Erik Brynjolfsson: Economics of AI, Social Networks, and Technology
技术与编程AI 与机器学习音乐与艺术心理与人性政治与社会
📋 章节目录
暂无章节信息
🔑 关键词
donbettermachinetruthdoingdatatechnologyimportanttaxeconomytwitterexponentialhumantechnologiesdoesnwaysothersproductivitygettingsaid
💬 精彩语录
暂无语录
🎙️ 完整对话（2411 条）
Lex Fridman (00:00.000)
The following is a conversation with Erik Brynjolfsson.
以下是与埃里克·布林约尔松 (Erik Brynjolfsson) 的对话。
Lex Fridman (00:03.200)
He's an economics professor at Stanford
他是斯坦福大学的经济学教授
Lex Fridman (00:05.800)
and the director of Stanford's Digital Economy Lab.
斯坦福大学数字经济实验室主任。
Lex Fridman (00:09.360)
Previously, he was a long, long time professor at MIT
此前，他是麻省理工学院的长期教授
Lex Fridman (00:13.440)
where he did groundbreaking work
他在那里做了开创性的工作
Erik Brynjolfsson (00:15.160)
on the economics of information.
关于信息经济学。
Lex Fridman (00:17.720)
He's the author of many books,
他是许多书的作者
Erik Brynjolfsson (00:19.760)
including The Second Machine Age
包括第二次机器时代
Lex Fridman (00:21.960)
and Machine Platform Crowd,
和机器平台人群，
Erik Brynjolfsson (00:24.520)
coauthored with Andrew McAfee.
与安德鲁·麦卡菲合着。
Lex Fridman (00:27.520)
Quick mention of each sponsor,
快速提及每个赞助商，
Erik Brynjolfsson (00:29.080)
followed by some thoughts related to the episode.
接下来是一些与这一集相关的想法。
Lex Fridman (00:31.560)
Ventura Watches, the maker of classy,
Ventura 手表，优雅的制造商，
Erik Brynjolfsson (00:34.040)
well performing watches.
性能良好的手表。
Lex Fridman (00:35.960)
Four Sigmatic, the maker of delicious mushroom coffee.
四西格玛蒂克 (Four Sigmatic)，美味蘑菇咖啡的制造商。
Erik Brynjolfsson (00:39.720)
ExpressVPN, the VPN I've used for many years
ExpressVPN，我用了很多年的VPN
Lex Fridman (00:42.760)
to protect my privacy on the internet.
保护我在互联网上的隐私。
Lex Fridman (00:44.920)
And CashApp, the app I use to send money to friends.
还有 CashApp，我用来向朋友汇款的应用程序。
Lex Fridman (00:48.920)
Please check out these sponsors in the description
请在说明中查看这些赞助商
Erik Brynjolfsson (00:50.840)
to get a discount and to support this podcast.
获得折扣并支持此播客。
Lex Fridman (00:54.480)
As a side note, let me say that the impact
Erik Brynjolfsson (00:56.800)
of artificial intelligence and automation
Lex Fridman (00:59.120)
on our economy and our world
Erik Brynjolfsson (01:01.640)
is something worth thinking deeply about.
Lex Fridman (01:04.360)
Like with many topics that are linked
Erik Brynjolfsson (01:06.280)
to predicting the future evolution of technology,
Lex Fridman (01:09.040)
it is often too easy to fall into one of two camps.
Erik Brynjolfsson (01:12.560)
The fear mongering camp
Lex Fridman (01:14.680)
or the technological utopianism camp.
Erik Brynjolfsson (01:18.160)
As always, the future will land us somewhere in between.
Lex Fridman (01:21.480)
I prefer to wear two hats in these discussions
Lex Fridman (01:24.240)
and alternate between them often.
Lex Fridman (01:26.400)
The hat of a pragmatic engineer
Lex Fridman (01:29.360)
and the hat of a futurist.
Lex Fridman (01:31.760)
This is probably a good time to mention Andrew Yang,
Erik Brynjolfsson (01:34.920)
the presidential candidate who has been
Lex Fridman (01:37.920)
one of the high profile thinkers on this topic.
Lex Fridman (01:41.040)
And I'm sure I will speak with him
Lex Fridman (01:42.680)
on this podcast eventually.
Erik Brynjolfsson (01:44.600)
A conversation with Andrew has been on the table many times.
Lex Fridman (01:48.520)
Our schedules just haven't aligned,
Erik Brynjolfsson (01:50.440)
especially because I have a strongly held to preference
Lex Fridman (01:54.280)
for long form, two, three, four hours or more,
Lex Fridman (01:58.040)
and in person.
Lex Fridman (02:00.000)
I work hard to not compromise on this.
Erik Brynjolfsson (02:02.880)
Trust me, it's not easy.
Lex Fridman (02:04.800)
Even more so in the times of COVID,
Erik Brynjolfsson (02:07.080)
which requires getting tested nonstop,
Lex Fridman (02:09.640)
staying isolated and doing a lot of costly
Lex Fridman (02:12.440)
and uncomfortable things that minimize risk for the guest.
Lex Fridman (02:15.760)
The reason I do this is because to me,
Erik Brynjolfsson (02:17.760)
something is lost in remote conversation.
Lex Fridman (02:20.720)
That something, that magic,
Erik Brynjolfsson (02:23.360)
I think is worth the effort,
Lex Fridman (02:25.120)
even if it ultimately leads to a failed conversation.
Erik Brynjolfsson (02:29.360)
This is how I approach life,
Lex Fridman (02:31.280)
treasuring the possibility of a rare moment of magic.
Erik Brynjolfsson (02:35.840)
I'm willing to go to the ends of the world
Lex Fridman (02:38.240)
for just such a moment.
Erik Brynjolfsson (02:40.680)
If you enjoy this thing, subscribe on YouTube,
Lex Fridman (02:43.080)
review it with five stars on Apple Podcast,
Erik Brynjolfsson (02:45.320)
follow on Spotify, support on Patreon,
Lex Fridman (02:47.960)
connect with me on Twitter at Lex Friedman.
Lex Fridman (02:51.080)
And now here's my conversation with Erik Brynjolfsson.
Lex Fridman (02:56.080)
You posted a quote on Twitter by Albert Bartlett
Erik Brynjolfsson (02:59.800)
saying that the greatest shortcoming of the human race
Lex Fridman (03:03.240)
is our inability to understand the exponential function.
Lex Fridman (03:07.760)
Why would you say the exponential growth
Lex Fridman (03:09.680)
is important to understand?
