans can do that the chipmunks
Lex Fridman (23:56.560)
and the eagles cannot?
Lex Fridman (23:58.880)
It's more to do with this logic based stuff, right,
Max Tegmark (24:01.600)
where we can extract out information
Lex Fridman (24:04.840)
in symbols, in language, and now even with equations
Lex Fridman (24:10.240)
if you're a scientist, right?
Lex Fridman (24:12.160)
So basically what happened was first we
Max Tegmark (24:13.920)
built these computers that could multiply numbers real fast
Lex Fridman (24:16.880)
and manipulate symbols.
Lex Fridman (24:18.080)
And we felt they were pretty dumb.
Lex Fridman (24:20.520)
And then we made neural networks that
Max Tegmark (24:22.800)
can see as well as a cat can and do
Lex Fridman (24:25.280)
a lot of this inscrutable black box neural networks.
Lex Fridman (24:30.040)
What we humans can do also is put the two together
Lex Fridman (24:33.000)
in a useful way.
Max Tegmark (24:34.040)
Yes, in our own brain.
Lex Fridman (24:36.120)
Yes, in our own brain.
Lex Fridman (24:37.360)
So if we ever want to get artificial general intelligence
Lex Fridman (24:40.920)
that can do all jobs as well as humans can, right,
Max Tegmark (24:45.160)
then that's what's going to be required
Lex Fridman (24:47.040)
to be able to combine the neural networks with symbolic,
Max Tegmark (24:53.120)
combine the old AI with the new AI in a good way.
Lex Fridman (24:55.200)
We do it in our brains.
Lex Fridman (24:57.200)
And there seems to be basically two strategies
Lex Fridman (24:59.760)
I see in industry now.
Max Tegmark (25:01.000)
One scares the heebie jeebies out of me,
Lex Fridman (25:03.600)
and the other one I find much more encouraging.
Lex Fridman (25:05.840)
OK, which one?
Lex Fridman (25:07.080)
Can we break them apart?
Lex Fridman (25:08.320)
Which of the two?
Lex Fridman (25:09.600)
The one that scares the heebie jeebies out of me
Max Tegmark (25:11.640)
is this attitude that we're just going
Lex Fridman (25:12.880)
to make ever bigger systems that we still
Max Tegmark (25:14.720)
don't understand until they can be as smart as humans.
Lex Fridman (25:19.280)
What could possibly go wrong?
Max Tegmark (25:22.200)
I think it's just such a reckless thing to do.
Lex Fridman (25:24.200)
And unfortunately, if we actually
Max Tegmark (25:27.000)
succeed as a species to build artificial general intelligence,
Lex Fridman (25:30.120)
then we still have no clue how it works.
Max Tegmark (25:31.840)
I think at least 50% chance we're
Lex Fridman (25:35.440)
going to be extinct before too long.
Max Tegmark (25:37.040)
It's just going to be an utter epic own goal.
Lex Fridman (25:40.480)
So it's that 44 minute losing money problem or the paper clip
Max Tegmark (25:46.600)
problem where we don't understand how it works,
Lex Fridman (25:49.480)
and it just in a matter of seconds
Max Tegmark (25:51.280)
runs away in some kind of direction
Lex Fridman (25:52.760)
that's going to be very problematic.
Max Tegmark (25:54.440)
Even long before, you have to worry about the machines
Lex Fridman (25:57.640)
themselves somehow deciding to do things.
Lex Fridman (26:01.400)
And to us, we have to worry about people using machines
Lex Fridman (26:06.840)
that are short of AGI and power to do bad things.
Max Tegmark (26:09.840)
I mean, just take a moment.
Lex Fridman (26:13.080)
And if anyone is not worried particularly about advanced AI,
Max Tegmark (26:18.040)
just take 10 seconds and just think
Lex Fridman (26:20.800)
about your least favorite leader on the planet right now.
Max Tegmark (26:23.720)
Don't tell me who it is.
Lex Fridman (26:25.120)
I want to keep this apolitical.
Lex Fridman (26:26.760)
But just see the face in front of you,
Lex Fridman (26:28.840)
that person, for 10 seconds.
Max Tegmark (26:30.480)
Now imagine that that person has this incredibly powerful AI
Lex Fridman (26:35.280)
under their control and can use it
Max Tegmark (26:37.120)
to impose their will on the whole planet.
Lex Fridman (26:38.760)
How does that make you feel?
Max Tegmark (26:42.840)
Yeah.
Lex Fridman (26:44.280)
So can we break that apart just briefly?
Max Tegmark (26:49.480)
For the 50% chance that we'll run
Lex Fridman (26:51.720)
to trouble with this approach, do you
Max Tegmark (26:53.880)
see the bigger worry in that leader or humans
Lex Fridman (26:58.040)
using the system to do damage?
Max Tegmark (27:00.600)
Or are you more worried, and I think I'm in this camp,
Lex Fridman (27:05.360)
more worried about accidental, unintentional destruction
Lex Fridman (27:09.800)
of everything?
Lex Fridman (27:10.840)
So humans trying to do good, and in a way
Max Tegmark (27:14.960)
where everyone agrees it's kind of good,
Lex Fridman (27:17.400)
it's just they're trying to do good without understanding.
Max Tegmark (27:20.040)
Because I think every evil leader in history
Lex Fridman (27:22.480)
thought they're, to some degree, thought
Max Tegmark (27:24.560)
they're trying to do good.
Lex Fridman (27:25.600)
Oh, yeah.
Max Tegmark (27:25.880)
I'm sure Hitler thought he was doing good.
Lex Fridman (27:28.080)
Yeah.
Max Tegmark (27:29.480)
I've been reading a lot about Stalin.
Lex Fridman (27:31.120)
I'm sure Stalin is from, he legitimately
Max Tegmark (27:34.240)
thought that communism was good for the world,
Lex Fridman (27:36.560)
and that he was doing good.
Max Tegmark (27:37.760)
I think Mao Zedong thought what he was doing with the Great
Lex Fridman (27:39.960)
Leap Forward was good too.
Max Tegmark (27:41.200)
Yeah.
Lex Fridman (27:42.880)
I'm actually concerned about both of those.
Max Tegmark (27:45.560)
Before, I promised to answer this in detail,
Lex Fridman (27:48.440)
but before we do that, let me finish
Max Tegmark (27:50.320)
answering the first question.
Lex Fridman (27:51.240)
Because I told you that there were two different routes we
Max Tegmark (27:53.520)
could get to artificial general intelligence,
Lex Fridman (27:55.400)
and one scares the hell out of me,
Max Tegmark (27:57.240)
which is this one where we build something,
Lex Fridman (27:59.320)
we just say bigger neural networks, ever more hardware,
Lex Fridman (28:02.040)
and just train the heck out of more data,
Lex Fridman (28:03.760)
and poof, now it's very powerful.
Max Tegmark (28:07.240)
That, I think, is the most unsafe and reckless approach.
Lex Fridman (28:11.800)
The alternative to that is the intelligible intelligence
Max Tegmark (28:16.480)
approach instead, where we say neural networks is just
Lex Fridman (28:22.840)
a tool for the first step to get the intuition,
Lex Fridman (28:27.000)
but then we're going to spend also
Lex Fridman (28:29.120)
serious resources on other AI techniques
Max Tegmark (28:33.280)
for demystifying this black box and figuring out
Lex Fridman (28:35.960)
what it's actually doing so we can convert it
Max Tegmark (28:38.680)
into something that's equally intelligent,
Lex Fridman (28:41.040)
but that we actually understand what it's doing.
Max Tegmark (28:44.040)
Maybe we can even prove theorems about it,
Lex Fridman (28:45.960)
that this car here will never be hacked when it's driving,
Max Tegmark (28:50.120)
because here is the proof.
Lex Fridman (28:53.800)
There is a whole science of this.
Max Tegmark (28:55.160)
It doesn't work for neural networks
Lex Fridman (28:57.040)
that are big black boxes, but it works well
Lex Fridman (28:58.800)
and works with certain other kinds of codes, right?
Lex Fridman (29:02.880)
That approach, I think, is much more promising.
Max Tegmark (29:05.160)
That's exactly why I'm working on it, frankly,
Lex Fridman (29:07.160)
not just because I think it's cool for science,
Lex Fridman (29:09.400)
but because I think the more we understand these systems,
Lex Fridman (29:14.080)
the better the chances that we can
Max Tegmark (29:16.160)
make them do the things that are good for us
Lex Fridman (29:18.400)
that are actually intended, not unintended.
Lex Fridman (29:21.520)
So you think it's possible to prove things
Lex Fridman (29:24.280)
about something as complicated as a neural network?
Lex Fridman (29:27.360)
That's the hope?
Lex Fridman (29:28.440)
Well, ideally, there's no reason it
Lex Fridman (29:30.840)
has to be a neural network in the end either, right?
Lex Fridman (29:34.320)
We discovered Newton's laws of gravity
Max Tegmark (29:36.480)
with neural network in Newton's head.
Lex Fridman (29:40.040)
But that's not the way it's programmed into the navigation
Max Tegmark (29:44.080)
system of Elon Musk's rocket anymore.
Lex Fridman (29:46.600)
It's written in C++, or I don't know
Lex Fridman (29:49.200)
what language he uses exactly.
Lex Fridman (29:51.360)
And then there are software tools called symbolic
Max Tegmark (29:53.400)
verification.
Lex Fridman (29:54.640)
DARPA and the US military has done a lot of really great
Max Tegmark (29:59.080)
research on this, because they really
Lex Fridman (2:00:00.920)
to do things that are both good for their people
Lex Fridman (2:00:03.320)
and don't screw up the rest of the world.
Lex Fridman (2:00:06.840)
And this is not just something for AI nerds to geek out on.
Max Tegmark (2:00:10.280)
This is an interesting challenge for political scientists,
Lex Fridman (2:00:13.080)
economists, and so many other thinkers.
Lex Fridman (2:00:16.800)
So one of the magical things
Lex Fridman (2:00:18.680)
that perhaps makes this earth quite unique
Max Tegmark (2:00:25.240)
is that it's home to conscious beings.
Lex Fridman (2:00:28.840)
So you mentioned consciousness.
Max Tegmark (2:00:31.640)
Perhaps as a small aside,
Lex Fridman (2:00:35.000)
because we didn't really get specific
Max Tegmark (2:00:36.720)
to how we might do the alignment.
Lex Fridman (2:00:39.440)
Like you said,
Max Tegmark (2:00:40.280)
is there just a really important research problem,
Lex Fridman (2:00:41.840)
but do you think engineering consciousness
Max Tegmark (2:00:44.720)
into AI systems is a possibility,
Lex Fridman (2:00:49.880)
is something that we might one day do,
Max Tegmark (2:00:53.040)
or is there something fundamental to consciousness
Lex Fridman (2:00:56.800)
that is, is there something about consciousness
Lex Fridman (2:00:59.880)
that is fundamental to humans and humans only?
Lex Fridman (2:01:03.400)
I think it's possible.
Max Tegmark (2:01:04.640)
I think both consciousness and intelligence
Lex Fridman (2:01:08.320)
are information processing.
Max Tegmark (2:01:10.760)
Certain types of information processing.
Lex Fridman (2:01:13.480)
And that fundamentally,
Max Tegmark (2:01:15.160)
it doesn't matter whether the information is processed
Lex Fridman (2:01:17.320)
by carbon atoms in the neurons and brains
Max Tegmark (2:01:21.280)
or by silicon atoms and so on in our technology.
Lex Fridman (2:01:27.280)
Some people disagree.
Max Tegmark (2:01:28.280)
This is what I think as a physicist.
Lex Fridman (2:01:32.960)
That consciousness is the same kind of,
Max Tegmark (2:01:34.960)
you said consciousness is information processing.
Lex Fridman (2:01:37.720)
So meaning, I think you had a quote of something like
Max Tegmark (2:01:43.000)
it's information knowing itself, that kind of thing.
Lex Fridman (2:01:47.760)
I think consciousness is, yeah,
Max Tegmark (2:01:49.280)
is the way information feels when it's being processed.
Lex Fridman (2:01:51.960)
One's being put in complex ways.
Max Tegmark (2:01:53.520)
We don't know exactly what those complex ways are.
Lex Fridman (2:01:56.120)
It's clear that most of the information processing
Max Tegmark (2:01:59.240)
in our brains does not create an experience.
Lex Fridman (2:02:01.720)
We're not even aware of it, right?
Max Tegmark (2:02:03.600)
Like for example,
Lex Fridman (2:02:05.520)
you're not aware of your heartbeat regulation right now,
Lex Fridman (2:02:07.880)
even though it's clearly being done by your body, right?
Lex Fridman (2:02:10.600)
It's just kind of doing its own thing.
Max Tegmark (2:02:12.120)
When you go jogging,
Lex Fridman (2:02:13.680)
there's a lot of complicated stuff
Max Tegmark (2:02:15.280)
about how you put your foot down and we know it's hard.
Lex Fridman (2:02:18.720)
That's why robots used to fall over so much,
Lex Fridman (2:02:20.560)
but you're mostly unaware about it.
Lex Fridman (2:02:22.720)
Your brain, your CEO consciousness module
Max Tegmark (2:02:25.760)
just sends an email,
Lex Fridman (2:02:26.600)
hey, I'm gonna keep jogging along this path.
Lex Fridman (2:02:29.160)
The rest is on autopilot, right?
Lex Fridman (2:02:31.560)
So most of it is not conscious,
Lex Fridman (2:02:33.200)
but somehow there is some of the information processing,
Lex Fridman (2:02:36.640)
which is we don't know what exactly.
Max Tegmark (2:02:41.680)
I think this is a science problem
Lex Fridman (2:02:44.120)
that I hope one day we'll have some equation for
Max Tegmark (2:02:47.680)
or something so we can be able to build
Lex Fridman (2:02:49.080)
a consciousness detector and say, yeah,
Max Tegmark (2:02:51.040)
here there is some consciousness, here there's not.
Lex Fridman (2:02:53.920)
Oh, don't boil that lobster because it's feeling pain
Max Tegmark (2:02:56.640)
or it's okay because it's not feeling pain.
Lex Fridman (2:02:59.880)
Right now we treat this as sort of just metaphysics,
Lex Fridman (2:03:03.440)
but it would be very useful in emergency rooms
Lex Fridman (2:03:06.920)
to know if a patient has locked in syndrome
Lex Fridman (2:03:09.760)
and is conscious or if they are actually just out.
Lex Fridman (2:03:14.560)
And in the future, if you build a very, very intelligent
Max Tegmark (2:03:17.720)
helper robot to take care of you,
Lex Fridman (2:03:20.120)
I think you'd like to know
Max Tegmark (2:03:21.480)
if you should feel guilty about shutting it down
Lex Fridman (2:03:24.120)
or if it's just like a zombie going through the motions
Lex Fridman (2:03:27.080)
like a fancy tape recorder, right?
Lex Fridman (2:03:29.720)
And once we can make progress
Max Tegmark (2:03:32.800)
on the science of consciousness
Lex Fridman (2:03:34.040)
and figure out what is conscious and what isn't,
Max Tegmark (2:03:38.320)
then assuming we want to create positive experiences
Lex Fridman (2:03:45.960)
and not suffering, we'll probably choose to build
Max Tegmark (2:03:48.880)
some machines that are deliberately unconscious
Lex Fridman (2:03:51.760)
that do incredibly boring, repetitive jobs
Max Tegmark (2:03:56.760)
in an iron mine somewhere or whatever.
Lex Fridman (2:03:59.680)
And maybe we'll choose to create helper robots
Max Tegmark (2:04:03.120)
for the elderly that are conscious
Lex Fridman (2:04:05.360)
so that people don't just feel creeped out
Max Tegmark (2:04:07.080)
that the robot is just faking it
Lex Fridman (2:04:10.160)
when it acts like it's sad or happy.
Max Tegmark (2:04:12.160)
Like you said, elderly,
Lex Fridman (2:04:13.440)
I think everybody gets pretty deeply lonely in this world.
Lex Fridman (2:04:16.920)
And so there's a place I think for everybody
Lex Fridman (2:04:19.640)
to have a connection with conscious beings,
Max Tegmark (2:04:21.640)
whether they're human or otherwise.
Lex Fridman (2:04:24.400)
But I know for sure that I would,
Max Tegmark (2:04:26.920)
if I had a robot, if I was gonna develop any kind
Lex Fridman (2:04:29.960)
of personal emotional connection with it,
Max Tegmark (2:04:32.760)
I would be very creeped out
Lex Fridman (2:04:33.840)
if I knew it in an intellectual level
Max Tegmark (2:04:35.280)
that the whole thing was just a fraud.
Lex Fridman (2:04:36.840)
Now today you can buy a little talking doll for a kid
Max Tegmark (2:04:43.000)
which will say things and the little child will often think
Lex Fridman (2:04:46.120)
that this is actually conscious
Lex Fridman (2:04:47.840)
and even real secrets to it that then go on the internet
Lex Fridman (2:04:50.440)
and with lots of the creepy repercussions.
Max Tegmark (2:04:52.520)
I would not wanna be just hacked and tricked like this.
Lex Fridman (2:04:58.040)
If I was gonna be developing real emotional connections
Max Tegmark (2:05:01.560)
with the robot, I would wanna know
Lex Fridman (2:05:04.200)
that this is actually real.
Max Tegmark (2:05:05.440)
It's acting conscious, acting happy
Lex Fridman (2:05:08.080)
because it actually feels it.
Lex Fridman (2:05:09.880)
And I think this is not sci fi.
Lex Fridman (2:05:11.400)
I think it's possible to measure, to come up with tools.
Max Tegmark (2:05:15.560)
After we understand the science of consciousness,
Lex Fridman (2:05:17.560)
you're saying we'll be able to come up with tools
Max Tegmark (2:05:19.760)
that can measure consciousness
Lex Fridman (2:05:21.400)
and definitively say like this thing is experiencing
Max Tegmark (2:05:25.120)
the things it says it's experiencing.
Lex Fridman (2:05:27.360)
Kind of by definition.
Max Tegmark (2:05:28.320)
If it is a physical phenomenon, information processing
Lex Fridman (2:05:31.560)
and we know that some information processing is conscious
Lex Fridman (2:05:34.040)
and some isn't, well, then there is something there
Lex Fridman (2:05:36.000)
to be discovered with the methods of science.
Max Tegmark (2:05:38.040)
Giulio Tononi has stuck his neck out the farthest
Lex Fridman (2:05:41.120)
and written down some equations for a theory.
Max Tegmark (2:05:43.640)
Maybe that's right, maybe it's wrong.
Lex Fridman (2:05:45.680)
We certainly don't know.
Lex Fridman (2:05:46.920)
But I applaud that kind of efforts to sort of take this,
Lex Fridman (2:05:50.760)
say this is not just something that philosophers
Max Tegmark (2:05:53.960)
can have beer and muse about,
Lex Fridman (2:05:56.320)
but something we can measure and study.
Lex Fridman (2:05:58.720)
And coming, bringing that back to us,
Lex Fridman (2:06:00.560)
I think what we would probably choose to do, as I said,
Max Tegmark (2:06:03.000)
is if we cannot figure this out,
Lex Fridman (2:06:05.680)
choose to make, to be quite mindful
Max Tegmark (2:06:09.000)
about what sort of consciousness, if any,
Lex Fridman (2:06:11.280)
we put in different machines that we have.