Erik Brynjolfsson (03:12.120)
Yeah, that quote, I remember posting that.
Lex Fridman (03:15.040)
It's actually a reprise of something Andy McAfee and I said
Erik Brynjolfsson (03:17.720)
in the second machine age,
Lex Fridman (03:19.240)
but I posted it in early March
Erik Brynjolfsson (03:21.240)
when COVID was really just beginning to take off
Lex Fridman (03:23.680)
and I was really scared.
Erik Brynjolfsson (03:25.600)
There were actually only a couple dozen cases,
Lex Fridman (03:28.080)
maybe less at that time,
Lex Fridman (03:29.800)
but they were doubling every like two or three days
Lex Fridman (03:32.040)
and I could see, oh my God, this is gonna be a catastrophe
Lex Fridman (03:35.280)
and it's gonna happen soon,
Lex Fridman (03:36.840)
but nobody was taking it very seriously
Erik Brynjolfsson (03:38.760)
or not a lot of people were taking it very seriously.
Lex Fridman (03:40.760)
In fact, I remember I did my last in person conference
Erik Brynjolfsson (03:45.000)
that week, I was flying back from Las Vegas
Lex Fridman (03:47.680)
and I was the only person on the plane wearing a mask
Lex Fridman (03:50.640)
and the flight attendant came over to me.
Lex Fridman (03:52.160)
She looked very concerned.
Erik Brynjolfsson (03:53.080)
She kind of put her hands on my shoulder.
Lex Fridman (03:54.240)
She was touching me all over, which I wasn't thrilled about
Lex Fridman (03:56.440)
and she goes, do you have some kind of anxiety disorder?
Lex Fridman (03:59.280)
Are you okay?
Lex Fridman (04:00.360)
And I was like, no, it's because of COVID.
Lex Fridman (04:02.760)
This is early March.
Erik Brynjolfsson (04:03.920)
Early March, but I was worried
Lex Fridman (04:06.720)
because I knew I could see or I suspected, I guess,
Erik Brynjolfsson (04:10.680)
that that doubling would continue and it did
Lex Fridman (04:13.200)
and pretty soon we had thousands of times more cases.
Erik Brynjolfsson (04:17.000)
Most of the time when I use that quote,
Lex Fridman (04:18.480)
I try to, it's motivated by more optimistic things
Erik Brynjolfsson (04:21.120)
like Moore's law and the wonders
Lex Fridman (04:23.040)
of having more computer power,
Lex Fridman (04:25.400)
but in either case, it can be very counterintuitive.
Lex Fridman (04:28.480)
I mean, if you walk for 10 minutes,
Erik Brynjolfsson (04:31.440)
you get about 10 times as far away
Lex Fridman (04:32.920)
as if you walk for one minute.
Erik Brynjolfsson (04:34.720)
That's the way our physical world works.
Lex Fridman (04:35.920)
That's the way our brains are wired,
Lex Fridman (04:38.040)
but if something doubles for 10 times as long,
Lex Fridman (04:41.480)
you don't get 10 times as much.
Erik Brynjolfsson (04:43.080)
You get a thousand times as much
Lex Fridman (04:45.240)
and after 20, it's a billion.
Erik Brynjolfsson (04:47.080)
After 30, it's a, no, sorry, after 20, it's a million.
Lex Fridman (04:50.720)
After 30, it's a billion.
Lex Fridman (04:53.400)
And pretty soon after that,
Lex Fridman (04:54.480)
it just gets to these numbers that you can barely grasp.
Erik Brynjolfsson (04:57.600)
Our world is becoming more and more exponential,
Lex Fridman (05:00.640)
mainly because of digital technologies.
Lex Fridman (05:03.440)
So more and more often our intuitions are out of whack
Lex Fridman (05:06.200)
and that can be good in the case of things creating wonders,
Lex Fridman (05:10.760)
but it can be dangerous in the case of viruses
Lex Fridman (05:13.520)
and other things.
Lex Fridman (05:14.440)
Do you think it generally applies,
Lex Fridman (05:16.280)
like is there spaces where it does apply
Lex Fridman (05:18.120)
and where it doesn't?
Lex Fridman (05:19.320)
How are we supposed to build an intuition
Erik Brynjolfsson (05:21.600)
about in which aspects of our society
Lex Fridman (05:25.160)
does exponential growth apply?
Erik Brynjolfsson (05:27.280)
Well, you can learn the math,
Lex Fridman (05:29.480)
but the truth is our brains, I think,
Erik Brynjolfsson (05:32.000)
tend to learn more from experiences.
Lex Fridman (05:35.280)
So we just start seeing it more and more often.
Lex Fridman (05:37.440)
So hanging around Silicon Valley,
Lex Fridman (05:39.320)
hanging around AI and computer researchers,
Erik Brynjolfsson (05:41.560)
I see this kind of exponential growth a lot more frequently
Lex Fridman (05:44.800)
and I'm getting used to it, but I still make mistakes.
Erik Brynjolfsson (05:46.760)
I still underestimate some of the progress
Lex Fridman (05:48.680)
in just talking to someone about GPT3
Lex Fridman (05:50.920)
and how rapidly natural language has improved.
Lex Fridman (05:54.040)
But I think that as the world becomes more exponential,
Erik Brynjolfsson (05:58.280)
we'll all start experiencing it more frequently.
Lex Fridman (06:01.760)
The danger is that we may make some mistakes in the meantime
Erik Brynjolfsson (06:05.360)
using our old kind of caveman intuitions
Lex Fridman (06:07.680)
about how the world works.
Erik Brynjolfsson (06:09.280)
Well, the weird thing is it always kind of looks linear
Lex Fridman (06:11.760)
in the moment.
Erik Brynjolfsson (06:12.640)
Like it's hard to feel,
Lex Fridman (06:16.920)
it's hard to like introspect
Lex Fridman (06:19.920)
and really acknowledge how much has changed
Lex Fridman (06:22.960)
in just a couple of years or five years or 10 years
Erik Brynjolfsson (06:26.320)
with the internet.
Lex Fridman (06:27.200)
If we just look at advancements of AI
Erik Brynjolfsson (06:29.600)
or even just social media,
Lex Fridman (06:31.680)
all the various technologies
Erik Brynjolfsson (06:33.680)
that go into the digital umbrella,
Lex Fridman (06:36.240)
it feels pretty calm and normal and gradual.
Erik Brynjolfsson (06:39.440)
Well, a lot of stuff,
Lex Fridman (06:40.760)
I think there are parts of the world,
Erik Brynjolfsson (06:42.000)
most of the world that is not exponential.
Lex Fridman (06:45.600)
The way humans learn,
Erik Brynjolfsson (06:47.000)
the way organizations change,
Lex Fridman (06:49.200)
the way our whole institutions adapt and evolve,
Erik Brynjolfsson (06:52.000)
those don't improve at exponential paces.
Lex Fridman (06:54.520)
And that leads to a mismatch oftentimes
Erik Brynjolfsson (06:56.320)
between these exponentially improving technologies
Lex Fridman (06:58.920)
or let's say changing technologies
Erik Brynjolfsson (07:00.440)
because some of them are exponentially more dangerous
Lex Fridman (07:03.040)
and our intuitions and our human skills
Lex Fridman (07:06.720)
and our institutions that just don't change very fast at all.
Lex Fridman (07:11.720)
And that mismatch I think is at the root
Erik Brynjolfsson (07:13.760)
of a lot of the problems in our society,
Lex Fridman (07:15.520)
the growing inequality
Lex Fridman (07:18.160)
and other dysfunctions in our political
Lex Fridman (07:22.560)
and economic systems.
Lex Fridman (07:24.240)
So one guy that talks about exponential functions
Lex Fridman (07:28.240)
a lot is Elon Musk.
Erik Brynjolfsson (07:29.440)
He seems to internalize this kind of way
Lex Fridman (07:32.040)
of exponential thinking.
Erik Brynjolfsson (07:34.680)
He calls it first principles thinking,
Lex Fridman (07:36.320)
sort of the kind of going to the basics,
Erik Brynjolfsson (07:39.480)
asking the question,
Lex Fridman (07:41.560)
like what were the assumptions of the past?
Lex Fridman (07:43.800)
How can we throw them out the window?
Lex Fridman (07:46.720)
How can we do this 10X much more efficiently
Lex Fridman (07:49.160)
and constantly practicing that process?
Lex Fridman (07:51.360)
And also using that kind of thinking
Erik Brynjolfsson (07:54.200)
to estimate sort of when, you know, create deadlines
Lex Fridman (08:01.200)
and estimate when you'll be able to deliver
Erik Brynjolfsson (08:04.040)
on some of these technologies.
Lex Fridman (08:06.520)
Now, it often gets him in trouble
Erik Brynjolfsson (08:09.560)
because he overestimates,
Lex Fridman (08:12.520)
like he doesn't meet the initial estimates of the deadlines,
Lex Fridman (08:17.360)
but he seems to deliver late but deliver.
Lex Fridman (08:22.360)
And which is kind of interesting.
Lex Fridman (08:25.080)
Like, what are your thoughts about this whole thing?
Lex Fridman (08:26.920)
I think we can all learn from Elon.
Erik Brynjolfsson (08:28.840)
I think going to first principles,
Lex Fridman (08:30.120)
I talked about two ways of getting more of a grip
Erik Brynjolfsson (08:32.840)
on the exponential function.
Lex Fridman (08:34.560)
And one of them just comes from first principles.
Erik Brynjolfsson (08:36.280)
You know, if you understand the math of it,
Lex Fridman (08:37.800)
you can see what's gonna happen.
Lex Fridman (08:39.040)
And even if it seems counterintuitive
Lex Fridman (08:41.040)
that a couple of dozen of COVID cases
Erik Brynjolfsson (08:42.960)
can become thousands or tens or hundreds of thousands
Lex Fridman (08:46.200)
of them in a month,
Erik Brynjolfsson (08:48.080)
it makes sense once you just do the math.
Lex Fridman (08:51.200)
And I think Elon tries to do that a lot.
Erik Brynjolfsson (08:53.680)
You know, in fairness, I think he also benefits
Lex Fridman (08:55.120)
from hanging out in Silicon Valley
Lex Fridman (08:56.960)
and he's experienced it in a lot of different applications.
Lex Fridman (09:00.600)
So, you know, it's not as much of a shock to him anymore,
Lex Fridman (09:04.080)
but that's something we can all learn from.
Lex Fridman (09:07.160)
In my own life, I remember one of my first experiences
Erik Brynjolfsson (09:10.400)
really seeing it was when I was a grad student
Lex Fridman (09:12.960)
and my advisor asked me to plot the growth of computer power
Erik Brynjolfsson (09:17.560)
in the US economy in different industries.
Lex Fridman (09:20.000)
And there are all these, you know,
Erik Brynjolfsson (09:21.560)
exponentially growing curves.
Lex Fridman (09:23.000)
And I was like, holy shit, look at this.
Erik Brynjolfsson (09:24.560)
In each industry, it was just taking off.
Lex Fridman (09:26.880)
And, you know, you didn't have to be a rocket scientist
Erik Brynjolfsson (09:29.240)
to extend that and say, wow,
Lex Fridman (09:30.640)
this means that this was in the late 80s and early 90s
Erik Brynjolfsson (09:33.600)
that, you know, if it goes anything like that,
Lex Fridman (09:35.880)
we're gonna have orders of magnitude more computer power
Erik Brynjolfsson (09:38.160)
than we did at that time.
Lex Fridman (09:39.480)
And of course we do.
Erik Brynjolfsson (09:41.320)
So, you know, when people look at Moore's law,
Lex Fridman (09:45.040)
they often talk about it as just,
Lex Fridman (09:46.880)
so the exponential function is actually
Lex Fridman (09:49.240)
a stack of S curves.
Lex Fridman (09:51.360)
So basically it's you milk or whatever,
Lex Fridman (09:57.240)
take the most advantage of a particular little revolution
Lex Fridman (10:01.240)
and then you search for another revolution.
Lex Fridman (10:03.000)
And it's basically revolutions stack on top of revolutions.
Lex Fridman (10:06.720)
Do you have any intuition about how the head humans
Lex Fridman (10:08.960)
keep finding ways to revolutionize things?
Erik Brynjolfsson (10:12.280)
Well, first, let me just unpack that first point
Lex Fridman (10:14.200)
that I talked about exponential curves,
Lex Fridman (10:17.160)
but no exponential curve continues forever.
Lex Fridman (10:21.480)
It's been said that if anything can't go on forever,
Erik Brynjolfsson (10:24.960)
eventually it will stop.
Lex Fridman (10:26.680)
And, and it's very profound, but it's,
Erik Brynjolfsson (10:29.840)
it seems that a lot of people don't appreciate
Lex Fridman (10:32.440)
that half of it as well either.