Lex Fridman (2:06:16.080)
And certainly, we wouldn't wanna make,
Lex Fridman (2:06:19.000)
we should not be making much machines that suffer
Lex Fridman (2:06:21.760)
without us even knowing it, right?
Lex Fridman (2:06:23.640)
And if at any point someone decides to upload themselves
Max Tegmark (2:06:28.320)
like Ray Kurzweil wants to do,
Lex Fridman (2:06:30.120)
I don't know if you've had him on your show.
Max Tegmark (2:06:31.440)
We agree, but then COVID happens,
Lex Fridman (2:06:33.040)
so we're waiting it out a little bit.
Max Tegmark (2:06:34.680)
Suppose he uploads himself into this robo Ray
Lex Fridman (2:06:38.520)
and it talks like him and acts like him and laughs like him.
Lex Fridman (2:06:42.200)
And before he powers off his biological body,
Lex Fridman (2:06:46.480)
he would probably be pretty disturbed
Max Tegmark (2:06:47.760)
if he realized that there's no one home.
Lex Fridman (2:06:49.600)
This robot is not having any subjective experience, right?
Max Tegmark (2:06:53.760)
If humanity gets replaced by machine descendants,
Lex Fridman (2:06:59.840)
which do all these cool things and build spaceships
Lex Fridman (2:07:02.320)
and go to intergalactic rock concerts,
Lex Fridman (2:07:05.640)
and it turns out that they are all unconscious,
Max Tegmark (2:07:10.000)
just going through the motions,
Lex Fridman (2:07:11.440)
wouldn't that be like the ultimate zombie apocalypse, right?
Lex Fridman (2:07:16.160)
Just a play for empty benches?
Lex Fridman (2:07:18.040)
Yeah, I have a sense that there's some kind of,
Max Tegmark (2:07:21.200)
once we understand consciousness better,
Lex Fridman (2:07:22.800)
we'll understand that there's some kind of continuum
Lex Fridman (2:07:25.640)
and it would be a greater appreciation.
Lex Fridman (2:07:28.000)
And we'll probably understand, just like you said,
Max Tegmark (2:07:30.440)
it'd be unfortunate if it's a trick.
Lex Fridman (2:07:32.400)
We'll probably definitely understand
Max Tegmark (2:07:33.920)
that love is indeed a trick that we'll play on each other,
Lex Fridman (2:07:37.760)
that we humans are, we convince ourselves we're conscious,
Lex Fridman (2:07:40.960)
but we're really, us and trees and dolphins
Lex Fridman (2:07:45.240)
are all the same kind of consciousness.
Max Tegmark (2:07:46.600)
Can I try to cheer you up a little bit
Lex Fridman (2:07:48.160)
with a philosophical thought here about the love part?
Max Tegmark (2:07:50.280)
Yes, let's do it.
Lex Fridman (2:07:51.360)
You know, you might say,
Max Tegmark (2:07:53.920)
okay, yeah, love is just a collaboration enabler.
Lex Fridman (2:07:58.120)
And then maybe you can go and get depressed about that.
Lex Fridman (2:08:01.800)
But I think that would be the wrong conclusion, actually.
Lex Fridman (2:08:04.640)
You know, I know that the only reason I enjoy food
Max Tegmark (2:08:08.640)
is because my genes hacked me
Lex Fridman (2:08:11.000)
and they don't want me to starve to death.
Max Tegmark (2:08:13.720)
Not because they care about me consciously
Lex Fridman (2:08:17.280)
enjoying succulent delights of pistachio ice cream,
Lex Fridman (2:08:21.080)
but they just want me to make copies of them.
Lex Fridman (2:08:23.360)
The whole thing, so in a sense,
Max Tegmark (2:08:24.520)
the whole enjoyment of food is also a scam like this.
Lex Fridman (2:08:28.960)
But does that mean I shouldn't take pleasure
Lex Fridman (2:08:31.280)
in this pistachio ice cream?
Lex Fridman (2:08:32.560)
I love pistachio ice cream.
Lex Fridman (2:08:34.040)
And I can tell you, I know this is an experimental fact.
Lex Fridman (2:08:38.200)
I enjoy pistachio ice cream every bit as much,
Max Tegmark (2:08:41.600)
even though I scientifically know exactly why,
Lex Fridman (2:08:45.560)
what kind of scam this was.
Max Tegmark (2:08:46.880)
Your genes really appreciate
Lex Fridman (2:08:48.640)
that you like the pistachio ice cream.
Lex Fridman (2:08:50.440)
Well, but I, my mind appreciates it too, you know?
Lex Fridman (2:08:53.080)
And I have a conscious experience right now.
Max Tegmark (2:08:55.800)
Ultimately, all of my brain is also just something
Lex Fridman (2:08:58.640)
the genes built to copy themselves.
Lex Fridman (2:09:00.440)
But so what?
Lex Fridman (2:09:01.600)
You know, I'm grateful that,
Max Tegmark (2:09:03.200)
yeah, thanks genes for doing this,
Lex Fridman (2:09:04.960)
but you know, now it's my brain that's in charge here
Lex Fridman (2:09:07.600)
and I'm gonna enjoy my conscious experience,
Lex Fridman (2:09:09.520)
thank you very much.
Lex Fridman (2:09:10.360)
And not just the pistachio ice cream,
Lex Fridman (2:09:12.480)
but also the love I feel for my amazing wife
Lex Fridman (2:09:15.440)
and all the other delights of being conscious.
Lex Fridman (2:09:19.280)
I don't, actually Richard Feynman,
Max Tegmark (2:09:22.240)
I think said this so well.
Lex Fridman (2:09:25.080)
He is also the guy, you know, really got me into physics.
Max Tegmark (2:09:29.680)
Some art friend said that,
Lex Fridman (2:09:31.240)
oh, science kind of just is the party pooper.
Lex Fridman (2:09:34.520)
It's kind of ruins the fun, right?
Lex Fridman (2:09:36.240)
When like you have a beautiful flowers as the artist
Lex Fridman (2:09:39.680)
and then the scientist is gonna deconstruct that
Lex Fridman (2:09:41.600)
into just a blob of quarks and electrons.
Lex Fridman (2:09:44.160)
And Feynman pushed back on that in such a beautiful way,
Lex Fridman (2:09:47.480)
which I think also can be used to push back
Lex Fridman (2:09:49.920)
and make you not feel guilty about falling in love.
Lex Fridman (2:09:53.440)
So here's what Feynman basically said.
Max Tegmark (2:09:55.000)
He said to his friend, you know,
Lex Fridman (2:09:56.920)
yeah, I can also as a scientist see
Max Tegmark (2:09:59.080)
that this is a beautiful flower, thank you very much.
Lex Fridman (2:10:00.960)
Maybe I can't draw as good a painting as you
Max Tegmark (2:10:03.280)
because I'm not as talented an artist,
Lex Fridman (2:10:04.560)
but yeah, I can really see the beauty in it.
Lex Fridman (2:10:06.800)
And it just, it also looks beautiful to me.
Lex Fridman (2:10:09.360)
But in addition to that, Feynman said, as a scientist,
Lex Fridman (2:10:12.200)
I see even more beauty that the artist did not see, right?
Lex Fridman (2:10:16.960)
Suppose this is a flower on a blossoming apple tree.
Max Tegmark (2:10:21.120)
You could say this tree has more beauty in it
Lex Fridman (2:10:23.840)
than just the colors and the fragrance.
Max Tegmark (2:10:26.400)
This tree is made of air, Feynman wrote.
Lex Fridman (2:10:29.040)
This is one of my favorite Feynman quotes ever.
Lex Fridman (2:10:31.240)
And it took the carbon out of the air
Lex Fridman (2:10:33.760)
and bound it in using the flaming heat of the sun,
Max Tegmark (2:10:36.160)
you know, to turn the air into a tree.
Lex Fridman (2:10:38.600)
And when you burn logs in your fireplace,
Max Tegmark (2:10:42.760)
it's really beautiful to think that this is being reversed.
Lex Fridman (2:10:45.120)
Now the tree is going, the wood is going back into air.
Lex Fridman (2:10:48.600)
And in this flaming, beautiful dance of the fire
Lex Fridman (2:10:52.520)
that the artist can see is the flaming light of the sun
Max Tegmark (2:10:56.000)
that was bound in to turn the air into tree.
Lex Fridman (2:10:59.120)
And then the ashes is the little residue
Max Tegmark (2:11:01.480)
that didn't come from the air
Lex Fridman (2:11:02.560)
that the tree sucked out of the ground, you know.
Max Tegmark (2:11:04.280)
Feynman said, these are beautiful things.
Lex Fridman (2:11:06.160)
And science just adds, it doesn't subtract.
Lex Fridman (2:11:10.040)
And I feel exactly that way about love
Lex Fridman (2:11:12.760)
and about pistachio ice cream also.
Max Tegmark (2:11:16.000)
I can understand that there is even more nuance
Lex Fridman (2:11:18.680)
to the whole thing, right?
Max Tegmark (2:11:20.480)
At this very visceral level,
Lex Fridman (2:11:22.480)
you can fall in love just as much as someone
Max Tegmark (2:11:24.560)
who knows nothing about neuroscience.
Lex Fridman (2:11:27.680)
But you can also appreciate this even greater beauty in it.
Max Tegmark (2:11:31.840)
Just like, isn't it remarkable that it came about
Lex Fridman (2:11:35.600)
from this completely lifeless universe,
Max Tegmark (2:11:38.560)
just a bunch of hot blob of plasma expanding.
Lex Fridman (2:11:43.080)
And then over the eons, you know, gradually,
Max Tegmark (2:11:46.160)
first the strong nuclear force decided
Lex Fridman (2:11:48.440)
to combine quarks together into nuclei.
Lex Fridman (2:11:50.920)
And then the electric force bound in electrons
Lex Fridman (2:11:53.040)
and made atoms.
Lex Fridman (2:11:53.880)
And then they clustered from gravity
Lex Fridman (2:11:55.240)
and you got planets and stars and this and that.
Lex Fridman (2:11:57.720)
And then natural selection came along
Lex Fridman (2:12:00.040)
and the genes had their little thing.
Lex Fridman (2:12:01.800)
And you started getting what went from seeming
Lex Fridman (2:12:04.640)
like a completely pointless universe
Max Tegmark (2:12:06.240)
that we're just trying to increase entropy
Lex Fridman (2:12:08.040)
and approach heat death into something
Max Tegmark (2:12:10.160)
that looked more goal oriented.
Lex Fridman (2:12:11.720)
Isn't that kind of beautiful?
Lex Fridman (2:12:13.280)
And then this goal orientedness through evolution
Lex Fridman (2:12:15.760)
got ever more sophisticated where you got ever more.
Lex Fridman (2:12:18.720)
And then you started getting this thing,
Lex Fridman (2:12:20.120)
which is kind of like DeepMind's mu zero and steroids,
Max Tegmark (2:12:25.280)
the ultimate self play is not what DeepMind's AI
Lex Fridman (2:12:29.400)
does against itself to get better at go.
Max Tegmark (2:12:32.080)
It's what all these little quark blobs did
Lex Fridman (2:12:34.440)
against each other in the game of survival of the fittest.
Max Tegmark (2:12:38.920)
Now, when you had really dumb bacteria
Lex Fridman (2:12:42.000)
living in a simple environment,
Max Tegmark (2:12:44.040)
there wasn't much incentive to get intelligent,
Lex Fridman (2:12:46.440)
but then the life made environment more complex.
Lex Fridman (2:12:50.880)
And then there was more incentive to get even smarter.
Lex Fridman (2:12:53.520)
And that gave the other organisms more of incentive
Max Tegmark (2:12:56.600)
to also get smarter.
Lex Fridman (2:12:57.520)
And then here we are now,
Max Tegmark (2:12:59.880)
just like mu zero learned to become world master at go
Lex Fridman (2:13:05.040)
and chess from playing against itself
Max Tegmark (2:13:07.200)
by just playing against itself.
Lex Fridman (2:13:08.560)
All the quirks here on our planet,
Max Tegmark (2:13:10.680)
the electrons have created giraffes and elephants
Lex Fridman (2:13:15.000)
and humans and love.
Max Tegmark (2:13:17.640)
I just find that really beautiful.
Lex Fridman (2:13:20.280)
And to me, that just adds to the enjoyment of love.
Max Tegmark (2:13:24.200)
It doesn't subtract anything.
Lex Fridman (2:13:25.640)
Do you feel a little more careful now?
Max Tegmark (2:13:27.320)
I feel way better, that was incredible.
Lex Fridman (2:13:30.640)
So this self play of quirks,
Max Tegmark (2:13:33.920)
taking back to the beginning of our conversation
Lex Fridman (2:13:36.320)
a little bit, there's so many exciting possibilities
Max Tegmark (2:13:39.520)
about artificial intelligence understanding
Lex Fridman (2:13:42.040)
the basic laws of physics.
Lex Fridman (2:13:44.240)
Do you think AI will help us unlock?
Lex Fridman (2:13:47.400)
There's been quite a bit of excitement
Max Tegmark (2:13:49.240)
throughout the history of physics
Lex Fridman (2:13:50.440)
of coming up with more and more general simple laws
Max Tegmark (2:13:55.440)
that explain the nature of our reality.
Lex Fridman (2:13:58.400)
And then the ultimate of that would be a theory
Max Tegmark (2:14:01.120)
of everything that combines everything together.
Lex Fridman (2:14:03.680)
Do you think it's possible that one, we humans,
Lex Fridman (2:14:07.440)
but perhaps AI systems will figure out a theory of physics
Lex Fridman (2:14:13.640)
that unifies all the laws of physics?
Max Tegmark (2:14:17.120)
Yeah, I think it's absolutely possible.
Lex Fridman (2:14:19.920)
I think it's very clear
Max Tegmark (2:14:21.360)
that we're gonna see a great boost to science.
Lex Fridman (2:14:24.960)
We're already seeing a boost actually
Max Tegmark (2:14:26.720)
from machine learning helping science.
Lex Fridman (2:14:28.760)
Alpha fold was an example,
Max Tegmark (2:14:30.280)
the decades old protein folding problem.
Lex Fridman (2:14:34.440)
So, and gradually, yeah, unless we go extinct
Max Tegmark (2:14:38.160)
by doing something dumb like we discussed,
Lex Fridman (2:14:39.720)
I think it's very likely
Max Tegmark (2:14:44.040)
that our understanding of physics will become so good
Lex Fridman (2:14:48.040)
that our technology will no longer be limited
Max Tegmark (2:14:53.040)
by human intelligence,
Lex Fridman (2:14:55.200)
but instead be limited by the laws of physics.
Lex Fridman (2:14:58.240)
So our tech today is limited
Lex Fridman (2:15:00.120)
by what we've been able to invent, right?
Max Tegmark (2:15:02.120)
I think as AI progresses,
Lex Fridman (2:15:04.920)
it'll just be limited by the speed of light
Lex Fridman (2:15:07.200)
and other physical limits,
Lex Fridman (2:15:09.240)
which would mean it's gonna be just dramatically beyond
Max Tegmark (2:15:13.960)
where we are now.
Lex Fridman (2:15:15.280)
Do you think it's a fundamentally mathematical pursuit
Max Tegmark (2:15:18.560)
of trying to understand like the laws
Lex Fridman (2:15:22.120)
of our universe from a mathematical perspective?
Lex Fridman (2:15:25.760)
So almost like if it's AI,
Lex Fridman (2:15:28.000)
it's exploring the space of like theorems
Lex Fridman (2:15:31.640)
and those kinds of things,
Lex Fridman (2:15:33.480)
or is there some other more computational ideas,
Lex Fridman (2:15:39.760)
more sort of empirical ideas?
Lex Fridman (2:15:41.280)
They're both, I would say.
Max Tegmark (2:15:43.120)
It's really interesting to look out at the landscape
Lex Fridman (2:15:45.920)
of everything we call science today.
Lex Fridman (2:15:48.000)
So here you come now with this big new hammer.
Lex Fridman (2:15:50.200)
It says machine learning on it
Lex Fridman (2:15:51.480)
and that's, you know, where are there some nails
Lex Fridman (2:15:53.360)
that you can help with here that you can hammer?
Max Tegmark (2:15:56.600)
Ultimately, if machine learning gets the point
Lex Fridman (2:16:00.120)
that it can do everything better than us,
Max Tegmark (2:16:02.800)
it will be able to help across the whole space of science.
Lex Fridman (2:16:06.000)
But maybe we can anchor it by starting a little bit
Max Tegmark (2:16:08.120)
right now near term and see how we kind of move forward.
Lex Fridman (2:16:11.640)
So like right now, first of all,
Lex Fridman (2:16:14.840)
you have a lot of big data science, right?
Lex Fridman (2:16:17.360)
Where, for example, with telescopes,
Max Tegmark (2:16:19.360)
we are able to collect way more data every hour
Lex Fridman (2:16:24.120)
than a grad student can just pour over
Lex Fridman (2:16:26.720)
like in the old times, right?
Lex Fridman (2:16:28.760)
And machine learning is already being used very effectively,
Max Tegmark (2:16:31.040)
even at MIT, to find planets around other stars,
Lex Fridman (2:16:34.680)
to detect exciting new signatures
Max Tegmark (2:16:36.560)
of new particle physics in the sky,
Lex Fridman (2:16:38.760)
to detect the ripples in the fabric of space time
Max Tegmark (2:16:42.960)
that we call gravitational waves
Lex Fridman (2:16:44.640)
caused by enormous black holes
Max Tegmark (2:16:46.520)
crashing into each other halfway
Lex Fridman (2:16:48.120)
across the observable universe.
Max Tegmark (2:16:49.920)
Machine learning is running and ticking right now,
Lex Fridman (2:16:52.680)
doing all these things,
Lex Fridman (2:16:53.800)
and it's really helping all these experimental fields.
Lex Fridman (2:16:58.440)
There is a separate front of physics,
Max Tegmark (2:17:01.880)
computational physics,
Lex Fridman (2:17:03.240)
which is getting an enormous boost also.
Lex Fridman (2:17:05.680)
So we had to do all our computations by hand, right?
Lex Fridman (2:17:09.520)
People would have these giant books
Max Tegmark (2:17:11.240)
with tables of logarithms,
Lex Fridman (2:17:12.800)
and oh my God, it pains me to even think
Lex Fridman (2:17:16.720)
how long time it would have taken to do simple stuff.
Lex Fridman (2:17:19.880)
Then we started to get little calculators and computers
Max Tegmark (2:17:23.560)
that could do some basic math for us.
Lex Fridman (2:17:26.520)
Now, what we're starting to see is
Max Tegmark (2:17:31.160)
kind of a shift from GOFI, computational physics,
Lex Fridman (2:17:35.600)
to neural network, computational physics.
Lex Fridman (2:17:40.000)
What I mean by that is most computational physics
Lex Fridman (2:17:44.520)
would be done by humans programming in
Max Tegmark (2:17:48.480)
the intelligence of how to do the computation
Lex Fridman (2:17:50.200)
into the computer.
Max Tegmark (2:17:52.440)
Just as when Garry Kasparov got his posterior kicked
Lex Fridman (2:17:55.400)
by IBM's Deep Blue in chess,
Max Tegmark (2:17:56.920)
humans had programmed in exactly how to play chess.