Lex Fridman (10:33.960)
And that's why all exponential functions eventually turn
Lex Fridman (10:36.560)
into some kind of S curve or stop in some other way,
Erik Brynjolfsson (10:39.800)
maybe catastrophically.
Lex Fridman (10:41.240)
And that's a cap with COVID as well.
Erik Brynjolfsson (10:42.800)
I mean, it was, it went up and then it sort of, you know,
Lex Fridman (10:44.560)
at some point it starts saturating the pool of people
Erik Brynjolfsson (10:47.640)
to be infected.
Lex Fridman (10:49.040)
There's a standard epidemiological model
Erik Brynjolfsson (10:51.320)
that's based on that.
Lex Fridman (10:52.960)
And it's beginning to happen with Moore's law
Erik Brynjolfsson (10:55.040)
or different generations of computer power.
Lex Fridman (10:56.920)
It happens with all exponential curves.
Erik Brynjolfsson (10:59.280)
The remarkable thing is you elude,
Lex Fridman (11:01.040)
the second part of your question is that we've been able
Erik Brynjolfsson (11:03.560)
to come up with a new S curve on top of the previous one
Lex Fridman (11:06.840)
and do that generation after generation with new materials,
Erik Brynjolfsson (11:10.640)
new processes, and just extend it further and further.
Lex Fridman (11:15.680)
I don't think anyone has a really good theory
Erik Brynjolfsson (11:17.400)
about why we've been so successful in doing that.
Lex Fridman (11:21.400)
It's great that we have been,
Lex Fridman (11:23.160)
and I hope it continues for some time,
Lex Fridman (11:26.400)
but it's, you know, one beginning of a theory
Erik Brynjolfsson (11:31.480)
is that there's huge incentives when other parts
Lex Fridman (11:34.440)
of the system are going on that clock speed
Erik Brynjolfsson (11:36.880)
of doubling every two to three years.
Lex Fridman (11:39.280)
If there's one component of it that's not keeping up,
Erik Brynjolfsson (11:42.120)
then the economic incentives become really large
Lex Fridman (11:44.800)
to improve that one part.
Erik Brynjolfsson (11:46.200)
It becomes a bottleneck and anyone who can do improvements
Lex Fridman (11:49.720)
in that part can reap huge returns
Lex Fridman (11:51.640)
so that the resources automatically get focused
Lex Fridman (11:54.000)
on whatever part of the system isn't keeping up.
Lex Fridman (11:56.560)
Do you think some version of the Moore's law will continue?
Lex Fridman (11:59.680)
Some version, yes, it is.
Erik Brynjolfsson (12:01.360)
I mean, one version that has become more important
Lex Fridman (12:04.560)
is something I call Coomey's law,
Erik Brynjolfsson (12:06.280)
which is named after John Coomey,
Lex Fridman (12:08.440)
who I should mention was also my college roommate,
Lex Fridman (12:10.280)
but he identified the fact that energy consumption
Lex Fridman (12:14.360)
has been declining by a factor of two.
Lex Fridman (12:17.280)
And for most of us, that's more important.
Lex Fridman (12:18.960)
The new iPhones came out today as we're recording this.
Erik Brynjolfsson (12:21.320)
I'm not sure when you're gonna make it available.
Lex Fridman (12:23.120)
Very soon after this, yeah.
Lex Fridman (12:24.920)
And for most of us, having the iPhone be twice as fast,
Lex Fridman (12:30.000)
it's nice, but having the battery lifelonger,
Erik Brynjolfsson (12:33.160)
that would be much more valuable.
Lex Fridman (12:35.440)
And the fact that a lot of the progress in chips now
Erik Brynjolfsson (12:38.200)
is reducing energy consumption is probably more important
Lex Fridman (12:42.800)
for many applications than just the raw speed.
Erik Brynjolfsson (12:46.040)
Other dimensions of Moore's law
Lex Fridman (12:47.480)
are in AI and machine learning.
Erik Brynjolfsson (12:51.560)
Those tend to be very parallelizable functions,
Lex Fridman (12:55.160)
especially deep neural nets.
Lex Fridman (12:58.440)
And so instead of having one chip,
Lex Fridman (13:01.280)
you can have multiple chips or you can have a GPU,
Erik Brynjolfsson (13:05.000)
graphic processing unit that goes faster.
Lex Fridman (13:07.000)
Now, special chips designed for machine learning
Erik Brynjolfsson (13:09.600)
like tensor processing units,
Lex Fridman (13:11.160)
each time you switch, there's another 10X
Erik Brynjolfsson (13:13.000)
or 100X improvement above and beyond Moore's law.
Lex Fridman (13:16.720)
So I think that the raw silicon
Erik Brynjolfsson (13:18.800)
isn't improving as much as it used to,
Lex Fridman (13:20.720)
but these other dimensions are becoming important,
Erik Brynjolfsson (13:23.880)
more important, and we're seeing progress in them.
Lex Fridman (13:26.240)
I don't know if you've seen the work by OpenAI
Erik Brynjolfsson (13:28.200)
where they show the exponential improvement
Lex Fridman (13:31.880)
of the training of neural networks
Erik Brynjolfsson (13:34.320)
just literally in the techniques used.
Lex Fridman (13:36.920)
So that's almost like the algorithm.
Lex Fridman (13:40.520)
It's fascinating to think like, can I actually continue?
Lex Fridman (13:43.640)
I was figuring out more and more tricks
Erik Brynjolfsson (13:45.160)
on how to train networks faster and faster.
Lex Fridman (13:47.000)
The progress has been staggering.
Erik Brynjolfsson (13:49.520)
If you look at image recognition, as you mentioned,
Lex Fridman (13:51.720)
I think it's a function of at least three things
Erik Brynjolfsson (13:53.440)
that are coming together.
Lex Fridman (13:54.520)
One, we just talked about faster chips,
Erik Brynjolfsson (13:56.560)
not just Moore's law, but GPUs, TPUs and other technologies.
Lex Fridman (14:00.400)
The second is just a lot more data.
Erik Brynjolfsson (14:02.760)
I mean, we are awash in digital data today
Lex Fridman (14:05.600)
in a way we weren't 20 years ago.
Erik Brynjolfsson (14:08.080)
Photography, I'm old enough to remember,
Lex Fridman (14:09.960)
it used to be chemical, and now everything is digital.