Lex Fridman (2:17:59.880)
Intelligence came from the humans.
Lex Fridman (2:18:01.160)
It wasn't learned, right?
Lex Fridman (2:18:03.840)
Mu zero can be not only Kasparov in chess,
Lex Fridman (2:18:08.480)
but also Stockfish,
Lex Fridman (2:18:09.880)
which is the best sort of GOFI chess program.
Max Tegmark (2:18:12.560)
By learning, and we're seeing more of that now,
Lex Fridman (2:18:16.560)
that shift beginning to happen in physics.
Lex Fridman (2:18:18.320)
So let me give you an example.
Lex Fridman (2:18:20.520)
So lattice QCD is an area of physics
Max Tegmark (2:18:24.120)
whose goal is basically to take the periodic table
Lex Fridman (2:18:27.320)
and just compute the whole thing from first principles.
Max Tegmark (2:18:31.120)
This is not the search for theory of everything.
Lex Fridman (2:18:33.920)
We already know the theory
Max Tegmark (2:18:36.360)
that's supposed to produce as output the periodic table,
Lex Fridman (2:18:39.720)
which atoms are stable, how heavy they are,
Max Tegmark (2:18:42.720)
all that good stuff, their spectral lines.
Lex Fridman (2:18:45.840)
It's a theory, lattice QCD,
Max Tegmark (2:18:48.120)
you can put it on your tshirt.
Lex Fridman (2:18:50.000)
Our colleague Frank Wilczek
Max Tegmark (2:18:51.160)
got the Nobel Prize for working on it.
Lex Fridman (2:18:54.520)
But the math is just too hard for us to solve.
Max Tegmark (2:18:56.600)
We have not been able to start with these equations
Lex Fridman (2:18:58.640)
and solve them to the extent that we can predict, oh yeah.
Lex Fridman (2:19:01.440)
And then there is carbon,
Lex Fridman (2:19:03.360)
and this is what the spectrum of the carbon atom looks like.
Lex Fridman (2:19:07.000)
But awesome people are building
Lex Fridman (2:19:09.960)
these supercomputer simulations
Max Tegmark (2:19:12.040)
where you just put in these equations
Lex Fridman (2:19:14.960)
and you make a big cubic lattice of space,
Max Tegmark (2:19:20.680)
or actually it's a very small lattice
Lex Fridman (2:19:22.080)
because you're going down to the subatomic scale,
Lex Fridman (2:19:25.640)
and you try to solve it.
Lex Fridman (2:19:26.880)
But it's just so computationally expensive
Max Tegmark (2:19:28.960)
that we still haven't been able to calculate things
Lex Fridman (2:19:31.840)
as accurately as we measure them in many cases.
Lex Fridman (2:19:34.960)
And now machine learning is really revolutionizing this.
Lex Fridman (2:19:37.520)
So my colleague Fiala Shanahan at MIT, for example,
Max Tegmark (2:19:40.040)
she's been using this really cool
Lex Fridman (2:19:43.280)
machine learning technique called normalizing flows,
Max Tegmark (2:19:47.560)
where she's realized she can actually speed up
Lex Fridman (2:19:49.800)
the calculation dramatically
Max Tegmark (2:19:52.160)
by having the AI learn how to do things faster.
Lex Fridman (2:19:55.680)
Another area like this
Max Tegmark (2:19:57.280)
where we suck up an enormous amount of supercomputer time
Lex Fridman (2:20:02.280)
to do physics is black hole collisions.
Lex Fridman (2:20:05.480)
So now that we've done the sexy stuff
Lex Fridman (2:20:06.880)
of detecting a bunch of this with LIGO and other experiments,
Max Tegmark (2:20:09.960)
we want to be able to know what we're seeing.
Lex Fridman (2:20:13.360)
And so it's a very simple conceptual problem.
Max Tegmark (2:20:16.480)
It's the two body problem.
Lex Fridman (2:20:19.000)
Newton solved it for classical gravity hundreds of years ago,
Lex Fridman (2:20:23.000)
but the two body problem is still not fully solved.
Lex Fridman (2:20:26.080)
For black holes.
Max Tegmark (2:20:26.920)
Black holes, yes, and Einstein's gravity
Lex Fridman (2:20:29.000)
because they won't just orbit in space,
Max Tegmark (2:20:31.120)
they won't just orbit each other forever anymore,
Lex Fridman (2:20:33.560)
two things, they give off gravitational waves
Lex Fridman (2:20:36.080)
and make sure they crash into each other.
Lex Fridman (2:20:37.800)
And the game, what you want to do is you want to figure out,
Max Tegmark (2:20:40.320)
okay, what kind of wave comes out
Lex Fridman (2:20:43.480)
as a function of the masses of the two black holes,
Max Tegmark (2:20:46.320)
as a function of how they're spinning,
Lex Fridman (2:20:48.120)
relative to each other, et cetera.
Lex Fridman (2:20:50.720)
And that is so hard.
Lex Fridman (2:20:52.040)
It can take months of supercomputer time
Lex Fridman (2:20:54.200)
and massive numbers of cores to do it.
Lex Fridman (2:20:56.200)
Now, wouldn't it be great if you can use machine learning
Lex Fridman (2:21:01.240)
to greatly speed that up, right?
Lex Fridman (2:21:04.760)
Now you can use the expensive old GoFi calculation
Max Tegmark (2:21:09.360)
as the truth, and then see if machine learning
Lex Fridman (2:21:11.920)
can figure out a smarter, faster way
Max Tegmark (2:21:13.600)
of getting the right answer.
Lex Fridman (2:21:16.320)
Yet another area, like computational physics.
Max Tegmark (2:21:20.000)
These are probably the big three
Lex Fridman (2:21:22.280)
that suck up the most computer time.
Max Tegmark (2:21:24.240)
Lattice QCD, black hole collisions,
Lex Fridman (2:21:27.160)
and cosmological simulations,
Max Tegmark (2:21:29.560)
where you take not a subatomic thing
Lex Fridman (2:21:32.280)
and try to figure out the mass of the proton,
Lex Fridman (2:21:34.400)
but you take something enormous
Lex Fridman (2:21:37.680)
and try to look at how all the galaxies get formed in there.
Max Tegmark (2:21:41.320)
There again, there are a lot of very cool ideas right now
Lex Fridman (2:21:44.720)
about how you can use machine learning
Max Tegmark (2:21:46.080)
to do this sort of stuff better.
Lex Fridman (2:21:49.760)
The difference between this and the big data
Lex Fridman (2:21:51.560)
is you kind of make the data yourself, right?
Lex Fridman (2:21:54.560)
So, and then finally,
Max Tegmark (2:21:58.440)
we're looking over the physics landscape
Lex Fridman (2:22:00.200)
and seeing what can we hammer with machine learning, right?
Lex Fridman (2:22:02.120)
So we talked about experimental data, big data,
Lex Fridman (2:22:05.520)
discovering cool stuff that we humans
Max Tegmark (2:22:07.880)
then look more closely at.
Lex Fridman (2:22:09.520)
Then we talked about taking the expensive computations
Max Tegmark (2:22:13.440)
we're doing now and figuring out
Lex Fridman (2:22:15.040)
how to do them much faster and better with AI.
Lex Fridman (2:22:18.560)
And finally, let's go really theoretical.
Lex Fridman (2:22:21.920)
So things like discovering equations,
Max Tegmark (2:22:25.000)
having deep fundamental insights,
Lex Fridman (2:22:30.240)
this is something closest to what I've been doing
Max Tegmark (2:22:33.040)
in my group.
Lex Fridman (2:22:33.880)
We talked earlier about the whole AI Feynman project,
Max Tegmark (2:22:35.920)
where if you just have some data,
Lex Fridman (2:22:37.920)
how do you automatically discover equations
Max Tegmark (2:22:39.840)
that seem to describe this well,
Lex Fridman (2:22:42.160)
that you can then go back as a human
Lex Fridman (2:22:44.120)
and then work with and test and explore.
Lex Fridman (2:22:46.640)
And you asked a really good question also
Max Tegmark (2:22:50.320)
about if this is sort of a search problem in some sense.
Lex Fridman (2:22:54.000)
That's very deep actually what you said, because it is.
Max Tegmark (2:22:56.880)
Suppose I ask you to prove some mathematical theorem.
Lex Fridman (2:23:01.680)
What is a proof in math?
Max Tegmark (2:23:02.960)
It's just a long string of steps, logical steps
Lex Fridman (2:23:05.360)
that you can write out with symbols.
Lex Fridman (2:23:07.920)
And once you find it, it's very easy to write a program
Lex Fridman (2:23:10.240)
to check whether it's a valid proof or not.
Lex Fridman (2:23:14.640)
So why is it so hard to prove it?
Lex Fridman (2:23:16.080)
Well, because there are ridiculously many possible
Lex Fridman (2:23:19.040)
candidate proofs you could write down, right?
Lex Fridman (2:23:21.600)
If the proof contains 10,000 symbols,
Max Tegmark (2:23:25.440)
even if there were only 10 options
Lex Fridman (2:23:27.760)
for what each symbol could be,
Max Tegmark (2:23:29.120)
that's 10 to the power of 1,000 possible proofs,
Lex Fridman (2:23:33.440)
which is way more than there are atoms in our universe.
Lex Fridman (2:23:36.080)
So you could say it's trivial to prove these things.
Lex Fridman (2:23:38.400)
You just write a computer, generate all strings,
Lex Fridman (2:23:41.200)
and then check, is this a valid proof?
Lex Fridman (2:23:43.680)
No.
Lex Fridman (2:23:44.520)
Is this a valid proof?
Lex Fridman (2:23:45.360)
Is this a valid proof?
Max Tegmark (2:23:46.400)
No.
Lex Fridman (2:23:47.720)
And then you just keep doing this forever.
Lex Fridman (2:23:51.960)
But there are a lot of,
Lex Fridman (2:23:53.160)
but it is fundamentally a search problem.
Max Tegmark (2:23:55.120)
You just want to search the space of all those,
Lex Fridman (2:23:57.000)
all strings of symbols to find one that is the proof, right?
Lex Fridman (2:24:03.880)
And there's a whole area of machine learning called search.
Lex Fridman (2:24:08.800)
How do you search through some giant space
Lex Fridman (2:24:10.600)
to find the needle in the haystack?
Lex Fridman (2:24:12.400)
And it's easier in cases
Max Tegmark (2:24:14.800)
where there's a clear measure of good,
Lex Fridman (2:24:17.160)
like you're not just right or wrong,
Lex Fridman (2:24:18.800)
but this is better and this is worse,
Lex Fridman (2:24:20.640)
so you can maybe get some hints
Max Tegmark (2:24:21.800)
as to which direction to go in.
Lex Fridman (2:24:23.800)
That's why we talked about neural networks work so well.
Max Tegmark (2:24:28.400)
I mean, that's such a human thing
Lex Fridman (2:24:30.680)
of that moment of genius
Max Tegmark (2:24:32.280)
of figuring out the intuition of good, essentially.
Lex Fridman (2:24:37.360)
I mean, we thought that that was...
Lex Fridman (2:24:38.680)
Or is it?
Lex Fridman (2:24:40.120)
Maybe it's not, right?
Lex Fridman (2:24:41.320)
We thought that about chess, right?
Lex Fridman (2:24:42.720)
That the ability to see like 10, 15,
Max Tegmark (2:24:46.880)
sometimes 20 steps ahead was not a calculation
Lex Fridman (2:24:50.680)
that humans were performing.
Max Tegmark (2:24:51.760)
It was some kind of weird intuition
Lex Fridman (2:24:53.720)
about different patterns, about board positions,
Max Tegmark (2:24:57.280)
about the relative positions,
Lex Fridman (2:24:59.440)
somehow stitching stuff together.
Lex Fridman (2:25:01.640)
And a lot of it is just like intuition,
Lex Fridman (2:25:03.920)
but then you have like alpha,
Max Tegmark (2:25:05.960)
I guess zero be the first one that did the self play.
Lex Fridman (2:25:10.400)
It just came up with this.
Max Tegmark (2:25:12.160)
It was able to learn through self play mechanism,
Lex Fridman (2:25:14.560)
this kind of intuition.
Max Tegmark (2:25:16.040)
Exactly.
Lex Fridman (2:25:16.880)
But just like you said, it's so fascinating to think,
Max Tegmark (2:25:19.960)
well, they're in the space of totally new ideas.
Lex Fridman (2:25:24.640)
Can that be done in developing theorems?
Max Tegmark (2:25:28.960)
We know it can be done by neural networks
Lex Fridman (2:25:30.800)
because we did it with the neural networks
Max Tegmark (2:25:32.280)
in the craniums of the great mathematicians of humanity.
Lex Fridman (2:25:36.280)
And I'm so glad you brought up alpha zero
Max Tegmark (2:25:38.640)
because that's the counter example.
Lex Fridman (2:25:39.960)
It turned out we were flattering ourselves
Max Tegmark (2:25:41.840)
when we said intuition is something different.
Lex Fridman (2:25:45.360)
Only humans can do it.
Max Tegmark (2:25:46.520)
It's not information processing.
Lex Fridman (2:25:50.880)
It used to be that way.
Max Tegmark (2:25:53.720)
Again, it's really instructive, I think,
Lex Fridman (2:25:56.200)
to compare the chess computer Deep Blue
Max Tegmark (2:25:58.480)
that beat Kasparov with alpha zero
Lex Fridman (2:26:02.040)
that beat Lisa Dahl at Go.
Max Tegmark (2:26:04.280)
Because for Deep Blue, there was no intuition.
Lex Fridman (2:26:08.640)
There was some, humans had programmed in some intuition.
Max Tegmark (2:26:12.000)
After humans had played a lot of games,
Lex Fridman (2:26:13.600)
they told the computer, count the pawn as one point,
Max Tegmark (2:26:16.520)
the bishop is three points, rook is five points,
Lex Fridman (2:26:19.920)
and so on, you add it all up,
Lex Fridman (2:26:21.120)
and then you add some extra points for past pawns
Lex Fridman (2:26:23.400)
and subtract if the opponent has it and blah, blah, blah.
Lex Fridman (2:26:28.280)
And then what Deep Blue did was just search.
Lex Fridman (2:26:32.520)
Just very brute force and tried many, many moves ahead,
Max Tegmark (2:26:34.960)
all these combinations and a prune tree search.
Lex Fridman (2:26:37.400)
And it could think much faster than Kasparov, and it won.
Lex Fridman (2:26:42.680)
And that, I think, inflated our egos
Lex Fridman (2:26:45.440)
in a way it shouldn't have,
Max Tegmark (2:26:46.560)
because people started to say, yeah, yeah,
Lex Fridman (2:26:48.760)
it's just brute force search, but it has no intuition.
Max Tegmark (2:26:52.280)
Alpha zero really popped our bubble there,
Lex Fridman (2:26:57.760)
because what alpha zero does,
Max Tegmark (2:27:00.880)
yes, it does also do some of that tree search,
Lex Fridman (2:27:03.880)
but it also has this intuition module,
Max Tegmark (2:27:06.560)
which in geek speak is called a value function,
Lex Fridman (2:27:09.560)
where it just looks at the board
Lex Fridman (2:27:11.120)
and comes up with a number for how good is that position.
Lex Fridman (2:27:14.960)
The difference was no human told it
Lex Fridman (2:27:17.960)
how good the position is, it just learned it.
Lex Fridman (2:27:22.480)
And mu zero is the coolest or scariest of all,
Max Tegmark (2:27:26.840)
depending on your mood,
Lex Fridman (2:27:28.320)
because the same basic AI system
Max Tegmark (2:27:33.040)
will learn what the good board position is,
Lex Fridman (2:27:35.320)
regardless of whether it's chess or Go or Shogi
Max Tegmark (2:27:38.640)
or Pacman or Lady Pacman or Breakout or Space Invaders
Lex Fridman (2:27:42.920)
or any number, a bunch of other games.
Max Tegmark (2:27:45.000)
You don't tell it anything,
Lex Fridman (2:27:45.840)
and it gets this intuition after a while for what's good.
Lex Fridman (2:27:49.760)
So this is very hopeful for science, I think,
Lex Fridman (2:27:52.760)
because if it can get intuition
Max Tegmark (2:27:55.240)
for what's a good position there,
Lex Fridman (2:27:57.280)
maybe it can also get intuition
Max Tegmark (2:27:58.880)
for what are some good directions to go
Lex Fridman (2:28:00.640)
if you're trying to prove something.
Max Tegmark (2:28:03.040)
I often, one of the most fun things in my science career
Lex Fridman (2:28:06.400)
is when I've been able to prove some theorem about something
Lex Fridman (2:28:08.600)
and it's very heavily intuition guided, of course.
Lex Fridman (2:28:12.160)
I don't sit and try all random strings.
Max Tegmark (2:28:14.080)
I have a hunch that, you know,
Lex Fridman (2:28:16.280)
this reminds me a little bit of about this other proof
Max Tegmark (2:28:18.840)
I've seen for this thing.
Lex Fridman (2:28:19.920)
So maybe I first, what if I try this?
Max Tegmark (2:28:22.520)
Nah, that didn't work out.
Lex Fridman (2:28:24.720)
But this reminds me actually,
Max Tegmark (2:28:25.840)
the way this failed reminds me of that.
Lex Fridman (2:28:28.560)
So combining the intuition with all these brute force
Max Tegmark (2:28:33.880)
capabilities, I think it's gonna be able to help physics too.
Lex Fridman (2:28:38.520)
Do you think there'll be a day when an AI system
Max Tegmark (2:28:42.880)
being the primary contributor, let's say 90% plus,
Lex Fridman (2:28:46.400)
wins the Nobel Prize in physics?
Max Tegmark (2:28:50.400)
Obviously they'll give it to the humans
Lex Fridman (2:28:51.960)
because we humans don't like to give prizes to machines.
Max Tegmark (2:28:54.800)
It'll give it to the humans behind the system.
Lex Fridman (2:28:57.560)
You could argue that AI has already been involved
Max Tegmark (2:28:59.920)
in some Nobel Prizes, probably,
Lex Fridman (2:29:01.560)
maybe something with black holes and stuff like that.
Max Tegmark (2:29:03.560)
Yeah, we don't like giving prizes to other life forms.
Lex Fridman (2:29:07.160)
If someone wins a horse racing contest,
Max Tegmark (2:29:09.720)
they don't give the prize to the horse either.
Lex Fridman (2:29:11.360)
That's true.
Lex Fridman (2:29:13.400)
But do you think that we might be able to see
Lex Fridman (2:29:16.000)
something like that in our lifetimes when AI,
Lex Fridman (2:29:19.200)
so like the first system I would say
Lex Fridman (2:29:21.840)
that makes us think about a Nobel Prize seriously
Max Tegmark (2:29:25.360)
is like Alpha Fold is making us think about
Lex Fridman (2:29:28.760)
in medicine, physiology, a Nobel Prize,
Max Tegmark (2:29:31.960)
perhaps discoveries that are direct result
Lex Fridman (2:29:34.080)
of something that's discovered by Alpha Fold.
Lex Fridman (2:29:36.640)
Do you think in physics we might be able
Lex Fridman (2:29:39.560)
to see that in our lifetimes?
Max Tegmark (2:29:41.520)
I think what's probably gonna happen
Lex Fridman (2:29:43.520)
is more of a blurring of the distinctions.
Lex Fridman (2:29:46.880)
So today if somebody uses a computer
Lex Fridman (2:29:53.000)
to do a computation that gives them the Nobel Prize,
Max Tegmark (2:29:54.920)
nobody's gonna dream of giving the prize to the computer.
Lex Fridman (2:29:57.160)
They're gonna be like, that was just a tool.
Max Tegmark (2:29:59.000)
I think for these things also,
Lex Fridman (2:30:02.120)
people are just gonna for a long time
Max Tegmark (2:30:04.000)
view the computer as a tool.
Lex Fridman (2:30:06.120)
But what's gonna change is the ubiquity of machine learning.
Max Tegmark (2:30:11.120)
I think at some point in my lifetime,
Lex Fridman (2:30:17.120)
finding a human physicist who knows nothing
Max Tegmark (2:30:21.400)
about machine learning is gonna be almost as hard
Lex Fridman (2:30:23.800)
as it is today finding a human physicist
Max Tegmark (2:30:25.960)
who doesn't says, oh, I don't know anything about computers
Lex Fridman (2:30:29.160)
or I don't use math.
Max Tegmark (2:30:30.880)
That would just be a ridiculous concept.
Lex Fridman (2:30:34.000)
You see, but the thing is there is a magic moment though,
Max Tegmark (2:30:38.240)
like with Alpha Zero, when the system surprises us
Lex Fridman (2:30:42.320)
in a way where the best people in the world
Max Tegmark (2:30:46.680)
truly learn something from the system
Lex Fridman (2:30:48.960)
in a way where you feel like it's another entity.
Max Tegmark (2:30:52.480)
Like the way people, the way Magnus Carlsen,
Lex Fridman (2:30:54.920)
the way certain people are looking at the work of Alpha Zero,
Max Tegmark (2:30:58.080)
it's like, it truly is no longer a tool
Lex Fridman (2:31:02.960)
in the sense that it doesn't feel like a tool.
Max Tegmark (2:31:06.680)
It feels like some other entity.
Lex Fridman (2:31:08.960)
So there's a magic difference like where you're like,
Max Tegmark (2:31:13.320)
if an AI system is able to come up with an insight
Lex Fridman (2:31:17.320)
that surprises everybody in some like major way
Max Tegmark (2:31:23.760)
that's a phase shift in our understanding
Lex Fridman (2:31:25.960)
of some particular science
Max Tegmark (2:31:27.760)
or some particular aspect of physics,
Lex Fridman (2:31:30.040)
I feel like that is no longer a tool.
Lex Fridman (2:31:32.680)
And then you can start to say
Lex Fridman (2:31:35.800)
that like it perhaps deserves the prize.
Lex Fridman (2:31:38.720)
So for sure, the more important
Lex Fridman (2:31:40.640)
and the more fundamental transformation
Max Tegmark (2:31:43.120)
of the 21st century science is exactly what you're saying,
Lex Fridman (2:31:46.640)
which is probably everybody will be doing machine learning.
Max Tegmark (2:31:50.680)
It's to some degree.
Lex Fridman (2:31:51.560)
Like if you want to be successful
Max Tegmark (2:31:54.760)
at unlocking the mysteries of science,
Lex Fridman (2:31:57.560)
you should be doing machine learning.
Lex Fridman (2:31:58.800)
But it's just exciting to think about like,
Lex Fridman (2:32:01.440)
whether there'll be one that comes along
Max Tegmark (2:32:03.080)
that's super surprising and they'll make us question
Lex Fridman (2:32:08.000)
like who the real inventors are in this world.
Max Tegmark (2:32:10.320)
Yeah.
Lex Fridman (2:32:11.640)
Yeah, I think the question of,
Lex Fridman (2:32:14.240)
isn't if it's gonna happen, but when?
Lex Fridman (2:32:15.880)
And, but it's important.
Max Tegmark (2:32:17.960)
Honestly, in my mind, the time when that happens
Lex Fridman (2:32:20.840)
is also more or less the same time
Max Tegmark (2:32:23.360)
when we get artificial general intelligence.
Lex Fridman (2:32:25.560)
And then we have a lot bigger things to worry about
Lex Fridman (2:32:28.160)
than whether we should get the Nobel prize or not, right?
Lex Fridman (2:32:31.000)
Yeah.
Max Tegmark (2:32:31.840)
Because when you have machines
Lex Fridman (2:32:35.000)
that can outperform our best scientists at science,
Max Tegmark (2:32:39.360)
they can probably outperform us
Lex Fridman (2:32:41.040)
at a lot of other stuff as well,
Max Tegmark (2:32:44.440)
which can at a minimum make them
Lex Fridman (2:32:46.440)
incredibly powerful agents in the world.
Lex Fridman (2:32:49.440)
And I think it's a mistake to think
Lex Fridman (2:32:53.160)
we only have to start worrying about loss of control
Max Tegmark (2:32:57.040)
when machines get to AGI across the board,
Lex Fridman (2:32:59.720)
where they can do everything, all our jobs.
Max Tegmark (2:33:02.160)
Long before that, they'll be hugely influential.
Lex Fridman (2:33:07.880)
We talked at length about how the hacking of our minds
Max Tegmark (2:33:12.560)
with algorithms trying to get us glued to our screens,
Lex Fridman (2:33:18.440)
right, has already had a big impact on society.
Max Tegmark (2:33:22.320)
That was an incredibly dumb algorithm
Lex Fridman (2:33:24.080)
in the grand scheme of things, right?
Max Tegmark (2:33:25.840)
The supervised machine learning,
Lex Fridman (2:33:27.840)
yet that had huge impact.
Lex Fridman (2:33:29.520)
So I just don't want us to be lulled
Lex Fridman (2:33:32.080)
into false sense of security
Lex Fridman (2:33:33.280)
and think there won't be any societal impact
Lex Fridman (2:33:35.560)
until things reach human level,
Max Tegmark (2:33:37.040)
because it's happening already.
Lex Fridman (2:33:38.280)
And I was just thinking the other week,
Max Tegmark (2:33:40.560)
when I see some scaremonger going,
Lex Fridman (2:33:44.880)
oh, the robots are coming,
Max Tegmark (2:33:47.080)
the implication is always that they're coming to kill us.
Lex Fridman (2:33:50.280)
Yeah.
Lex Fridman (2:33:51.120)
And maybe you should have worried about that
Lex Fridman (2:33:52.360)
if you were in Nagorno Karabakh
Max Tegmark (2:33:54.720)
during the recent war there.
Lex Fridman (2:33:55.720)
But more seriously, the robots are coming right now,
Lex Fridman (2:34:01.440)
but they're mainly not coming to kill us.
Lex Fridman (2:34:03.160)
They're coming to hack us.
Max Tegmark (2:34:06.000)
They're coming to hack our minds,
Lex Fridman (2:34:08.200)
into buying things that maybe we didn't need,
Max Tegmark (2:34:11.280)
to vote for people who may not have
Lex Fridman (2:34:13.200)
our best interest in mind.
Lex Fridman (2:34:15.360)
And it's kind of humbling, I think,
Lex Fridman (2:34:17.600)
actually, as a human being to admit
Max Tegmark (2:34:20.120)
that it turns out that our minds are actually
Lex Fridman (2:34:22.360)
much more hackable than we thought.
Lex Fridman (2:34:24.760)
And the ultimate insult is that we are actually
Lex Fridman (2:34:27.040)
getting hacked by the machine learning algorithms
Max Tegmark (2:34:30.400)
that are, in some objective sense,
Lex Fridman (2:34:31.560)
much dumber than us, you know?
Lex Fridman (2:34:33.960)
But maybe we shouldn't be so surprised
Lex Fridman (2:34:35.720)
because, you know, how do you feel about cute puppies?
Max Tegmark (2:34:40.520)
Love them.
Lex Fridman (2:34:41.600)
So, you know, you would probably argue
Max Tegmark (2:34:43.640)
that in some across the board measure,
Lex Fridman (2:34:46.120)
you're more intelligent than they are,
Lex Fridman (2:34:47.680)
but boy, are cute puppies good at hacking us, right?
Lex Fridman (2:34:50.760)
Yeah.
Max Tegmark (2:34:51.600)
They move into our house, persuade us to feed them
Lex Fridman (2:34:53.720)
and do all these things.
Lex Fridman (2:34:54.560)
And what do they ever do but for us?
Lex Fridman (2:34:56.600)
Yeah.
Lex Fridman (2:34:57.440)
Other than being cute and making us feel good, right?
Lex Fridman (2:35:00.520)
So if puppies can hack us,
Max Tegmark (2:35:03.080)
maybe we shouldn't be so surprised
Lex Fridman (2:35:04.920)
if pretty dumb machine learning algorithms can hack us too.
Max Tegmark (2:35:09.040)
Not to speak of cats, which is another level.
Lex Fridman (2:35:11.680)
And I think we should,
Max Tegmark (2:35:13.400)
to counter your previous point about there,
Lex Fridman (2:35:15.640)
let us not think about evil creatures in this world.
Max Tegmark (2:35:18.040)
We can all agree that cats are as close
Lex Fridman (2:35:20.480)
to objective evil as we can get.
Lex Fridman (2:35:22.960)
But that's just me saying that.
Lex Fridman (2:35:24.400)
Okay, so you have.
Lex Fridman (2:35:25.480)
Have you seen the cartoon?
Lex Fridman (2:35:27.320)
I think it's maybe the onion
Max Tegmark (2:35:31.760)
with this incredibly cute kitten.
Lex Fridman (2:35:33.720)
And it just says, it's underneath something
Max Tegmark (2:35:36.840)
that thinks about murder all day.
Lex Fridman (2:35:38.920)
Exactly.
Max Tegmark (2:35:41.560)
That's accurate.
Lex Fridman (2:35:43.080)
You've mentioned offline that there might be a link
Max Tegmark (2:35:45.200)
between post biological AGI and SETI.
Lex Fridman (2:35:47.960)
So last time we talked,
Max Tegmark (2:35:52.520)
you've talked about this intuition
Lex Fridman (2:35:54.920)
that we humans might be quite unique
Max Tegmark (2:35:59.280)
in our galactic neighborhood.
Lex Fridman (2:36:02.360)
Perhaps our galaxy,
Max Tegmark (2:36:03.680)
perhaps the entirety of the observable universe
Lex Fridman (2:36:06.360)
who might be the only intelligent civilization here,
Max Tegmark (2:36:10.840)
which is, and you argue pretty well for that thought.
Lex Fridman (2:36:17.720)
So I have a few little questions around this.
Max Tegmark (2:36:21.240)
One, the scientific question,
Lex Fridman (2:36:24.680)
in which way would you be,
Max Tegmark (2:36:29.240)
if you were wrong in that intuition,
Lex Fridman (2:36:33.960)
in which way do you think you would be surprised?
Lex Fridman (2:36:36.680)
Like why were you wrong?
Lex Fridman (2:36:38.520)
We find out that you ended up being wrong.
Lex Fridman (2:36:41.600)
Like in which dimension?
Lex Fridman (2:36:43.880)
So like, is it because we can't see them?
Max Tegmark (2:36:48.400)
Is it because the nature of their intelligence
Lex Fridman (2:36:51.320)
or the nature of their life is totally different
Lex Fridman (2:36:54.760)
than we can possibly imagine?
Lex Fridman (2:36:56.760)
Is it because the,
Max Tegmark (2:37:00.680)
I mean, something about the great filters
Lex Fridman (2:37:02.640)
and surviving them,
Max Tegmark (2:37:04.440)
or maybe because we're being protected from signals,
Lex Fridman (2:37:08.760)
all those explanations for why we haven't heard
Max Tegmark (2:37:15.120)
a big, loud, like red light that says we're here.
Lex Fridman (2:37:21.680)
So there are actually two separate things there
Max Tegmark (2:37:23.520)
that I could be wrong about,
Lex Fridman (2:37:24.720)
two separate claims that I made, right?
Max Tegmark (2:37:28.920)
One of them is, I made the claim,
Lex Fridman (2:37:32.240)
I think most civilizations,
Max Tegmark (2:37:36.960)
when you're going from simple bacteria like things
Lex Fridman (2:37:41.800)
to space colonizing civilizations,
Max Tegmark (2:37:47.840)
they spend only a very, very tiny fraction
Lex Fridman (2:37:50.840)
of their life being where we are.
Max Tegmark (2:37:55.160)
That I could be wrong about.
Lex Fridman (2:37:57.280)
The other one I could be wrong about
Max Tegmark (2:37:58.760)
is the quite different statement that I think that actually
Lex Fridman (2:38:01.520)
I'm guessing that we are the only civilization
Max Tegmark (2:38:04.680)
in our observable universe
Lex Fridman (2:38:06.120)
from which light has reached us so far
Max Tegmark (2:38:08.240)
that's actually gotten far enough to invent telescopes.
Lex Fridman (2:38:12.320)
So let's talk about maybe both of them in turn
Max Tegmark (2:38:13.960)
because they really are different.
Lex Fridman (2:38:15.000)
The first one, if you look at the N equals one,
Lex Fridman (2:38:19.880)
the data point we have on this planet, right?
Lex Fridman (2:38:22.080)
So we spent four and a half billion years
Lex Fridman (2:38:25.880)
fluxing around on this planet with life, right?
Lex Fridman (2:38:28.240)
We got, and most of it was pretty lame stuff
Max Tegmark (2:38:32.080)
from an intelligence perspective,
Lex Fridman (2:38:33.640)
you know, it was bacteria and then the dinosaurs spent,
Lex Fridman (2:38:39.200)
then the things gradually accelerated, right?
Lex Fridman (2:38:41.280)
Then the dinosaurs spent over a hundred million years
Max Tegmark (2:38:43.600)
stomping around here without even inventing smartphones.
Lex Fridman (2:38:46.960)
And then very recently, you know,
Max Tegmark (2:38:50.240)
it's only, we've only spent 400 years
Lex Fridman (2:38:52.120)
going from Newton to us, right?
Max Tegmark (2:38:55.320)
In terms of technology.
Lex Fridman (2:38:56.480)
And look what we've done even, you know,
Max Tegmark (2:39:00.160)
when I was a little kid, there was no internet even.
Lex Fridman (2:39:02.600)
So it's, I think it's pretty likely for,
Lex Fridman (2:39:05.880)
in this case of this planet, right?
Lex Fridman (2:39:08.160)
That we're either gonna really get our act together
Lex Fridman (2:39:12.200)
and start spreading life into space, the century,
Lex Fridman (2:39:15.080)
and doing all sorts of great things,
Max Tegmark (2:39:16.440)
or we're gonna wipe out.
Lex Fridman (2:39:18.880)
It's a little hard.
Max Tegmark (2:39:20.080)
If I, I could be wrong in the sense that maybe
Lex Fridman (2:39:23.480)
what happened on this earth is very atypical.
Lex Fridman (2:39:25.800)
And for some reason, what's more common on other planets
Lex Fridman (2:39:28.520)
is that they spend an enormously long time
Max Tegmark (2:39:31.440)
futzing around with the ham radio and things,
Lex Fridman (2:39:33.720)
but they just never really take it to the next level
Max Tegmark (2:39:36.200)
for reasons I don't, I haven't understood.
Lex Fridman (2:39:38.400)
I'm humble and open to that.
Lex Fridman (2:39:40.200)
But I would bet at least 10 to one
Lex Fridman (2:39:42.880)
that our situation is more typical
Max Tegmark (2:39:45.160)
because the whole thing with Moore's law
Lex Fridman (2:39:46.760)
and accelerating technology,
Max Tegmark (2:39:48.160)
it's pretty obvious why it's happening.
Lex Fridman (2:39:51.200)
Everything that grows exponentially,
Max Tegmark (2:39:52.880)
we call it an explosion,
Lex Fridman (2:39:54.080)
whether it's a population explosion or a nuclear explosion,
Max Tegmark (2:39:56.640)
it's always caused by the same thing.
Lex Fridman (2:39:58.000)
It's that the next step triggers a step after that.
Lex Fridman (2:40:01.480)
So I, we, tomorrow's technology,
Lex Fridman (2:40:04.320)
today's technology enables tomorrow's technology
Lex Fridman (2:40:06.760)
and that enables the next level.
Lex Fridman (2:40:09.080)
And as I think, because the technology is always better,
Max Tegmark (2:40:13.800)
of course, the steps can come faster and faster.
Lex Fridman (2:40:17.200)
On the other question that I might be wrong about,
Max Tegmark (2:40:19.160)
that's the much more controversial one, I think.
Lex Fridman (2:40:22.320)
But before we close out on this thing about,
Max Tegmark (2:40:24.920)
if, the first one, if it's true
Lex Fridman (2:40:27.080)
that most civilizations spend only a very short amount
Max Tegmark (2:40:30.520)
of their total time in the stage, say,
Lex Fridman (2:40:32.880)
between inventing
Max Tegmark (2:40:37.320)
telescopes or mastering electricity
Lex Fridman (2:40:40.760)
and leaving there and doing space travel,
Max Tegmark (2:40:43.880)
if that's actually generally true,
Lex Fridman (2:40:46.200)
then that should apply also elsewhere out there.
Lex Fridman (2:40:49.000)
So we should be very, very,
Lex Fridman (2:40:51.040)
we should be very, very surprised
Max Tegmark (2:40:52.920)
if we find some random civilization
Lex Fridman (2:40:55.480)
and we happen to catch them exactly
Max Tegmark (2:40:56.920)
in that very, very short stage.
Lex Fridman (2:40:58.800)
It's much more likely
Max Tegmark (2:40:59.640)
that we find a planet full of bacteria.
Lex Fridman (2:41:02.960)
Or that we find some civilization
Max Tegmark (2:41:05.560)
that's already post biological
Lex Fridman (2:41:07.480)
and has done some really cool galactic construction projects
Max Tegmark (2:41:11.880)
in their galaxy.
Lex Fridman (2:41:13.360)
Would we be able to recognize them, do you think?
Max Tegmark (2:41:15.200)
Is it possible that we just can't,
Lex Fridman (2:41:17.480)
I mean, this post biological world,
Lex Fridman (2:41:21.120)
could it be just existing in some other dimension?
Lex Fridman (2:41:23.520)
It could be just all a virtual reality game
Max Tegmark (2:41:26.280)
for them or something, I don't know,
Lex Fridman (2:41:28.480)
that it changes completely
Max Tegmark (2:41:30.560)
where we won't be able to detect.
Lex Fridman (2:41:32.880)
We have to be honestly very humble about this.
Max Tegmark (2:41:35.280)
I think I said earlier the number one principle
Lex Fridman (2:41:39.000)
of being a scientist is you have to be humble
Lex Fridman (2:41:40.840)
and willing to acknowledge that everything we think,
Lex Fridman (2:41:42.960)
guess might be totally wrong.
Max Tegmark (2:41:45.040)
Of course, you could imagine some civilization
Lex Fridman (2:41:46.960)
where they all decide to become Buddhists
Lex Fridman (2:41:48.640)
and very inward looking
Lex Fridman (2:41:49.880)
and just move into their little virtual reality
Lex Fridman (2:41:52.360)
and not disturb the flora and fauna around them
Lex Fridman (2:41:55.120)
and we might not notice them.