Erik Brynjolfsson (14:12.800)
I took probably 50 digital photos yesterday.
Lex Fridman (14:16.440)
I wouldn't have done that if it was chemical.
Lex Fridman (14:17.880)
And we have the internet of things
Lex Fridman (14:20.680)
and all sorts of other types of data.
Erik Brynjolfsson (14:22.920)
When we walk around with our phone,
Lex Fridman (14:24.120)
it's just broadcasting a huge amounts of digital data
Erik Brynjolfsson (14:27.240)
that can be used as training sets.
Lex Fridman (14:29.240)
And then last but not least, as they mentioned at OpenAI,
Erik Brynjolfsson (14:34.240)
there've been significant improvements in the techniques.
Lex Fridman (14:37.160)
The core idea of deep neural nets
Erik Brynjolfsson (14:39.240)
has been around for a few decades,
Lex Fridman (14:41.160)
but the advances in making it work more efficiently
Erik Brynjolfsson (14:44.200)
have also improved a couple of orders of magnitude or more.
Lex Fridman (14:48.160)
So you multiply together,
Erik Brynjolfsson (14:49.640)
a hundred fold improvement in computer power,
Lex Fridman (14:52.480)
a hundred fold or more improvement in data,
Erik Brynjolfsson (14:55.320)
a hundred fold improvement in techniques
Lex Fridman (14:59.000)
of software and algorithms,
Lex Fridman (15:00.560)
and soon you're getting into a million fold improvements.
Lex Fridman (15:03.840)
So somebody brought this up, this idea with GPT3 that,
Lex Fridman (15:09.920)
so it's trained in a self supervised way
Lex Fridman (15:11.960)
on basically internet data.
Lex Fridman (15:15.080)
And that's one of the, I've seen arguments made
Lex Fridman (15:18.920)
and they seem to be pretty convincing
Erik Brynjolfsson (15:21.120)
that the bottleneck there is going to be
Lex Fridman (15:23.440)
how much data there is on the internet,
Erik Brynjolfsson (15:25.640)
which is a fascinating idea that it literally
Lex Fridman (15:29.120)
will just run out of human generated data to train on.
Erik Brynjolfsson (15:33.280)
Right, I know we make it to the point where it's consumed
Lex Fridman (15:35.720)
basically all of human knowledge
Erik Brynjolfsson (15:37.480)
or all digitized human knowledge, yeah.
Lex Fridman (15:39.120)
And that will be the bottleneck.
Lex Fridman (15:40.880)
But the interesting thing with bottlenecks
Lex Fridman (15:44.520)
is people often use bottlenecks
Erik Brynjolfsson (15:47.360)
as a way to argue against exponential growth.
Lex Fridman (15:49.920)
They say, well, there's no way
Erik Brynjolfsson (15:51.400)
you can overcome this bottleneck,
Lex Fridman (15:53.840)
but we seem to somehow keep coming up in new ways
Erik Brynjolfsson (15:56.960)
to like overcome whatever bottlenecks
Lex Fridman (15:59.200)
the critics come up with, which is fascinating.
Erik Brynjolfsson (16:01.920)
I don't know how you overcome the data bottleneck,
Lex Fridman (16:04.600)
but probably more efficient training algorithms.
Erik Brynjolfsson (16:07.000)
Yeah, well, you already mentioned that,
Lex Fridman (16:08.240)
that these training algorithms are getting much better
Erik Brynjolfsson (16:10.240)
at using smaller amounts of data.
Lex Fridman (16:12.440)
We also are just capturing a lot more data than we used to,
Erik Brynjolfsson (16:15.880)
especially in China, but all around us.
Lex Fridman (16:18.880)
So those are both important.
Erik Brynjolfsson (16:20.680)
In some applications, you can simulate the data,
Lex Fridman (16:24.160)
video games, some of the self driving car systems
Erik Brynjolfsson (16:28.920)
are simulating driving, and of course,
Lex Fridman (16:32.520)
that has some risks and weaknesses,
Lex Fridman (16:34.240)
but you can also, if you want to exhaust
Lex Fridman (16:38.080)
all the different ways you could beat a video game,
Erik Brynjolfsson (16:39.840)
you could just simulate all the options.
Lex Fridman (16:42.280)
Can we take a step in that direction of autonomous vehicles?
Erik Brynjolfsson (16:44.920)
Next, you're talking to the CTO of Waymo tomorrow.
Lex Fridman (16:48.640)
And obviously, I'm talking to Elon again in a couple of weeks.
Lex Fridman (16:53.920)
What's your thoughts on autonomous vehicles?
Lex Fridman (16:57.040)
Like where do we stand as a problem
Lex Fridman (17:01.800)
that has the potential of revolutionizing the world?
Lex Fridman (17:04.480)
Well, I'm really excited about that,
Lex Fridman (17:06.760)
but it's become much clearer
Lex Fridman (17:09.000)
that the original way that I thought about it,
Erik Brynjolfsson (17:10.680)
most people thought about like,
Lex Fridman (17:11.520)
you know, will we have a self driving car or not
Erik Brynjolfsson (17:13.440)
is way too simple.
Lex Fridman (17:15.640)
The better way to think about it
Erik Brynjolfsson (17:17.400)
is that there's a whole continuum
Lex Fridman (17:19.360)
of how much driving and assisting the car can do.
Erik Brynjolfsson (17:22.320)
I noticed that you're right next door
Lex Fridman (17:24.760)
to the Toyota Research Institute.
Erik Brynjolfsson (17:25.960)
That is a total accident.
Lex Fridman (17:27.600)
I love the TRI folks, but yeah.
Lex Fridman (17:29.320)
Have you talked to Gil Pratt?
Lex Fridman (17:30.800)
Yeah, we're supposed to talk.
Erik Brynjolfsson (17:34.080)
It's kind of hilarious.
Lex Fridman (17:34.960)
So there's kind of the,
Erik Brynjolfsson (17:35.800)
I think it's a good counterpart to say what Elon is doing.
Lex Fridman (17:38.720)
And hopefully they can be frank
Erik Brynjolfsson (17:40.040)
in what they think about each other,
Lex Fridman (17:41.440)
because I've heard both of them talk about it.
Lex Fridman (17:43.920)
But they're much more, you know,
Lex Fridman (17:45.400)
this is an assistive, a guardian angel
Erik Brynjolfsson (17:47.440)
that watches over you as opposed to try to do everything.