Lex Fridman (2:41:58.120)
But this is a numbers game, right?
Lex Fridman (2:41:59.960)
If you have millions of civilizations out there
Max Tegmark (2:42:02.280)
or billions of them,
Lex Fridman (2:42:03.680)
all it takes is one with a more ambitious mentality
Max Tegmark (2:42:08.080)
that decides, hey, we are gonna go out
Lex Fridman (2:42:10.280)
and settle a bunch of other solar systems
Lex Fridman (2:42:15.520)
and maybe galaxies.
Lex Fridman (2:42:17.560)
And then it doesn't matter
Max Tegmark (2:42:18.440)
if they're a bunch of quiet Buddhists,
Lex Fridman (2:42:19.640)
we're still gonna notice that expansionist one, right?
Lex Fridman (2:42:23.040)
And it seems like quite the stretch to assume that,
Lex Fridman (2:42:26.560)
now we know even in our own galaxy
Max Tegmark (2:42:28.120)
that there are probably a billion or more planets
Lex Fridman (2:42:33.120)
that are pretty Earth like.
Lex Fridman (2:42:35.280)
And many of them are formed over a billion years
Lex Fridman (2:42:37.680)
before ours, so had a big head start.
Lex Fridman (2:42:40.640)
So if you actually assume also
Lex Fridman (2:42:43.600)
that life happens kind of automatically
Max Tegmark (2:42:46.120)
on an Earth like planet,
Lex Fridman (2:42:48.440)
I think it's quite the stretch to then go and say,
Max Tegmark (2:42:52.080)
okay, so there are another billion civilizations out there
Lex Fridman (2:42:55.280)
that also have our level of tech
Lex Fridman (2:42:56.840)
and they all decided to become Buddhists
Lex Fridman (2:42:59.280)
and not a single one decided to go Hitler on the galaxy
Lex Fridman (2:43:02.880)
and say, we need to go out and colonize
Lex Fridman (2:43:05.280)
or not a single one decided for more benevolent reasons
Max Tegmark (2:43:08.840)
to go out and get more resources.
Lex Fridman (2:43:11.480)
That seems like a bit of a stretch, frankly.
Lex Fridman (2:43:13.840)
And this leads into the second thing
Lex Fridman (2:43:16.560)
you challenged me that I might be wrong about,
Lex Fridman (2:43:18.560)
how rare or common is life, you know?
Lex Fridman (2:43:22.320)
So Francis Drake, when he wrote down the Drake equation,
Max Tegmark (2:43:25.120)
multiplied together a huge number of factors
Lex Fridman (2:43:27.560)
and then we don't know any of them.
Lex Fridman (2:43:29.320)
So we know even less about what you get
Lex Fridman (2:43:31.480)
when you multiply together the whole product.
Max Tegmark (2:43:35.120)
Since then, a lot of those factors
Lex Fridman (2:43:37.200)
have become much better known.
Max Tegmark (2:43:38.880)
One of his big uncertainties was
Lex Fridman (2:43:40.840)
how common is it that a solar system even has a planet?
Max Tegmark (2:43:44.360)
Well, now we know it very common.
Lex Fridman (2:43:46.280)
Earth like planets, we know we have better.
Max Tegmark (2:43:48.320)
There are a dime a dozen, there are many, many of them,
Lex Fridman (2:43:50.440)
even in our galaxy.
Max Tegmark (2:43:52.080)
At the same time, you know, we have thanks to,
Lex Fridman (2:43:55.000)
I'm a big supporter of the SETI project and its cousins
Lex Fridman (2:43:58.840)
and I think we should keep doing this
Lex Fridman (2:44:00.520)
and we've learned a lot.
Max Tegmark (2:44:02.400)
We've learned that so far,
Lex Fridman (2:44:03.800)
all we have is still unconvincing hints, nothing more, right?
Lex Fridman (2:44:08.040)
And there are certainly many scenarios
Lex Fridman (2:44:10.320)
where it would be dead obvious.
Max Tegmark (2:44:13.080)
If there were a hundred million
Lex Fridman (2:44:15.920)
other human like civilizations in our galaxy,
Max Tegmark (2:44:19.000)
it would not be that hard to notice some of them
Lex Fridman (2:44:21.600)
with today's technology and we haven't, right?
Lex Fridman (2:44:23.440)
So what we can say is, well, okay,
Lex Fridman (2:44:27.720)
we can rule out that there is a human level of civilization
Max Tegmark (2:44:30.560)
on the moon and in fact, the many nearby solar systems
Lex Fridman (2:44:34.120)
where we cannot rule out, of course,
Max Tegmark (2:44:37.600)
that there is something like Earth sitting in a galaxy
Lex Fridman (2:44:41.560)
five billion light years away.
Lex Fridman (2:44:45.120)
But we've ruled out a lot
Lex Fridman (2:44:46.400)
and that's already kind of shocking
Lex Fridman (2:44:48.480)
given that there are all these planets there, you know?
Lex Fridman (2:44:50.320)
So like, where are they?
Lex Fridman (2:44:51.440)
Where are they all?
Lex Fridman (2:44:52.280)
That's the classic Fermi paradox.
Lex Fridman (2:44:54.880)
And so my argument, which might very well be wrong,
Lex Fridman (2:44:59.240)
it's very simple really, it just goes like this.
Max Tegmark (2:45:01.400)
Okay, we have no clue about this.
Lex Fridman (2:45:05.240)
It could be the probability of getting life
Max Tegmark (2:45:07.800)
on a random planet, it could be 10 to the minus one
Lex Fridman (2:45:11.240)
a priori or 10 to the minus five, 10, 10 to the minus 20,
Max Tegmark (2:45:14.680)
10 to the minus 30, 10 to the minus 40.
Lex Fridman (2:45:17.400)
Basically every order of magnitude is about equally likely.
Max Tegmark (2:45:21.400)
When then do the math and ask the question,
Lex Fridman (2:45:24.120)
how close is our nearest neighbor?
Max Tegmark (2:45:27.400)
It's again, equally likely that it's 10 to the 10 meters away,
Lex Fridman (2:45:30.520)
10 to 20 meters away, 10 to the 30 meters away.
Max Tegmark (2:45:33.440)
We have some nerdy ways of talking about this
Lex Fridman (2:45:35.640)
with Bayesian statistics and a uniform log prior,
Lex Fridman (2:45:38.080)
but that's irrelevant.
Lex Fridman (2:45:39.360)
This is the simple basic argument.
Lex Fridman (2:45:42.040)
And now comes the data.
Lex Fridman (2:45:43.320)
So we can say, okay, there are all these orders
Max Tegmark (2:45:46.320)
of magnitude, 10 to the 26 meters away,
Lex Fridman (2:45:49.280)
there's the edge of our observable universe.
Max Tegmark (2:45:51.960)
If it's farther than that, light hasn't even reached us yet.
Lex Fridman (2:45:54.840)
If it's less than 10 to the 16 meters away,
Max Tegmark (2:45:58.040)
well, it's within Earth's,
Lex Fridman (2:46:02.320)
it's no farther away than the sun.
Max Tegmark (2:46:03.800)
We can definitely rule that out.
Lex Fridman (2:46:07.200)
So I think about it like this,
Max Tegmark (2:46:08.520)
a priori before we looked at the telescopes,
Lex Fridman (2:46:11.840)
it could be 10 to the 10 meters, 10 to the 20,
Max Tegmark (2:46:14.320)
10 to the 30, 10 to the 40, 10 to the 50, 10 to blah, blah, blah.
Lex Fridman (2:46:16.520)
Equally likely anywhere here.
Lex Fridman (2:46:18.040)
And now we've ruled out like this chunk.
Lex Fridman (2:46:21.760)
And here is the edge of our observable universe already.
Lex Fridman (2:46:27.880)
So I'm certainly not saying I don't think
Lex Fridman (2:46:30.560)
there's any life elsewhere in space.
Max Tegmark (2:46:32.480)
If space is infinite,
Lex Fridman (2:46:33.680)
then you're basically a hundred percent guaranteed
Max Tegmark (2:46:35.640)
that there is, but the probability that there is life,
Lex Fridman (2:46:41.200)
that the nearest neighbor,
Max Tegmark (2:46:42.280)
it happens to be in this little region
Lex Fridman (2:46:43.760)
between where we would have seen it already
Lex Fridman (2:46:47.120)
and where we will never see it.
Lex Fridman (2:46:48.680)
There's actually significantly less than one, I think.
Lex Fridman (2:46:51.920)
And I think there's a moral lesson from this,
Lex Fridman (2:46:54.280)
which is really important,
Max Tegmark (2:46:55.840)
which is to be good stewards of this planet
Lex Fridman (2:47:00.120)
and this shot we've had.
Max Tegmark (2:47:01.440)
It can be very dangerous to say,
Lex Fridman (2:47:03.640)
oh, it's fine if we nuke our planet or ruin the climate
Max Tegmark (2:47:07.640)
or mess it up with unaligned AI,
Lex Fridman (2:47:10.280)
because I know there is this nice Star Trek fleet out there.
Max Tegmark (2:47:15.160)
They're gonna swoop in and take over where we failed.
Lex Fridman (2:47:18.040)
Just like it wasn't the big deal
Max Tegmark (2:47:19.840)
that the Easter Island losers wiped themselves out.
Lex Fridman (2:47:23.040)
That's a dangerous way of lulling yourself
Max Tegmark (2:47:25.200)
into false sense of security.
Lex Fridman (2:47:27.760)
If it's actually the case that it might be up to us
Lex Fridman (2:47:32.000)
and only us, the whole future of intelligent life
Lex Fridman (2:47:35.000)
in our observable universe,
Max Tegmark (2:47:37.680)
then I think it really puts a lot of responsibility
Lex Fridman (2:47:42.440)
on our shoulders.
Max Tegmark (2:47:43.280)
It's inspiring, it's a little bit terrifying,
Lex Fridman (2:47:45.320)
but it's also inspiring.
Lex Fridman (2:47:46.480)
But it's empowering, I think, most of all,
Lex Fridman (2:47:48.600)
because the biggest problem today is,
Max Tegmark (2:47:50.240)
I see this even when I teach,
Lex Fridman (2:47:53.120)
so many people feel that it doesn't matter what they do
Max Tegmark (2:47:56.360)
or we do, we feel disempowered.
Lex Fridman (2:47:58.760)
Oh, it makes no difference.
Max Tegmark (2:48:02.560)
This is about as far from that as you can come.
Lex Fridman (2:48:05.080)
But we realize that what we do
Max Tegmark (2:48:07.760)
on our little spinning ball here in our lifetime
Lex Fridman (2:48:12.200)
could make the difference for the entire future of life
Max Tegmark (2:48:15.440)
in our universe.
Lex Fridman (2:48:17.080)
How empowering is that?
Max Tegmark (2:48:18.720)
Yeah, survival of consciousness.
Lex Fridman (2:48:20.280)
I mean, a very similar kind of empowering aspect
Max Tegmark (2:48:25.840)
of the Drake equation is,
Lex Fridman (2:48:27.680)
say there is a huge number of intelligent civilizations
Max Tegmark (2:48:31.120)
that spring up everywhere,
Lex Fridman (2:48:32.920)
but because of the Drake equation,
Max Tegmark (2:48:34.760)
which is the lifetime of a civilization,
Lex Fridman (2:48:38.000)
maybe many of them hit a wall.
Lex Fridman (2:48:39.880)
And just like you said, it's clear that that,
Lex Fridman (2:48:43.360)
for us, the great filter,
Max Tegmark (2:48:45.920)
the one possible great filter seems to be coming
Lex Fridman (2:48:49.040)
in the next 100 years.
Lex Fridman (2:48:51.240)
So it's also empowering to say,
Lex Fridman (2:48:53.720)
okay, well, we have a chance to not,
Max Tegmark (2:48:58.720)
I mean, the way great filters work,
Lex Fridman (2:49:00.120)
they just get most of them.
Max Tegmark (2:49:02.080)
Exactly.
Lex Fridman (2:49:02.920)
Nick Bostrom has articulated this really beautifully too.
Max Tegmark (2:49:06.120)
Every time yet another search for life on Mars
Lex Fridman (2:49:09.480)
comes back negative or something,
Max Tegmark (2:49:11.120)
I'm like, yes, yes.
Lex Fridman (2:49:14.760)
Our odds for us surviving is the best.
Lex Fridman (2:49:17.840)
You already made the argument in broad brush there, right?
Lex Fridman (2:49:20.960)
But just to unpack it, right?
Max Tegmark (2:49:22.560)
The point is we already know
Lex Fridman (2:49:26.880)
there is a crap ton of planets out there
Max Tegmark (2:49:28.640)
that are Earth like,
Lex Fridman (2:49:29.640)
and we also know that most of them do not seem
Max Tegmark (2:49:33.160)
to have anything like our kind of life on them.
Lex Fridman (2:49:35.080)
So what went wrong?
Max Tegmark (2:49:37.240)
There's clearly one step along the evolutionary,
Lex Fridman (2:49:39.520)
at least one filter or roadblock
Max Tegmark (2:49:42.360)
in going from no life to spacefaring life.
Lex Fridman (2:49:45.600)
And where is it?
Lex Fridman (2:49:48.160)
Is it in front of us or is it behind us, right?
Lex Fridman (2:49:51.640)
If there's no filter behind us,
Lex Fridman (2:49:54.080)
and we keep finding all sorts of little mice on Mars
Lex Fridman (2:50:00.520)
or whatever, right?
Max Tegmark (2:50:01.880)
That's actually very depressing
Lex Fridman (2:50:03.120)
because that makes it much more likely
Max Tegmark (2:50:04.440)
that the filter is in front of us.
Lex Fridman (2:50:06.280)
And that what actually is going on
Max Tegmark (2:50:08.080)
is like the ultimate dark joke
Lex Fridman (2:50:11.080)
that whenever a civilization
Max Tegmark (2:50:13.800)
invents sufficiently powerful tech,
Lex Fridman (2:50:15.640)
it's just, you just set your clock.
Lex Fridman (2:50:17.240)
And then after a little while it goes poof
Lex Fridman (2:50:19.160)
for one reason or other and wipes itself out.
Max Tegmark (2:50:21.840)
Now wouldn't that be like utterly depressing
Lex Fridman (2:50:24.240)
if we're actually doomed?
Max Tegmark (2:50:26.120)
Whereas if it turns out that there is a really,
Lex Fridman (2:50:29.720)
there is a great filter early on
Max Tegmark (2:50:31.720)
that for whatever reason seems to be really hard
Lex Fridman (2:50:33.960)
to get to the stage of sexually reproducing organisms
Lex Fridman (2:50:39.160)
or even the first ribosome or whatever, right?
Lex Fridman (2:50:43.320)
Or maybe you have lots of planets with dinosaurs and cows,
Lex Fridman (2:50:47.160)
but for some reason they tend to get stuck there
Lex Fridman (2:50:48.880)
and never invent smartphones.
Max Tegmark (2:50:50.840)
All of those are huge boosts for our own odds
Lex Fridman (2:50:55.200)
because been there done that, you know?
Max Tegmark (2:50:58.840)
It doesn't matter how hard or unlikely it was
Lex Fridman (2:51:01.720)
that we got past that roadblock
Max Tegmark (2:51:03.800)
because we already did.
Lex Fridman (2:51:05.120)
And then that makes it likely
Max Tegmark (2:51:07.520)
that the future is in our own hands, we're not doomed.
Lex Fridman (2:51:11.440)
So that's why I think the fact
Max Tegmark (2:51:14.800)
that life is rare in the universe,
Lex Fridman (2:51:18.280)
it's not just something that there is some evidence for,
Lex Fridman (2:51:21.440)
but also something we should actually hope for.
Lex Fridman (2:51:26.680)
So that's the end, the mortality,
Max Tegmark (2:51:29.920)
the death of human civilization
Lex Fridman (2:51:31.520)
that we've been discussing in life,
Max Tegmark (2:51:33.120)
maybe prospering beyond any kind of great filter.
Lex Fridman (2:51:36.680)
Do you think about your own death?
Max Tegmark (2:51:39.440)
Does it make you sad that you may not witness some of the,
Lex Fridman (2:51:45.760)
you know, you lead a research group
Max Tegmark (2:51:47.440)
on working some of the biggest questions
Lex Fridman (2:51:49.040)
in the universe actually,
Lex Fridman (2:51:51.080)
both on the physics and the AI side?
Lex Fridman (2:51:53.720)
Does it make you sad that you may not be able
Max Tegmark (2:51:55.560)
to see some of these exciting things come to fruition
Lex Fridman (2:51:58.840)
that we've been talking about?
Max Tegmark (2:52:00.640)
Of course, of course it sucks, the fact that I'm gonna die.
Lex Fridman (2:52:04.840)
I remember once when I was much younger,
Max Tegmark (2:52:07.200)
my dad made this remark that life is fundamentally tragic.
Lex Fridman (2:52:10.800)
And I'm like, what are you talking about, daddy?
Lex Fridman (2:52:13.080)
And then many years later, I felt,
Lex Fridman (2:52:15.640)
now I feel I totally understand what he means.
Max Tegmark (2:52:17.320)
You know, we grow up, we're little kids
Lex Fridman (2:52:19.040)
and everything is infinite and it's so cool.
Lex Fridman (2:52:21.920)
And then suddenly we find out that actually, you know,
Lex Fridman (2:52:25.800)
you got to serve only,
Max Tegmark (2:52:26.840)
this is the, you're gonna get game over at some point.
Lex Fridman (2:52:30.280)
So of course it's something that's sad.
Lex Fridman (2:52:36.400)
Are you afraid?
Lex Fridman (2:52:42.640)
No, not in the sense that I think anything terrible
Max Tegmark (2:52:46.000)
is gonna happen after I die or anything like that.
Lex Fridman (2:52:48.240)
No, I think it's really gonna be a game over,
Lex Fridman (2:52:50.960)
but it's more that it makes me very acutely aware
Lex Fridman (2:52:56.280)
of what a wonderful gift this is
Max Tegmark (2:52:57.920)
that I get to be alive right now.
Lex Fridman (2:53:00.200)
And is a steady reminder to just live life to the fullest
Lex Fridman (2:53:04.680)
and really enjoy it because it is finite, you know.
Lex Fridman (2:53:08.000)
And I think actually, and we know we all get
Max Tegmark (2:53:11.240)
the regular reminders when someone near and dear to us dies
Lex Fridman (2:53:14.280)
that one day it's gonna be our turn.
Max Tegmark (2:53:19.560)
It adds this kind of focus.
Lex Fridman (2:53:21.480)
I wonder what it would feel like actually
Max Tegmark (2:53:23.680)
to be an immortal being if they might even enjoy
Lex Fridman (2:53:26.960)
some of the wonderful things of life a little bit less
Max Tegmark (2:53:29.440)
just because there isn't that.
Lex Fridman (2:53:33.400)
Finiteness?
Max Tegmark (2:53:34.320)
Yeah.
Lex Fridman (2:53:35.160)
Do you think that could be a feature, not a bug,
Lex Fridman (2:53:38.040)
the fact that we beings are finite?
Lex Fridman (2:53:42.040)
Maybe there's lessons for engineering
Max Tegmark (2:53:44.320)
in artificial intelligence systems as well
Lex Fridman (2:53:46.940)
that are conscious.