Lex Fridman (17:50.440)
I think there's some things like driving on a highway,
Erik Brynjolfsson (17:53.320)
you know, from LA to Phoenix,
Lex Fridman (17:55.320)
where it's mostly good weather, straight roads.
Erik Brynjolfsson (17:58.120)
That's close to a solved problem, let's face it.
Lex Fridman (18:01.240)
In other situations, you know,
Erik Brynjolfsson (18:02.560)
driving through the snow in Boston
Lex Fridman (18:04.640)
where the roads are kind of crazy.
Lex Fridman (18:06.160)
And most importantly, you have to make a lot of judgments
Lex Fridman (18:08.200)
about what the other driver is gonna do
Erik Brynjolfsson (18:09.440)
at these intersections that aren't really right angles
Lex Fridman (18:11.680)
and aren't very well described.
Erik Brynjolfsson (18:13.400)
It's more like game theory.
Lex Fridman (18:15.320)
That's a much harder problem
Lex Fridman (18:17.080)
and requires understanding human motivations.
Lex Fridman (18:22.080)
So there's a continuum there of some places
Erik Brynjolfsson (18:24.320)
where the cars will work very well
Lex Fridman (18:27.560)
and others where it could probably take decades.
Lex Fridman (18:30.920)
What do you think about the Waymo?
Lex Fridman (18:33.480)
So you mentioned two companies
Erik Brynjolfsson (18:36.000)
that actually have cars on the road.
Lex Fridman (18:38.040)
There's the Waymo approach that it's more like
Erik Brynjolfsson (18:40.640)
we're not going to release anything until it's perfect
Lex Fridman (18:42.800)
and we're gonna be very strict
Erik Brynjolfsson (18:45.320)
about the streets that we travel on,
Lex Fridman (18:47.680)
but it better be perfect.
Erik Brynjolfsson (18:49.160)
Yeah.
Lex Fridman (18:50.200)
Well, I'm smart enough to be humble
Lex Fridman (18:53.920)
and not try to get between.
Lex Fridman (18:55.000)
I know there's very bright people
Erik Brynjolfsson (18:56.600)
on both sides of the argument.
Lex Fridman (18:57.440)
I've talked to them and they make convincing arguments to me
Erik Brynjolfsson (19:00.000)
about how careful they need to be and the social acceptance.
Lex Fridman (19:04.000)
Some people thought that when the first few people died
Erik Brynjolfsson (19:07.400)
from self driving cars, that would shut down the industry,
Lex Fridman (19:09.840)
but it was more of a blip actually.
Erik Brynjolfsson (19:11.800)
And, you know, so that was interesting.
Lex Fridman (19:14.440)
Of course, there's still a concern
Erik Brynjolfsson (19:16.040)
that if there could be setbacks, if we do this wrong,
Lex Fridman (19:20.560)
you know, your listeners may be familiar
Erik Brynjolfsson (19:22.600)
with the different levels of self driving,
Lex Fridman (19:24.160)
you know, level one, two, three, four, five.
Erik Brynjolfsson (19:26.800)
I think Andrew Ng has convinced me that this idea
Lex Fridman (19:29.640)
of really focusing on level four,
Erik Brynjolfsson (19:32.600)
where you only go in areas that are well mapped
Lex Fridman (19:35.000)
rather than just going out in the wild
Erik Brynjolfsson (19:37.480)
is the way things are gonna evolve.
Lex Fridman (19:39.720)
But you can just keep expanding those areas
Erik Brynjolfsson (19:42.600)
where you've mapped things really well,
Lex Fridman (19:44.040)
where you really understand them
Lex Fridman (19:45.040)
and eventually all become kind of interconnected.
Lex Fridman (19:47.960)
And that could be a kind of another way of progressing
Erik Brynjolfsson (19:51.160)
to make it more feasible over time.
Lex Fridman (19:55.240)
I mean, that's kind of like the Waymo approach,
Erik Brynjolfsson (19:57.400)
which is they just now released,
Lex Fridman (19:59.520)
I think just like a day or two ago,
Erik Brynjolfsson (1:00:01.560)
in technology, there's just too many ways to go wrong.
Lex Fridman (1:00:04.640)
There's a lot of ways to blow yourself up.
Lex Fridman (1:00:06.200)
And people, or I should say species,
Lex Fridman (1:00:09.640)
end up falling into one of those traps.
Erik Brynjolfsson (1:00:12.520)
The great filter.
Lex Fridman (1:00:13.580)
The great filter.
Erik Brynjolfsson (1:00:14.960)
I mean, there's an optimistic view of that.
Lex Fridman (1:00:16.720)
If there is literally no intelligent life out there
Erik Brynjolfsson (1:00:19.380)
in the universe, or at least in our galaxy,
Lex Fridman (1:00:22.340)
that means that we've passed at least one
Erik Brynjolfsson (1:00:25.140)
of the great filters or some of the great filters
Lex Fridman (1:00:27.840)
that we survived.
Erik Brynjolfsson (1:00:30.040)
Yeah, no, I think Robin Hansen has a good way of,
Lex Fridman (1:00:32.240)
maybe others have a good way of thinking about this,
Erik Brynjolfsson (1:00:33.920)
that if there are no other intelligence creatures out there
Lex Fridman (1:00:38.920)
that we've been able to detect,
Erik Brynjolfsson (1:00:40.640)
one possibility is that there's a filter ahead of us.
Lex Fridman (1:00:43.440)
And when you get a little more advanced,
Erik Brynjolfsson (1:00:44.780)
maybe in a hundred or a thousand or 10,000 years,
Lex Fridman (1:00:47.600)
things just get destroyed for some reason.
Erik Brynjolfsson (1:00:50.560)
The other one is the great filters behind us.
Lex Fridman (1:00:53.000)
That'll be good, is that most planets don't even evolve life
Erik Brynjolfsson (1:00:57.700)
or if they don't evolve life,
Lex Fridman (1:00:58.920)
they don't evolve intelligent life.
Erik Brynjolfsson (1:01:00.280)
Maybe we've gotten past that.
Lex Fridman (1:01:02.040)
And so now maybe we're on the good side
Erik Brynjolfsson (1:01:03.960)
of the great filter.
Lex Fridman (1:01:05.680)
So if we sort of rewind back and look at the thing
Erik Brynjolfsson (1:01:10.480)
where we could say something a little bit more comfortably
Lex Fridman (1:01:12.760)
at five years and 10 years out,
Erik Brynjolfsson (1:01:15.860)
you've written about jobs
Lex Fridman (1:01:20.200)
and the impact on sort of our economy and the jobs
Erik Brynjolfsson (1:01:24.680)
in terms of artificial intelligence that it might have.