Max Tegmark (2:53:48.400)
Like do you think it makes, is it possible
Lex Fridman (2:53:53.920)
that the reason the pistachio ice cream is delicious
Max Tegmark (2:53:56.960)
is the fact that you're going to die one day
Lex Fridman (2:53:59.920)
and you will not have all the pistachio ice cream
Lex Fridman (2:54:03.720)
that you could eat because of that fact?
Lex Fridman (2:54:06.200)
Well, let me say two things.
Max Tegmark (2:54:07.560)
First of all, it's actually quite profound
Lex Fridman (2:54:09.660)
what you're saying.
Max Tegmark (2:54:10.500)
I do think I appreciate the pistachio ice cream
Lex Fridman (2:54:12.300)
a lot more knowing that I will,
Max Tegmark (2:54:14.400)
there's only a finite number of times I get to enjoy that.
Lex Fridman (2:54:17.760)
And I can only remember a finite number of times
Max Tegmark (2:54:19.900)
in the past.
Lex Fridman (2:54:21.720)
And moreover, my life is not so long
Max Tegmark (2:54:25.120)
that it just starts to feel like things are repeating
Lex Fridman (2:54:26.800)
themselves in general.
Max Tegmark (2:54:28.120)
It's so new and fresh.
Lex Fridman (2:54:30.520)
I also think though that death is a little bit overrated
Max Tegmark (2:54:36.400)
in the sense that it comes from a sort of outdated view
Lex Fridman (2:54:42.020)
of physics and what life actually is.
Max Tegmark (2:54:45.640)
Because if you ask, okay, what is it that's gonna die
Lex Fridman (2:54:49.120)
exactly, what am I really?
Max Tegmark (2:54:52.040)
When I say I feel sad about the idea of myself dying,
Lex Fridman (2:54:56.000)
am I really sad that this skin cell here is gonna die?
Max Tegmark (2:54:59.180)
Of course not, because it's gonna die next week anyway
Lex Fridman (2:55:01.600)
and I'll grow a new one, right?
Lex Fridman (2:55:04.020)
And it's not any of my cells that I'm associating really
Lex Fridman (2:55:08.440)
with who I really am.
Max Tegmark (2:55:11.000)
Nor is it any of my atoms or quarks or electrons.
Lex Fridman (2:55:15.640)
In fact, basically all of my atoms get replaced
Lex Fridman (2:55:19.380)
on a regular basis, right?
Lex Fridman (2:55:20.520)
So what is it that's really me
Lex Fridman (2:55:22.880)
from a more modern physics perspective?
Lex Fridman (2:55:24.320)
It's the information in processing me.
Max Tegmark (2:55:28.800)
That's where my memory, that's my memories,
Lex Fridman (2:55:31.520)
that's my values, my dreams, my passion, my love.
Max Tegmark (2:55:40.560)
That's what's really fundamentally me.
Lex Fridman (2:55:43.580)
And frankly, not all of that will die when my body dies.
Max Tegmark (2:55:48.580)
Like Richard Feynman, for example, his body died of cancer,
Lex Fridman (2:55:55.100)
but many of his ideas that he felt made him very him
Max Tegmark (2:55:59.720)
actually live on.
Lex Fridman (2:56:01.400)
This is my own little personal tribute to Richard Feynman.
Max Tegmark (2:56:04.100)
I try to keep a little bit of him alive in myself.
Lex Fridman (2:56:07.500)
I've even quoted him today, right?
Max Tegmark (2:56:09.620)
Yeah, he almost came alive for a brief moment
Lex Fridman (2:56:11.740)
in this conversation, yeah.
Max Tegmark (2:56:13.320)
Yeah, and this honestly gives me some solace.
Lex Fridman (2:56:17.500)
When I work as a teacher, I feel,
Max Tegmark (2:56:20.780)
if I can actually share a bit about myself
Lex Fridman (2:56:25.820)
that my students feel worthy enough to copy and adopt
Max Tegmark (2:56:30.740)
as some part of things that they know
Lex Fridman (2:56:33.140)
or they believe or aspire to,
Lex Fridman (2:56:36.140)
now I live on also a little bit in them, right?
Lex Fridman (2:56:39.540)
And so being a teacher is a little bit
Max Tegmark (2:56:44.540)
of what I, that's something also that contributes
Lex Fridman (2:56:49.740)
to making me a little teeny bit less mortal, right?
Max Tegmark (2:56:53.740)
Because I'm not, at least not all gonna die all at once,
Lex Fridman (2:56:56.740)
right?
Lex Fridman (2:56:57.580)
And I find that a beautiful tribute to people
Lex Fridman (2:56:59.820)
we do not respect.
Max Tegmark (2:57:01.020)
If we can remember them and carry in us
Lex Fridman (2:57:05.740)
the things that we felt was the most awesome about them,
Max Tegmark (2:57:10.260)
right, then they live on.
Lex Fridman (2:57:11.620)
And I'm getting a bit emotional here,
Lex Fridman (2:57:13.580)
but it's a very beautiful idea you bring up there.
Lex Fridman (2:57:16.140)
I think we should stop this old fashioned materialism
Lex Fridman (2:57:19.620)
and just equate who we are with our quirks and electrons.
Lex Fridman (2:57:25.220)
There's no scientific basis for that really.
Lex Fridman (2:57:27.820)
And it's also very uninspiring.
Lex Fridman (2:57:33.180)
Now, if you look a little bit towards the future, right?
Max Tegmark (2:57:36.980)
One thing which really sucks about humans dying is that even
Lex Fridman (2:57:40.740)
though some of their teachings and memories and stories
Lex Fridman (2:57:43.300)
and ethics and so on will be copied by those around them,
Lex Fridman (2:57:47.540)
hopefully, a lot of it can't be copied
Lex Fridman (2:57:50.260)
and just dies with them, with their brain.
Lex Fridman (2:57:51.980)
And that really sucks.
Max Tegmark (2:57:53.140)
That's the fundamental reason why we find it so tragic
Lex Fridman (2:57:56.860)
when someone goes from having all this information there
Lex Fridman (2:57:59.660)
to the more just gone, ruined, right?
Lex Fridman (2:58:03.460)
With more post biological intelligence,
Lex Fridman (2:58:07.460)
that's going to shift a lot, right?
Lex Fridman (2:58:10.940)
The only reason it's so hard to make a backup of your brain
Max Tegmark (2:58:13.980)
in its entirety is exactly
Lex Fridman (2:58:15.380)
because it wasn't built for that, right?
Max Tegmark (2:58:17.580)
If you have a future machine intelligence,
Lex Fridman (2:58:21.540)
there's no reason for why it has to die at all.
Max Tegmark (2:58:24.300)
If you want to copy it, whatever it is,
Lex Fridman (2:58:28.300)
into some other machine intelligence,
Lex Fridman (2:58:30.780)
whatever it is, into some other quark blob, right?
Lex Fridman (2:58:36.660)
You can copy not just some of it, but all of it, right?
Lex Fridman (2:58:39.540)
And so in that sense,
Lex Fridman (2:58:45.020)
you can get immortality because all the information
Max Tegmark (2:58:48.300)
can be copied out of any individual entity.
Lex Fridman (2:58:51.940)
And it's not just mortality that will change
Max Tegmark (2:58:54.220)
if we get to more post biological life.
Lex Fridman (2:58:56.900)
It's also with that, very much the whole individualism
Lex Fridman (2:59:03.180)
we have now, right?
Lex Fridman (2:59:04.020)
The reason that we make such a big difference
Max Tegmark (2:59:05.740)
between me and you is exactly because
Lex Fridman (2:59:09.100)
we're a little bit limited in how much we can copy.
Max Tegmark (2:59:10.940)
Like I would just love to go like this
Lex Fridman (2:59:13.300)
and copy your Russian skills, Russian speaking skills.
Lex Fridman (2:59:17.780)
Wouldn't it be awesome?
Lex Fridman (2:59:18.820)
But I can't, I have to actually work for years
Max Tegmark (2:59:21.980)
if I want to get better on it.
Lex Fridman (2:59:23.900)
But if we were robots.
Max Tegmark (2:59:27.940)
Just copy and paste freely, then that loses completely.
Lex Fridman (2:59:31.820)
It washes away the sense of what immortality is.
Lex Fridman (2:59:35.140)
And also individuality a little bit, right?
Lex Fridman (2:59:37.460)
We would start feeling much more,
Max Tegmark (2:59:40.620)
maybe we would feel much more collaborative with each other
Lex Fridman (2:59:43.540)
if we can just, hey, you know, I'll give you my Russian,
Max Tegmark (2:59:45.620)
you can give me your Russian
Lex Fridman (2:59:46.540)
and I'll give you whatever,
Lex Fridman (2:59:47.940)
and suddenly you can speak Swedish.
Lex Fridman (2:59:50.220)
Maybe that's less a bad trade for you,
Lex Fridman (2:59:52.060)
but whatever else you want from my brain, right?
Lex Fridman (2:59:54.620)
And there've been a lot of sci fi stories
Max Tegmark (2:59:58.060)
about hive minds and so on,
Lex Fridman (2:59:59.540)
where people, where experiences
Max Tegmark (30:01.160)
want to understand that when they build weapon systems,
Lex Fridman (30:03.760)
they don't just go fire at random or malfunction, right?
Lex Fridman (30:07.480)
And there is even a whole operating system
Lex Fridman (30:10.720)
called Cell 3 that's been developed by a DARPA grant,
Max Tegmark (30:12.920)
where you can actually mathematically prove
Lex Fridman (30:16.160)
that this thing can never be hacked.
Max Tegmark (30:18.800)
Wow.
Lex Fridman (30:20.360)
One day, I hope that will be something
Max Tegmark (30:22.280)
you can say about the OS that's running on our laptops too.
Lex Fridman (30:25.120)
As you know, we're not there.
Lex Fridman (30:27.040)
But I think we should be ambitious, frankly.
Lex Fridman (30:30.080)
And if we can use machine learning
Max Tegmark (30:34.120)
to help do the proofs and so on as well,
Lex Fridman (30:36.320)
then it's much easier to verify that a proof is correct
Max Tegmark (30:40.040)
than to come up with a proof in the first place.
Lex Fridman (30:42.960)
That's really the core idea here.
Max Tegmark (30:45.000)
If someone comes on your podcast and says
Lex Fridman (30:47.480)
they proved the Riemann hypothesis
Max Tegmark (30:49.760)
or some sensational new theorem, it's
Lex Fridman (30:55.480)
much easier for someone else, take some smart grad,
Max Tegmark (30:58.640)
math grad students to check, oh, there's an error here
Lex Fridman (31:01.000)
on equation five, or this really checks out,
Max Tegmark (31:04.000)
than it was to discover the proof.
Lex Fridman (31:07.080)
Yeah, although some of those proofs are pretty complicated.
Lex Fridman (31:09.000)
But yes, it's still nevertheless much easier
Lex Fridman (31:11.080)
to verify the proof.
Max Tegmark (31:12.880)
I love the optimism.
Lex Fridman (31:14.680)
We kind of, even with the security of systems,
Max Tegmark (31:17.480)
there's a kind of cynicism that pervades people
Lex Fridman (31:21.760)
who think about this, which is like, oh, it's hopeless.
Max Tegmark (31:24.920)
I mean, in the same sense, exactly like you're saying
Lex Fridman (31:27.080)
when you own networks, oh, it's hopeless to understand
Max Tegmark (31:29.000)
what's happening.
Lex Fridman (31:30.440)
With security, people are just like, well,
Max Tegmark (31:32.560)
it's always going, there's always going to be
Lex Fridman (31:36.240)
attack vectors, like ways to attack the system.
Lex Fridman (31:40.800)
But you're right, we're just very new
Lex Fridman (31:42.200)
with these computational systems.
Max Tegmark (31:44.080)
We're new with these intelligent systems.
Lex Fridman (31:46.400)
And it's not out of the realm of possibly,
Max Tegmark (31:49.560)
just like people that understand the movement
Lex Fridman (31:51.840)
of the stars and the planets and so on.
Max Tegmark (31:54.600)
It's entirely possible that within, hopefully soon,
Lex Fridman (31:58.320)
but it could be within 100 years,
Max Tegmark (32:00.360)
we start to have an obvious laws of gravity
Lex Fridman (32:03.600)
about intelligence and God forbid about consciousness too.
Max Tegmark (32:09.280)
That one is...
Lex Fridman (32:10.960)
Agreed.
Max Tegmark (32:12.320)
I think, of course, if you're selling computers
Lex Fridman (32:15.240)
that get hacked a lot, that's in your interest
Max Tegmark (32:16.720)
as a company that people think it's impossible
Lex Fridman (32:18.640)
to make it safe, but he's going to get the idea
Max Tegmark (32:20.640)
of suing you.
Lex Fridman (32:21.480)
I want to really inject optimism here.
Max Tegmark (32:24.840)
It's absolutely possible to do much better
Lex Fridman (32:29.480)
than we're doing now.
Lex Fridman (32:30.320)
And your laptop does so much stuff.
Lex Fridman (32:34.840)
You don't need the music player to be super safe
Lex Fridman (32:37.960)
in your future self driving car, right?
Lex Fridman (32:42.120)
If someone hacks it and starts playing music
Max Tegmark (32:43.840)
you don't like, the world won't end.
Lex Fridman (32:47.880)
But what you can do is you can break out
Lex Fridman (32:49.560)
and say that your drive computer that controls your safety
Lex Fridman (32:53.080)
must be completely physically decoupled entirely
Max Tegmark (32:55.920)
from the entertainment system.
Lex Fridman (32:57.600)
And it must physically be such that it can't take on
Max Tegmark (33:01.080)
over the air updates while you're driving.
Lex Fridman (33:03.040)
And it can have ultimately some operating system on it
Max Tegmark (33:09.920)
which is symbolically verified and proven
Lex Fridman (33:13.320)
that it's always going to do what it's supposed to do, right?
Max Tegmark (33:17.760)
We can basically have, and companies should take
Lex Fridman (33:19.960)
that attitude too.
Max Tegmark (33:20.800)
They should look at everything they do and say
Lex Fridman (33:22.440)
what are the few systems in our company
Max Tegmark (33:25.840)
that threaten the whole life of the company
Lex Fridman (33:27.400)
if they get hacked and have the highest standards for them.
Lex Fridman (33:31.800)
And then they can save money by going for the el cheapo
Lex Fridman (33:34.560)
poorly understood stuff for the rest.
Max Tegmark (33:36.920)
This is very feasible, I think.
Lex Fridman (33:38.920)
And coming back to the bigger question
Max Tegmark (33:41.720)
that you worried about that there'll be unintentional
Lex Fridman (33:45.000)
failures, I think there are two quite separate risks here.
Lex Fridman (33:47.720)
Right?
Lex Fridman (33:48.560)
We talked a lot about one of them
Max Tegmark (33:49.600)
which is that the goals are noble of the human.
Lex Fridman (33:52.640)
The human says, I want this airplane to not crash
Max Tegmark (33:56.920)
because this is not Muhammad Atta
Lex Fridman (33:58.640)
now flying the airplane, right?
Lex Fridman (34:00.480)
And now there's this technical challenge
Lex Fridman (34:03.240)
of making sure that the autopilot is actually
Max Tegmark (34:05.960)
gonna behave as the pilot wants.
Lex Fridman (34:11.000)
If you set that aside, there's also the separate question.
Lex Fridman (34:13.360)
How do you make sure that the goals of the pilot
Lex Fridman (34:17.440)
are actually aligned with the goals of the passenger?
Lex Fridman (34:19.680)
How do you make sure very much more broadly
Lex Fridman (34:22.480)
that if we can all agree as a species
Max Tegmark (34:24.640)
that we would like things to kind of go well
Lex Fridman (34:26.200)
for humanity as a whole, that the goals are aligned here.
Max Tegmark (34:30.320)
The alignment problem.
Lex Fridman (34:31.560)
And yeah, there's been a lot of progress
Max Tegmark (34:36.000)
in the sense that there's suddenly huge amounts
Lex Fridman (34:39.880)
of research going on on it about it.
Max Tegmark (34:42.040)
I'm very grateful to Elon Musk
Lex Fridman (34:43.400)
for giving us that money five years ago
Lex Fridman (34:44.960)
so we could launch the first research program
Lex Fridman (34:46.680)
on technical AI safety and alignment.
Max Tegmark (34:49.480)
There's a lot of stuff happening.
Lex Fridman (34:51.280)
But I think we need to do more than just make sure
Max Tegmark (34:54.920)
little machines do always what their owners do.
Lex Fridman (34:58.200)
That wouldn't have prevented September 11th
Max Tegmark (35:00.240)
if Muhammad Atta said, okay, autopilot,
Lex Fridman (35:03.040)
please fly into World Trade Center.
Lex Fridman (35:06.720)
And it's like, okay.
Lex Fridman (35:08.960)
That even happened in a different situation.
Lex Fridman (35:11.840)
There was this depressed pilot named Andreas Lubitz, right?
Lex Fridman (35:15.680)
Who told his German wings passenger jet
Max Tegmark (35:17.640)
to fly into the Alps.
Lex Fridman (35:19.040)
He just told the computer to change the altitude
Max Tegmark (35:21.640)
to a hundred meters or something like that.
Lex Fridman (35:23.280)
And you know what the computer said?
Max Tegmark (35:25.360)
Okay.
Lex Fridman (35:26.600)
And it had the freaking topographical map of the Alps
Max Tegmark (35:29.560)
in there, it had GPS, everything.
Lex Fridman (35:31.440)
No one had bothered teaching it
Max Tegmark (35:33.120)
even the basic kindergarten ethics of like,
Lex Fridman (35:35.600)
no, we never want airplanes to fly into mountains
Max Tegmark (35:39.600)
under any circumstances.
Lex Fridman (35:41.040)
And so we have to think beyond just the technical issues
Lex Fridman (35:48.520)
and think about how do we align in general incentives
Lex Fridman (35:51.120)
on this planet for the greater good?
Lex Fridman (35:53.760)
So starting with simple stuff like that,
Lex Fridman (35:55.520)
every airplane that has a computer in it
Max Tegmark (35:58.160)
should be taught whatever kindergarten ethics
Lex Fridman (36:00.840)
that's smart enough to understand.
Max Tegmark (36:02.280)
Like, no, don't fly into fixed objects
Lex Fridman (36:05.040)
if the pilot tells you to do so.
Max Tegmark (36:07.280)
Then go on autopilot mode.
Lex Fridman (36:10.000)
Send an email to the cops and land at the latest airport,
Max Tegmark (36:13.480)
nearest airport, you know.
Lex Fridman (36:14.840)
Any car with a forward facing camera
Max Tegmark (36:18.240)
should just be programmed by the manufacturer
Lex Fridman (36:20.720)
so that it will never accelerate into a human ever.
Max Tegmark (36:24.760)
That would avoid things like the NIS attack
Lex Fridman (36:28.760)
and many horrible terrorist vehicle attacks
Lex Fridman (36:31.120)
where they deliberately did that, right?
Lex Fridman (36:33.720)
This was not some sort of thing,
Max Tegmark (36:35.160)
oh, you know, US and China, different views on,
Lex Fridman (36:38.400)
no, there was not a single car manufacturer
Max Tegmark (36:41.880)
in the world, right, who wanted the cars to do this.
Lex Fridman (36:44.080)
They just hadn't thought to do the alignment.
Lex Fridman (36:45.920)
And if you look at more broadly problems
Lex Fridman (36:48.520)
that happen on this planet,
Max Tegmark (36:51.280)
the vast majority have to do a poor alignment.
Lex Fridman (36:53.840)
I mean, think about, let's go back really big
Max Tegmark (36:57.160)
because I know you're so good at that.
Lex Fridman (36:59.080)
Let's go big, yeah.
Max Tegmark (36:59.920)
Yeah, so long ago in evolution, we had these genes.
Lex Fridman (37:03.840)
And they wanted to make copies of themselves.
Max Tegmark (37:06.400)
That's really all they cared about.
Lex Fridman (37:07.640)
So some genes said, hey, I'm gonna build a brain
Max Tegmark (37:13.160)
on this body I'm in so that I can get better
Lex Fridman (37:15.880)
at making copies of myself.
Lex Fridman (37:17.280)
And then they decided for their benefit
Lex Fridman (37:20.240)
to get copied more, to align your brain's incentives
Max Tegmark (37:23.320)
with their incentives.
Lex Fridman (37:24.560)
So it didn't want you to starve to death.
Lex Fridman (37:29.080)
So it gave you an incentive to eat
Lex Fridman (37:31.520)
and it wanted you to make copies of the genes.
Lex Fridman (37:35.080)
So it gave you incentive to fall in love
Lex Fridman (37:37.680)
and do all sorts of naughty things
Lex Fridman (37:40.960)
to make copies of itself, right?
Lex Fridman (37:44.120)
So that was successful value alignment done on the genes.
Max Tegmark (37:47.720)
They created something more intelligent than themselves,
Lex Fridman (37:50.440)
but they made sure to try to align the values.
Lex Fridman (37:52.920)
But then something went a little bit wrong
Lex Fridman (37:55.800)
against the idea of what the genes wanted
Max Tegmark (37:58.400)
because a lot of humans discovered,
Lex Fridman (38:00.360)
hey, you know, yeah, we really like this business
Max Tegmark (38:03.280)
about sex that the genes have made us enjoy,
Lex Fridman (38:06.640)
but we don't wanna have babies right now.
Lex Fridman (38:09.440)
So we're gonna hack the genes and use birth control.
Lex Fridman (38:13.800)
And I really feel like drinking a Coca Cola right now,
Lex Fridman (38:18.640)
but I don't wanna get a potbelly,
Lex Fridman (38:20.080)
so I'm gonna drink Diet Coke.
Max Tegmark (38:21.960)
We have all these things we've figured out
Lex Fridman (38:24.600)
because we're smarter than the genes,
Lex Fridman (38:26.400)
how we can actually subvert their intentions.
Lex Fridman (38:29.040)
So it's not surprising that we humans now,
Max Tegmark (38:33.440)
when we are in the role of these genes,
Lex Fridman (38:34.800)
creating other nonhuman entities with a lot of power,
Max Tegmark (38:37.640)
have to face the same exact challenge.
Lex Fridman (38:39.400)
How do we make other powerful entities
Lex Fridman (38:41.720)
have incentives that are aligned with ours?
Lex Fridman (38:45.280)
And so they won't hack them.
Lex Fridman (38:47.000)
Corporations, for example, right?
Lex Fridman (38:48.720)
We humans decided to create corporations
Max Tegmark (38:51.280)
because it can benefit us greatly.
Lex Fridman (38:53.440)
Now all of a sudden there's a supermarket.
Max Tegmark (38:55.120)
I can go buy food there.
Lex Fridman (38:56.240)
I don't have to hunt.
Max Tegmark (38:57.200)
Awesome, and then to make sure that this corporation
Lex Fridman (39:02.880)
would do things that were good for us and not bad for us,
Max Tegmark (39:05.960)
we created institutions to keep them in check.
Lex Fridman (39:08.280)
Like if the local supermarket sells poisonous food,
Max Tegmark (39:12.520)
then the owners of the supermarket
Lex Fridman (39:17.920)
have to spend some years reflecting behind bars, right?
Lex Fridman (39:22.160)
So we created incentives to align them.
Lex Fridman (39:25.720)
But of course, just like we were able to see
Max Tegmark (39:27.480)
through this thing and you develop birth control,
Lex Fridman (39:30.640)
if you're a powerful corporation,
Max Tegmark (39:31.840)
you also have an incentive to try to hack the institutions
Lex Fridman (39:35.080)
that are supposed to govern you.
Max Tegmark (39:36.320)
Because you ultimately, as a corporation,
Lex Fridman (39:38.160)
have an incentive to maximize your profit.
Max Tegmark (39:40.920)
Just like you have an incentive
Lex Fridman (39:42.080)
to maximize the enjoyment your brain has,
Max Tegmark (39:44.160)
not for your genes.
Lex Fridman (39:46.000)
So if they can figure out a way of bribing regulators,
Max Tegmark (39:50.480)
then they're gonna do that.
Lex Fridman (39:52.400)
In the US, we kind of caught onto that
Lex Fridman (39:54.440)
and made laws against corruption and bribery.
Lex Fridman (39:58.560)
Then in the late 1800s, Teddy Roosevelt realized that,
Max Tegmark (3:00:02.140)
can be more broadly shared.
Lex Fridman (3:00:05.500)
And I think one, we don't,
Max Tegmark (3:00:08.540)
I don't pretend to know what it would feel like
Lex Fridman (3:00:12.140)
to be a super intelligent machine,
Lex Fridman (3:00:16.940)
but I'm quite confident that however it feels
Lex Fridman (3:00:20.420)
about mortality and individuality
Max Tegmark (3:00:22.420)
will be very, very different from how it is for us.
Lex Fridman (3:00:26.660)
Well, for us, mortality and finiteness
Max Tegmark (3:00:30.500)
seems to be pretty important at this particular moment.
Lex Fridman (3:00:34.100)
And so all good things must come to an end.
Max Tegmark (3:00:37.460)
Just like this conversation, Max.
Lex Fridman (3:00:39.100)
I saw that coming.
Max Tegmark (3:00:40.660)
Sorry, this is the world's worst translation.
Lex Fridman (3:00:44.660)
I could talk to you forever.
Max Tegmark (3:00:45.820)
It's such a huge honor that you've spent time with me.
Lex Fridman (3:00:49.100)
The honor is mine.
Max Tegmark (3:00:50.140)
Thank you so much for getting me essentially
Lex Fridman (3:00:53.380)
to start this podcast by doing the first conversation,
Max Tegmark (3:00:55.980)
making me realize falling in love
Lex Fridman (3:00:58.500)
with conversation in itself.
Lex Fridman (3:01:01.140)
And thank you so much for inspiring
Lex Fridman (3:01:03.220)
so many people in the world with your books,
Max Tegmark (3:01:05.380)
with your research, with your talking,
Lex Fridman (3:01:07.740)
and with the other, like this ripple effect of friends,
Max Tegmark (3:01:12.780)
including Elon and everybody else that you inspire.
Lex Fridman (3:01:15.460)
So thank you so much for talking today.
Max Tegmark (3:01:18.140)
Thank you, I feel so fortunate
Lex Fridman (3:01:21.540)
that you're doing this podcast
Lex Fridman (3:01:23.620)
and getting so many interesting voices out there
Lex Fridman (3:01:27.780)
into the ether and not just the five second sound bites,
Lex Fridman (3:01:30.940)
but so many of the interviews I've watched you do.
Lex Fridman (3:01:33.060)
You really let people go in into depth
Max Tegmark (3:01:36.140)
in a way which we sorely need in this day and age.
Lex Fridman (3:01:38.660)
That I got to be number one, I feel super honored.
Max Tegmark (3:01:41.740)
Yeah, you started it.
Lex Fridman (3:01:43.500)
Thank you so much, Max.
Max Tegmark (3:01:45.620)
Thanks for listening to this conversation
Lex Fridman (3:01:47.180)
with Max Tegmark, and thank you to our sponsors,
Max Tegmark (3:01:50.260)
the Jordan Harbinger Show, For Sigmatic Mushroom Coffee,
Lex Fridman (3:01:54.860)
BetterHelp Online Therapy, and ExpressVPN.
Lex Fridman (3:01:58.940)
So the choice is wisdom, caffeine, sanity, or privacy.
Lex Fridman (3:02:04.420)
Choose wisely, my friends.
Lex Fridman (3:02:05.820)
And if you wish, click the sponsor links below
Lex Fridman (3:02:08.740)
to get a discount and to support this podcast.
Lex Fridman (3:02:11.860)
And now let me leave you with some words from Max Tegmark.
Lex Fridman (3:02:15.100)
If consciousness is the way that information feels
Max Tegmark (3:02:18.860)
when it's processed in certain ways,
Lex Fridman (3:02:21.380)
then it must be substrate independent.
Max Tegmark (3:02:24.220)
It's only the structure of information processing
Lex Fridman (3:02:26.660)
that matters, not the structure of the matter
Max Tegmark (3:02:29.100)
doing the information processing.
Lex Fridman (3:02:31.900)
Thank you for listening, and hope to see you next time.
Max Tegmark (40:03.760)
no, we were still being kind of hacked
Lex Fridman (40:05.360)
because the Massachusetts Railroad companies
Max Tegmark (40:07.280)
had like a bigger budget than the state of Massachusetts
Lex Fridman (40:10.120)
and they were doing a lot of very corrupt stuff.
Lex Fridman (40:13.600)
So he did the whole trust busting thing
Lex Fridman (40:15.480)
to try to align these other nonhuman entities,
Max Tegmark (40:18.440)
the companies, again,
Lex Fridman (40:19.440)
more with the incentives of Americans as a whole.
Max Tegmark (40:23.040)
It's not surprising, though,
Lex Fridman (40:24.080)
that this is a battle you have to keep fighting.
Max Tegmark (40:26.160)
Now we have even larger companies than we ever had before.
Lex Fridman (40:30.560)
And of course, they're gonna try to, again,
Max Tegmark (40:34.320)
subvert the institutions.
Lex Fridman (40:37.800)
Not because, I think people make a mistake
Max Tegmark (40:41.040)
of getting all too,
Lex Fridman (40:44.280)
thinking about things in terms of good and evil.
Max Tegmark (40:46.960)
Like arguing about whether corporations are good or evil,
Lex Fridman (40:50.360)
or whether robots are good or evil.
Max Tegmark (40:53.080)
A robot isn't good or evil, it's a tool.
Lex Fridman (40:57.040)
And you can use it for great things
Max Tegmark (40:58.400)
like robotic surgery or for bad things.
Lex Fridman (41:01.080)
And a corporation also is a tool, of course.
Lex Fridman (41:04.120)
And if you have good incentives to the corporation,
Lex Fridman (41:06.480)
it'll do great things,
Max Tegmark (41:07.520)
like start a hospital or a grocery store.
Lex Fridman (41:10.000)
If you have any bad incentives,
Max Tegmark (41:12.680)
then it's gonna start maybe marketing addictive drugs
Lex Fridman (41:15.800)
to people and you'll have an opioid epidemic, right?
Max Tegmark (41:18.600)
It's all about,
Lex Fridman (41:21.440)
we should not make the mistake of getting into
Max Tegmark (41:23.480)
some sort of fairytale, good, evil thing
Lex Fridman (41:25.640)
about corporations or robots.
Max Tegmark (41:27.920)
We should focus on putting the right incentives in place.
Lex Fridman (41:30.800)
My optimistic vision is that if we can do that,
Max Tegmark (41:34.280)
then we can really get good things.
Lex Fridman (41:35.840)
We're not doing so great with that right now,
Max Tegmark (41:38.000)
either on AI, I think,
Lex Fridman (41:39.240)
or on other intelligent nonhuman entities,
Lex Fridman (41:42.680)
like big companies, right?
Lex Fridman (41:43.920)
We just have a new second generation of AI
Lex Fridman (41:47.440)
and a secretary of defense who's gonna start up now
Lex Fridman (41:51.160)
in the Biden administration,
Max Tegmark (41:53.640)
who was an active member of the board of Raytheon,
Lex Fridman (41:58.120)
for example.
Max Tegmark (41:59.240)
So, I have nothing against Raytheon.
Lex Fridman (42:04.720)
I'm not a pacifist,
Lex Fridman (42:05.680)
but there's an obvious conflict of interest
Lex Fridman (42:08.560)
if someone is in the job where they decide
Max Tegmark (42:12.360)
who they're gonna contract with.
Lex Fridman (42:14.240)
And I think somehow we have,
Max Tegmark (42:16.680)
maybe we need another Teddy Roosevelt to come along again
Lex Fridman (42:19.520)
and say, hey, you know,
Max Tegmark (42:20.640)
we want what's good for all Americans,
Lex Fridman (42:23.480)
and we need to go do some serious realigning again
Max Tegmark (42:26.600)
of the incentives that we're giving to these big companies.
Lex Fridman (42:30.760)
And then we're gonna be better off.
Max Tegmark (42:33.880)
It seems that naturally with human beings,
Lex Fridman (42:35.800)
just like you beautifully described the history
Max Tegmark (42:37.720)
of this whole thing,
Lex Fridman (42:38.880)
of it all started with the genes
Lex Fridman (42:40.760)
and they're probably pretty upset
Lex Fridman (42:42.680)
by all the unintended consequences that happened since.
Lex Fridman (42:45.600)
But it seems that it kind of works out,
Lex Fridman (42:48.680)
like it's in this collective intelligence
Max Tegmark (42:51.120)
that emerges at the different levels.
Lex Fridman (42:53.480)
It seems to find sometimes last minute
Max Tegmark (42:56.920)
a way to realign the values or keep the values aligned.
Lex Fridman (43:00.920)
It's almost, it finds a way,
Max Tegmark (43:03.800)
like different leaders, different humans pop up
Lex Fridman (43:07.560)
all over the place that reset the system.
Lex Fridman (43:10.680)
Do you want, I mean, do you have an explanation why that is?
Lex Fridman (43:15.240)
Or is that just survivor bias?
Lex Fridman (43:17.240)
And also is that different,
Lex Fridman (43:19.600)
somehow fundamentally different than with AI systems
Max Tegmark (43:23.120)
where you're no longer dealing with something
Lex Fridman (43:26.440)
that was a direct, maybe companies are the same,
Lex Fridman (43:30.200)
a direct byproduct of the evolutionary process?
Lex Fridman (43:33.360)
I think there is one thing which has changed.
Max Tegmark (43:36.200)
That's why I'm not all optimistic.
Lex Fridman (43:40.280)
That's why I think there's about a 50% chance
Max Tegmark (43:42.280)
if we take the dumb route with artificial intelligence
Lex Fridman (43:46.120)
that humanity will be extinct in this century.
Max Tegmark (43:51.680)
First, just the big picture.
Lex Fridman (43:53.320)
Yeah, companies need to have the right incentives.
Lex Fridman (43:57.880)
Even governments, right?
Lex Fridman (43:59.000)
We used to have governments,
Max Tegmark (44:02.120)
usually there were just some king,
Lex Fridman (44:04.200)
who was the king because his dad was the king.
Lex Fridman (44:07.160)
And then there were some benefits
Lex Fridman (44:10.600)
of having this powerful kingdom or empire of any sort
Max Tegmark (44:15.280)
because then it could prevent a lot of local squabbles.
Lex Fridman (44:17.960)
So at least everybody in that region
Max Tegmark (44:19.360)
would stop warring against each other.
Lex Fridman (44:20.800)
And their incentives of different cities in the kingdom
Lex Fridman (44:24.200)
became more aligned, right?
Lex Fridman (44:25.160)
That was the whole selling point.
Max Tegmark (44:27.200)
Harare, Noel Harare has a beautiful piece
Lex Fridman (44:31.520)
on how empires were collaboration enablers.
Lex Fridman (44:35.320)
And then we also, Harare says,
Lex Fridman (44:36.760)
invented money for that reason
Lex Fridman (44:38.280)
so we could have better alignment
Lex Fridman (44:40.640)
and we could do trade even with people we didn't know.
Lex Fridman (44:44.160)
So this sort of stuff has been playing out
Lex Fridman (44:45.840)
since time immemorial, right?
Max Tegmark (44:47.880)
What's changed is that it happens on ever larger scales,
Lex Fridman (44:51.520)
right?
Max Tegmark (44:52.360)
The technology keeps getting better
Lex Fridman (44:53.480)
because science gets better.
Lex Fridman (44:54.760)
So now we can communicate over larger distances,
Lex Fridman (44:57.600)
transport things fast over larger distances.
Lex Fridman (44:59.840)
And so the entities get ever bigger,
Lex Fridman (45:02.960)
but our planet is not getting bigger anymore.
Lex Fridman (45:05.480)
So in the past, you could have one experiment
Lex Fridman (45:08.120)
that just totally screwed up like Easter Island,
Max Tegmark (45:11.920)
where they actually managed to have such poor alignment
Lex Fridman (45:15.160)
that when they went extinct, people there,
Lex Fridman (45:17.600)
there was no one else to come back and replace them, right?
Lex Fridman (45:21.520)
If Elon Musk doesn't get us to Mars
Lex Fridman (45:24.000)
and then we go extinct on a global scale,
Lex Fridman (45:27.680)
then we're not coming back.
Max Tegmark (45:28.920)
That's the fundamental difference.
Lex Fridman (45:31.480)
And that's a mistake we don't make for that reason.
Lex Fridman (45:35.800)
In the past, of course, history is full of fiascos, right?
Lex Fridman (45:39.800)
But it was never the whole planet.
Lex Fridman (45:42.160)
And then, okay, now there's this nice uninhabited land here.
Lex Fridman (45:45.960)
Some other people could move in and organize things better.
Max Tegmark (45:49.400)
This is different.
Lex Fridman (45:50.720)
The second thing, which is also different
Lex Fridman (45:52.680)
is that technology gives us so much more empowerment, right?
Lex Fridman (45:58.200)
Both to do good things and also to screw up.
Max Tegmark (46:00.520)
In the stone age, even if you had someone
Lex Fridman (46:02.920)
whose goals were really poorly aligned,
Max Tegmark (46:04.760)
like maybe he was really pissed off
Lex Fridman (46:06.680)
because his stone age girlfriend dumped him
Lex Fridman (46:08.760)
and he just wanted to,
Lex Fridman (46:09.920)
if he wanted to kill as many people as he could,
Lex Fridman (46:12.640)
how many could he really take out with a rock and a stick
Lex Fridman (46:15.160)
before he was overpowered, right?
Lex Fridman (46:17.200)
Just handful, right?
Lex Fridman (46:18.920)
Now, with today's technology,
Max Tegmark (46:23.760)
if we have an accidental nuclear war
Lex Fridman (46:25.640)
between Russia and the US,
Max Tegmark (46:27.880)
which we almost have about a dozen times,
Lex Fridman (46:31.080)
and then we have a nuclear winter,
Max Tegmark (46:32.280)
it could take out seven billion people
Lex Fridman (46:34.760)
or six billion people, we don't know.
Lex Fridman (46:37.280)
So the scale of the damage is bigger that we can do.
Lex Fridman (46:40.440)
And there's obviously no law of physics
Max Tegmark (46:45.520)
that says that technology will never get powerful enough
Lex Fridman (46:48.080)
that we could wipe out our species entirely.