Lex Fridman (1:01:28.240)
It's a fascinating question of what kind of jobs are safe,
Lex Fridman (1:01:30.560)
what kind of jobs are not.
Lex Fridman (1:01:32.520)
Can you maybe speak to your intuition
Erik Brynjolfsson (1:01:34.560)
about how we should think about AI changing
Lex Fridman (1:01:38.320)
the landscape of work?
Erik Brynjolfsson (1:01:39.940)
Sure, absolutely.
Lex Fridman (1:01:40.880)
Well, this is a really important question
Erik Brynjolfsson (1:01:42.600)
because I think we're very far
Lex Fridman (1:01:43.900)
from artificial general intelligence,
Erik Brynjolfsson (1:01:45.720)
which is AI that can just do the full breadth
Lex Fridman (1:01:48.120)
of what humans can do.
Lex Fridman (1:01:49.520)
But we do have human level or superhuman level
Lex Fridman (1:01:52.980)
narrow intelligence, narrow artificial intelligence.
Lex Fridman (1:01:56.800)
And obviously my calculator can do math a lot better
Lex Fridman (1:01:59.880)
than I can.
Lex Fridman (1:02:00.720)
And there's a lot of other things
Lex Fridman (1:02:01.560)
that machines can do better than I can.
Lex Fridman (1:02:03.160)
So which is which?
Lex Fridman (1:02:04.440)
We actually set out to address that question
Erik Brynjolfsson (1:02:06.860)
with Tom Mitchell.
Lex Fridman (1:02:08.160)
I wrote a paper called what can machine learning do
Erik Brynjolfsson (1:02:12.160)
that was in science.
Lex Fridman (1:02:13.440)
And we went and interviewed a whole bunch of AI experts
Lex Fridman (1:02:16.840)
and kind of synthesized what they thought machine learning
Lex Fridman (1:02:20.440)
was good at and wasn't good at.
Lex Fridman (1:02:22.220)
And we came up with what we called a rubric,
Lex Fridman (1:02:25.540)
basically a set of questions you can ask about any task
Erik Brynjolfsson (1:02:28.160)
that will tell you whether it's likely to score high or low
Lex Fridman (1:02:30.960)
on suitability for machine learning.
Lex Fridman (1:02:33.720)
And then we've applied that
Lex Fridman (1:02:34.760)
to a bunch of tasks in the economy.
Erik Brynjolfsson (1:02:36.940)
In fact, there's a data set of all the tasks
Lex Fridman (1:02:39.080)
in the US economy, believe it or not, it's called ONET.
Erik Brynjolfsson (1:02:41.600)
The US government put it together,
Lex Fridman (1:02:43.120)
part of the Bureau of Labor Statistics.
Erik Brynjolfsson (1:02:45.000)
They divide the economy into about 970 occupations
Lex Fridman (1:02:48.680)
like bus driver, economist, primary school teacher,
Erik Brynjolfsson (1:02:52.140)
radiologist, and then for each one of them,
Lex Fridman (1:02:54.800)
they describe which tasks need to be done.
Erik Brynjolfsson (1:02:57.580)
Like for radiologists, there are 27 distinct tasks.
Lex Fridman (1:03:00.720)
So we went through all those tasks
Erik Brynjolfsson (1:03:02.160)
to see whether or not a machine could do them.
Lex Fridman (1:03:04.960)
And what we found interestingly was...
Erik Brynjolfsson (1:03:06.680)
Brilliant study by the way, that's so awesome.
Lex Fridman (1:03:08.880)
Yeah, thank you.
Lex Fridman (1:03:10.240)
So what we found was that there was no occupation
Lex Fridman (1:03:13.760)
in our data set where machine learning just ran the table
Lex Fridman (1:03:16.240)
and did everything.
Lex Fridman (1:03:17.520)
And there was almost no occupation
Erik Brynjolfsson (1:03:18.980)
where machine learning didn't have
Lex Fridman (1:03:19.900)
like a significant ability to do things.
Erik Brynjolfsson (1:03:22.120)
Like take radiology, a lot of people I hear saying,
Lex Fridman (1:03:24.360)
you know, it's the end of radiology.
Lex Fridman (1:03:26.680)
And one of the 27 tasks is read medical images.
Lex Fridman (1:03:29.880)
Really important one, like it's kind of a core job.
Lex Fridman (1:03:31.960)
And machines have basically gotten as good
Lex Fridman (1:03:34.640)
or better than radiologists.
Erik Brynjolfsson (1:03:35.880)
There was just an article in Nature last week,
Lex Fridman (1:03:38.360)
but they've been publishing them for the past few years
Erik Brynjolfsson (1:03:42.440)
showing that machine learning can do as well as humans
Lex Fridman (1:03:46.480)
on many kinds of diagnostic imaging tasks.
Lex Fridman (1:03:49.600)
But other things that radiologists do,
Lex Fridman (1:03:51.120)
they sometimes administer conscious sedation.
Erik Brynjolfsson (1:03:54.440)
They sometimes do physical exams.
Lex Fridman (1:03:55.940)
They have to synthesize the results
Lex Fridman (1:03:57.320)
and explain it to the other doctors or to the patients.
Lex Fridman (1:04:01.680)
In all those categories,
Erik Brynjolfsson (1:04:02.520)
machine learning isn't really up to snuff yet.
Lex Fridman (1:04:05.560)
So that job, we're gonna see a lot of restructuring.
Erik Brynjolfsson (1:04:09.300)
Parts of the job, they'll hand over to machines.
Lex Fridman (1:04:11.400)
Others, humans will do more of.
Erik Brynjolfsson (1:04:13.160)
That's been more or less the pattern all of them.
Lex Fridman (1:04:15.080)
So, you know, to oversimplify a bit,
Erik Brynjolfsson (1:04:17.080)
we're gonna see a lot of restructuring,
Lex Fridman (1:04:19.080)
reorganization of work.
Lex Fridman (1:04:20.400)
And it's real gonna be a great time.
Lex Fridman (1:04:22.300)
It is a great time for smart entrepreneurs and managers
Erik Brynjolfsson (1:04:24.720)
to do that reinvention of work.
Lex Fridman (1:04:27.280)
I'm not gonna see mass unemployment.