Max Tegmark (46:51.720)
That would just be fantasy to think
Lex Fridman (46:53.640)
that science is somehow doomed
Lex Fridman (46:55.080)
to not get more powerful than that, right?
Lex Fridman (46:57.240)
And it's not at all unfeasible in our lifetime
Max Tegmark (47:00.280)
that someone could design a designer pandemic
Lex Fridman (47:03.120)
which spreads as easily as COVID,
Lex Fridman (47:04.640)
but just basically kills everybody.
Lex Fridman (47:06.880)
We already had smallpox.
Max Tegmark (47:08.480)
It killed one third of everybody who got it.
Lex Fridman (47:13.000)
What do you think of the, here's an intuition,
Max Tegmark (47:15.320)
maybe it's completely naive
Lex Fridman (47:16.840)
and this optimistic intuition I have,
Max Tegmark (47:18.960)
which it seems, and maybe it's a biased experience
Lex Fridman (47:22.880)
that I have, but it seems like the most brilliant people
Max Tegmark (47:25.920)
I've met in my life all are really like
Lex Fridman (47:31.600)
fundamentally good human beings.
Lex Fridman (47:33.680)
And not like naive good, like they really wanna do good
Lex Fridman (47:37.440)
for the world in a way that, well, maybe is aligned
Max Tegmark (47:39.880)
to my sense of what good means.
Lex Fridman (47:41.800)
And so I have a sense that the people
Max Tegmark (47:47.840)
that will be defining the very cutting edge of technology,
Lex Fridman (47:51.000)
there'll be much more of the ones that are doing good
Max Tegmark (47:53.960)
versus the ones that are doing evil.
Lex Fridman (47:55.840)
So the race, I'm optimistic on the,
Max Tegmark (48:00.160)
us always like last minute coming up with a solution.
Lex Fridman (48:03.080)
So if there's an engineered pandemic
Max Tegmark (48:06.480)
that has the capability to destroy
Lex Fridman (48:09.280)
most of the human civilization,
Max Tegmark (48:11.640)
it feels like to me either leading up to that before
Lex Fridman (48:15.880)
or as it's going on, there will be,
Max Tegmark (48:19.240)
we're able to rally the collective genius
Lex Fridman (48:22.520)
of the human species.
Max Tegmark (48:23.800)
I can tell by your smile that you're
Lex Fridman (48:26.160)
at least some percentage doubtful,
Lex Fridman (48:30.080)
but could that be a fundamental law of human nature?
Lex Fridman (48:35.000)
That evolution only creates, like karma is beneficial,
Max Tegmark (48:40.880)
good is beneficial, and therefore we'll be all right.
Lex Fridman (48:44.280)
I hope you're right.
Max Tegmark (48:46.960)
I would really love it if you're right,
Lex Fridman (48:48.720)
if there's some sort of law of nature that says
Max Tegmark (48:51.000)
that we always get lucky in the last second
Lex Fridman (48:53.080)
with karma, but I prefer not playing it so close
Lex Fridman (49:01.160)
and gambling on that.
Lex Fridman (49:03.040)
And I think, in fact, I think it can be dangerous
Max Tegmark (49:06.480)
to have too strong faith in that
Lex Fridman (49:08.120)
because it makes us complacent.
Max Tegmark (49:10.800)
Like if someone tells you, you never have to worry
Lex Fridman (49:12.520)
about your house burning down,
Max Tegmark (49:13.760)
then you're not gonna put in a smoke detector
Lex Fridman (49:15.360)
because why would you need to?
Max Tegmark (49:17.000)
Even if it's sometimes very simple precautions,
Lex Fridman (49:19.040)
we don't take them.
Max Tegmark (49:20.000)
If you're like, oh, the government is gonna take care
Lex Fridman (49:22.360)
of everything for us, I can always trust my politicians.
Max Tegmark (49:24.760)
I can always, we advocate our own responsibility.
Lex Fridman (49:27.520)
I think it's a healthier attitude to say,
Max Tegmark (49:29.080)
yeah, maybe things will work out.
Lex Fridman (49:30.840)
Maybe I'm actually gonna have to myself step up
Lex Fridman (49:33.560)
and take responsibility.
Lex Fridman (49:37.160)
And the stakes are so huge.
Max Tegmark (49:38.360)
I mean, if we do this right, we can develop
Lex Fridman (49:41.840)
all this ever more powerful technology
Lex Fridman (49:43.640)
and cure all diseases and create a future
Lex Fridman (49:46.360)
where humanity is healthy and wealthy
Max Tegmark (49:48.040)
for not just the next election cycle,
Lex Fridman (49:50.080)
but like billions of years throughout our universe.
Max Tegmark (49:52.960)
That's really worth working hard for
Lex Fridman (49:54.760)
and not just sitting and hoping
Max Tegmark (49:58.000)
for some sort of fairytale karma.
Lex Fridman (49:59.520)
Well, I just mean, so you're absolutely right.
Max Tegmark (50:01.600)
From the perspective of the individual,
Lex Fridman (50:03.080)
like for me, the primary thing should be
Max Tegmark (50:05.600)
to take responsibility and to build the solutions
Lex Fridman (50:09.720)
that your skillset allows.
Max Tegmark (50:11.320)
Yeah, which is a lot.
Lex Fridman (50:12.720)
I think we underestimate often very much
Lex Fridman (50:14.560)
how much good we can do.
Lex Fridman (50:16.360)
If you or anyone listening to this
Max Tegmark (50:19.520)
is completely confident that our government
Lex Fridman (50:23.000)
would do a perfect job on handling any future crisis
Max Tegmark (50:25.720)
with engineered pandemics or future AI,
Lex Fridman (50:29.920)
I actually reflect a bit on what actually happened in 2020.
Lex Fridman (50:36.360)
Do you feel that the government by and large
Lex Fridman (50:39.680)
around the world has handled this flawlessly?
Max Tegmark (50:42.680)
That's a really sad and disappointing reality
Lex Fridman (50:45.160)
that hopefully is a wake up call for everybody.
Max Tegmark (50:48.720)
For the scientists, for the engineers,
Lex Fridman (50:52.280)
for the researchers in AI especially,
Max Tegmark (50:54.240)
it was disappointing to see how inefficient we were
Lex Fridman (51:01.000)
at collecting the right amount of data
Max Tegmark (51:04.120)
in a privacy preserving way and spreading that data
Lex Fridman (51:07.080)
and utilizing that data to make decisions,
Max Tegmark (51:09.200)
all that kind of stuff.
Lex Fridman (51:10.440)
Yeah, I think when something bad happens to me,
Max Tegmark (51:13.360)
I made myself a promise many years ago
Lex Fridman (51:17.280)
that I would not be a whiner.
Lex Fridman (51:21.760)
So when something bad happens to me,
Lex Fridman (51:23.680)
of course it's a process of disappointment,
Lex Fridman (51:27.280)
but then I try to focus on what did I learn from this
Lex Fridman (51:30.520)
that can make me a better person in the future.
Lex Fridman (51:32.600)
And there's usually something to be learned when I fail.
Lex Fridman (51:35.720)
And I think we should all ask ourselves,
Lex Fridman (51:38.200)
what can we learn from the pandemic
Lex Fridman (51:41.480)
about how we can do better in the future?
Lex Fridman (51:43.400)
And you mentioned there a really good lesson.
Lex Fridman (51:46.360)
We were not as resilient as we thought we were
Lex Fridman (51:50.480)
and we were not as prepared maybe as we wish we were.
Lex Fridman (51:53.960)
You can even see very stark contrast around the planet.
Max Tegmark (51:57.280)
South Korea, they have over 50 million people.
Lex Fridman (52:01.760)
Do you know how many deaths they have from COVID
Lex Fridman (52:03.520)
last time I checked?
Lex Fridman (52:05.600)
No.
Max Tegmark (52:06.440)
It's about 500.
Lex Fridman (52:08.880)
Why is that?
Max Tegmark (52:10.280)
Well, the short answer is that they had prepared.
Lex Fridman (52:16.760)
They were incredibly quick,
Max Tegmark (52:19.200)
incredibly quick to get on it
Lex Fridman (52:21.520)
with very rapid testing and contact tracing and so on,
Max Tegmark (52:25.520)
which is why they never had more cases
Lex Fridman (52:28.080)
than they could contract trace effectively, right?
Max Tegmark (52:30.040)
They never even had to have the kind of big lockdowns
Lex Fridman (52:32.040)
we had in the West.
Lex Fridman (52:33.720)
But the deeper answer to,
Lex Fridman (52:36.560)
it's not just the Koreans are just somehow better people.
Max Tegmark (52:39.080)
The reason I think they were better prepared
Lex Fridman (52:40.800)
was because they had already had a pretty bad hit
Max Tegmark (52:45.320)
from the SARS pandemic,
Lex Fridman (52:47.560)
or which never became a pandemic,
Max Tegmark (52:49.920)
something like 17 years ago, I think.
Lex Fridman (52:52.120)
So it was kind of fresh memory
Max Tegmark (52:53.400)
that we need to be prepared for pandemics.
Lex Fridman (52:56.000)
So they were, right?
Lex Fridman (52:59.080)
So maybe this is a lesson here
Lex Fridman (53:01.240)
for all of us to draw from COVID
Max Tegmark (53:03.280)
that rather than just wait for the next pandemic
Lex Fridman (53:06.360)
or the next problem with AI getting out of control
Max Tegmark (53:09.840)
or anything else,
Lex Fridman (53:11.320)
maybe we should just actually set aside
Max Tegmark (53:14.720)
a tiny fraction of our GDP
Lex Fridman (53:17.680)
to have people very systematically
Max Tegmark (53:19.320)
do some horizon scanning and say,
Lex Fridman (53:20.680)
okay, what are the things that could go wrong?
Lex Fridman (53:23.320)
And let's duke it out and see
Lex Fridman (53:24.600)
which are the more likely ones
Lex Fridman (53:25.800)
and which are the ones that are actually actionable
Lex Fridman (53:28.760)
and then be prepared.
Lex Fridman (53:29.800)
So one of the observations as one little ant slash human
Lex Fridman (53:36.560)
that I am of disappointment
Max Tegmark (53:38.560)
is the political division over information
Lex Fridman (53:44.040)
that has been observed, that I observed this year,
Max Tegmark (53:47.440)
that it seemed the discussion was less about
Lex Fridman (53:54.040)
sort of what happened and understanding
Lex Fridman (53:57.600)
what happened deeply and more about
Lex Fridman (54:00.680)
there's different truths out there.
Lex Fridman (54:04.080)
And it's like an argument,
Lex Fridman (54:05.400)
my truth is better than your truth.
Lex Fridman (54:07.640)
And it's like red versus blue or different.
Lex Fridman (54:10.840)
It was like this ridiculous discourse
Max Tegmark (54:13.280)
that doesn't seem to get at any kind of notion of the truth.
Lex Fridman (54:16.520)
It's not like some kind of scientific process.
Max Tegmark (54:19.000)
Even science got politicized in ways
Lex Fridman (54:21.000)
that's very heartbreaking to me.
Max Tegmark (54:24.360)
You have an exciting project on the AI front
Lex Fridman (54:28.680)
of trying to rethink one of the,
Max Tegmark (54:32.560)
you mentioned corporations.
Lex Fridman (54:34.240)
There's one of the other collective intelligence systems
Max Tegmark (54:37.360)
that have emerged through all of this is social networks.
Lex Fridman (54:40.480)
And just the spread, the internet is the spread
Max Tegmark (54:43.600)
of information on the internet,
Lex Fridman (54:46.400)
our ability to share that information.
Max Tegmark (54:48.320)
There's all different kinds of news sources and so on.
Lex Fridman (54:50.640)
And so you said like that's from first principles,
Max Tegmark (54:53.200)
let's rethink how we think about the news,
Lex Fridman (54:57.320)
how we think about information.
Lex Fridman (54:59.080)
Can you talk about this amazing effort
Lex Fridman (55:02.480)
that you're undertaking?
Max Tegmark (55:03.640)
Oh, I'd love to.
Lex Fridman (55:04.560)
This has been my big COVID project
Lex Fridman (55:06.400)
and nights and weekends on ever since the lockdown.
Lex Fridman (55:11.920)
To segue into this actually,
Max Tegmark (55:13.080)
let me come back to what you said earlier
Lex Fridman (55:14.520)
that you had this hope that in your experience,
Max Tegmark (55:17.040)
people who you felt were very talented
Lex Fridman (55:18.800)
were often idealistic and wanted to do good.
Max Tegmark (55:21.240)
Frankly, I feel the same about all people by and large,
Lex Fridman (55:25.160)
there are always exceptions,
Lex Fridman (55:26.120)
but I think the vast majority of everybody,
Lex Fridman (55:28.480)
regardless of education and whatnot,
Lex Fridman (55:30.320)
really are fundamentally good, right?
Lex Fridman (55:33.280)
So how can it be that people still do so much nasty stuff?
Max Tegmark (55:37.920)
I think it has everything to do with this,
Lex Fridman (55:40.040)
with the information that we're given.
Max Tegmark (55:41.920)
Yes.
Lex Fridman (55:42.760)
If you go into Sweden 500 years ago
Lex Fridman (55:46.240)
and you start telling all the farmers
Lex Fridman (55:47.360)
that those Danes in Denmark,
Max Tegmark (55:49.160)
they're so terrible people, and we have to invade them
Lex Fridman (55:52.840)
because they've done all these terrible things
Max Tegmark (55:55.320)
that you can't fact check yourself.
Lex Fridman (55:56.840)
A lot of people, Swedes did that, right?
Lex Fridman (55:59.720)
And we're seeing so much of this today in the world,
Lex Fridman (56:06.680)
both geopolitically, where we are told that China is bad
Lex Fridman (56:11.760)
and Russia is bad and Venezuela is bad,
Lex Fridman (56:13.960)
and people in those countries are often told
Max Tegmark (56:16.000)
that we are bad.
Lex Fridman (56:17.320)
And we also see it at a micro level where people are told
Max Tegmark (56:21.840)
that, oh, those who voted for the other party are bad people.
Lex Fridman (56:24.640)
It's not just an intellectual disagreement,
Lex Fridman (56:26.480)
but they're bad people and we're getting ever more divided.
Lex Fridman (56:32.880)
So how do you reconcile this with this intrinsic goodness
Lex Fridman (56:39.000)
in people?
Lex Fridman (56:39.840)
I think it's pretty obvious that it has, again,
Lex Fridman (56:41.640)
to do with the information that we're fed and given, right?
Lex Fridman (56:46.280)
We evolved to live in small groups
Lex Fridman (56:49.800)
where you might know 30 people in total, right?
Lex Fridman (56:52.080)
So you then had a system that was quite good
Max Tegmark (56:55.440)
for assessing who you could trust and who you could not.
Lex Fridman (56:57.760)
And if someone told you that Joe there is a jerk,
Lex Fridman (57:02.840)
but you had interacted with him yourself
Lex Fridman (57:05.000)
and seen him in action,
Lex Fridman (57:06.400)
and you would quickly realize maybe
Lex Fridman (57:08.320)
that that's actually not quite accurate, right?
Lex Fridman (57:11.680)
But now that the most people on the planet
Lex Fridman (57:13.520)
are people we've never met,
Max Tegmark (57:15.280)
it's very important that we have a way
Lex Fridman (57:17.200)
of trusting the information we're given.
Lex Fridman (57:19.400)
And so, okay, so where does the news project come in?
Lex Fridman (57:23.160)
Well, throughout history, you can go read Machiavelli,
Max Tegmark (57:26.560)
from the 1400s, and you'll see how already then
Lex Fridman (57:28.680)
they were busy manipulating people
Max Tegmark (57:30.040)
with propaganda and stuff.
Lex Fridman (57:31.640)
Propaganda is not new at all.
Lex Fridman (57:35.720)
And the incentives to manipulate people
Lex Fridman (57:37.720)
is just not new at all.
Lex Fridman (57:40.040)
What is it that's new?
Lex Fridman (57:41.240)
What's new is machine learning meets propaganda.
Max Tegmark (57:44.680)
That's what's new.
Lex Fridman (57:45.760)
That's why this has gotten so much worse.
Max Tegmark (57:47.880)
Some people like to blame certain individuals,
Lex Fridman (57:50.680)
like in my liberal university bubble,
Max Tegmark (57:53.120)
many people blame Donald Trump and say it was his fault.
Lex Fridman (57:56.920)
I see it differently.
Max Tegmark (57:59.120)
I think Donald Trump just had this extreme skill
Lex Fridman (58:03.840)
at playing this game in the machine learning algorithm age.
Max Tegmark (58:07.560)
A game he couldn't have played 10 years ago.
Lex Fridman (58:09.920)
So what's changed?
Max Tegmark (58:10.920)
What's changed is, well, Facebook and Google
Lex Fridman (58:13.200)
and other companies, and I'm not badmouthing them,
Max Tegmark (58:16.640)
I have a lot of friends who work for these companies,
Lex Fridman (58:18.800)
good people, they deployed machine learning algorithms
Max Tegmark (58:22.720)
just to increase their profit a little bit,
Lex Fridman (58:24.280)
to just maximize the time people spent watching ads.
Lex Fridman (58:28.520)
And they had totally underestimated
Lex Fridman (58:30.520)
how effective they were gonna be.
Max Tegmark (58:32.360)
This was, again, the black box, non intelligible intelligence.
Lex Fridman (58:37.560)
They just noticed, oh, we're getting more ad revenue.
Max Tegmark (58:39.360)
Great.
Lex Fridman (58:40.200)
It took a long time until they even realized why and how
Lex Fridman (58:42.400)
and how damaging this was for society.
Lex Fridman (58:45.760)
Because of course, what the machine learning figured out
Max Tegmark (58:47.960)
was that the by far most effective way of gluing you
Lex Fridman (58:52.080)
to your little rectangle was to show you things
Max Tegmark (58:55.040)
that triggered strong emotions, anger, et cetera, resentment,
Lex Fridman (58:59.800)
and if it was true or not, it didn't really matter.
Max Tegmark (59:04.720)
It was also easier to find stories that weren't true.
Lex Fridman (59:07.520)
If you weren't limited, that's just the limitation,
Max Tegmark (59:09.320)
is to show people.
Lex Fridman (59:10.600)
That's a very limiting fact.
Lex Fridman (59:12.360)
And before long, we got these amazing filter bubbles
Lex Fridman (59:16.960)
on a scale we had never seen before.
Max Tegmark (59:18.960)
A couple of days to the fact that also the online news media
Lex Fridman (59:24.600)
were so effective that they killed a lot of people
Max Tegmark (59:27.560)
that were so effective that they killed a lot of print
Lex Fridman (59:30.200)
journalism.
Max Tegmark (59:30.800)
There's less than half as many journalists
Lex Fridman (59:34.120)
now in America, I believe, as there was a generation ago.
Max Tegmark (59:39.640)
You just couldn't compete with the online advertising.
Lex Fridman (59:42.800)
So all of a sudden, most people are not
Max Tegmark (59:47.240)
getting even reading newspapers.
Lex Fridman (59:48.680)
They get their news from social media.
Lex Fridman (59:51.320)
And most people only get news in their little bubble.
Lex Fridman (59:55.000)
So along comes now some people like Donald Trump,
Max Tegmark (59:58.400)
who figured out, among the first successful politicians,
🔗 相关节目