Erik Brynjolfsson (1:04:30.600)
To get more specifically to your question,
Lex Fridman (1:04:33.120)
the kinds of tasks that machines tend to be good at
Erik Brynjolfsson (1:04:36.560)
are a lot of routine problem solving,
Lex Fridman (1:04:39.040)
mapping inputs X into outputs Y.
Erik Brynjolfsson (1:04:42.560)
If you have a lot of data on the Xs and the Ys,
Lex Fridman (1:04:44.840)
the inputs and the outputs,
Erik Brynjolfsson (1:04:45.680)
you can do that kind of mapping and find the relationships.
Lex Fridman (1:04:48.520)
They tend to not be very good at,
Erik Brynjolfsson (1:04:50.660)
even now, fine motor control and dexterity.
Lex Fridman (1:04:53.680)
Emotional intelligence and human interactions
Lex Fridman (1:04:58.960)
and thinking outside the box, creative work.
Lex Fridman (1:05:01.700)
If you give it a well structured task,
Erik Brynjolfsson (1:05:03.220)
machines can be very good at it.
Lex Fridman (1:05:05.040)
But even asking the right questions, that's hard.
Erik Brynjolfsson (1:05:08.680)
There's a quote that Andrew McAfee and I use
Lex Fridman (1:05:10.680)
in our book, Second Machine Age.
Erik Brynjolfsson (1:05:12.980)
Apparently Pablo Picasso was shown an early computer
Lex Fridman (1:05:16.840)
and he came away kind of unimpressed.
Erik Brynjolfsson (1:05:18.460)
He goes, well, I don't see all the fusses.
Lex Fridman (1:05:20.660)
All that does is answer questions.
Lex Fridman (1:05:23.900)
And to him, the interesting thing was asking the questions.
Lex Fridman (1:05:26.740)
Yeah, try to replace me, GPT3, I dare you.
Erik Brynjolfsson (1:05:31.260)
Although some people think I'm a robot.
Lex Fridman (1:05:33.160)
You have this cool plot that shows,
Erik Brynjolfsson (1:05:37.020)
I just remember where economists land,
Lex Fridman (1:05:39.640)
where I think the X axis is the income.
Lex Fridman (1:05:43.380)
And then the Y axis is, I guess,
Lex Fridman (1:05:46.220)
aggregating the information of how replaceable the job is.
Erik Brynjolfsson (1:05:49.380)
Or I think there's an index.
Lex Fridman (1:05:50.780)
There's a suitability for machine learning index.
Erik Brynjolfsson (1:05:51.620)
Exactly.
Lex Fridman (1:05:52.460)
So we have all 970 occupations on that chart.
Erik Brynjolfsson (1:05:55.300)
It's a cool plot.
Lex Fridman (1:05:56.500)
And there's scatters in all four corners
Erik Brynjolfsson (1:05:59.200)
have some occupations.
Lex Fridman (1:06:01.040)
But there is a definite pattern,
Erik Brynjolfsson (1:06:02.700)
which is the lower wage occupations tend to have more tasks
Lex Fridman (1:06:05.660)
that are suitable for machine learning, like cashiers.
Erik Brynjolfsson (1:06:07.960)
I mean, anyone who's gone to a supermarket or CVS
Lex Fridman (1:06:10.400)
knows that they not only read barcodes,
Lex Fridman (1:06:12.380)
but they can recognize an apple and an orange
Lex Fridman (1:06:14.520)
and a lot of things cashiers, humans used to be needed for.
Erik Brynjolfsson (1:06:19.520)
At the other end of the spectrum,
Lex Fridman (1:06:21.020)
there are some jobs like airline pilot
Erik Brynjolfsson (1:06:23.580)
that are among the highest paid in our economy,
Lex Fridman (1:06:26.640)
but also a lot of them are suitable for machine learning.
Erik Brynjolfsson (1:06:28.780)
A lot of those tasks are.
Lex Fridman (1:06:30.940)
And then, yeah, you mentioned economists.
Erik Brynjolfsson (1:06:32.500)
I couldn't help peeking at those
Lex Fridman (1:06:33.820)
and they're paid a fair amount,
Erik Brynjolfsson (1:06:36.100)
maybe not as much as some of us think they should be.
Lex Fridman (1:06:39.120)
But they have some tasks that are suitable
Erik Brynjolfsson (1:06:43.620)
for machine learning, but for now at least,
Lex Fridman (1:06:45.540)
most of the tasks of economists
Erik Brynjolfsson (1:06:47.180)
didn't end up being in that category.
Lex Fridman (1:06:48.540)
And I should say, I didn't like create that data.
Erik Brynjolfsson (1:06:50.640)
We just took the analysis and that's what came out of it.
Lex Fridman (1:06:54.480)
And over time, that scatter plot will be updated
Erik Brynjolfsson (1:06:57.320)
as the technology improves.
Lex Fridman (1:06:59.940)
But it was just interesting to see the pattern there.
Lex Fridman (1:07:02.860)
And it is a little troubling in so far
Lex Fridman (1:07:05.140)
as if you just take the technology as it is today,
Erik Brynjolfsson (1:07:08.100)
it's likely to worsen income inequality
Lex Fridman (1:07:10.520)
on a lot of dimensions.
Lex Fridman (1:07:12.260)
So on this topic of the effect of AI
Lex Fridman (1:07:16.480)
on our landscape of work,
Erik Brynjolfsson (1:07:21.060)
one of the people that have been speaking about it
Lex Fridman (1:07:23.660)
in the public domain, public discourse
Erik Brynjolfsson (1:07:25.800)
is the presidential candidate, Andrew Yang.
Lex Fridman (1:07:28.100)
Yeah.
Lex Fridman (1:07:29.040)
What are your thoughts about Andrew?
Lex Fridman (1:07:31.900)
What are your thoughts about UBI,
Erik Brynjolfsson (1:07:34.340)
that universal basic income
Lex Fridman (1:07:36.700)
that he made one of the core ideas,
Erik Brynjolfsson (1:07:39.100)
by the way, he has like hundreds of ideas
Lex Fridman (1:07:40.780)
about like everything, it's kind of interesting.
Lex Fridman (1:07:44.020)
But what are your thoughts about him
Lex Fridman (1:07:45.380)
and what are your thoughts about UBI?
Erik Brynjolfsson (1:07:46.740)
Let me answer the question about his broader approach first.
Lex Fridman (1:07:52.060)
I mean, I just love that.
Erik Brynjolfsson (1:07:52.900)
He's really thoughtful, analytical.
Lex Fridman (1:07:56.460)
I agree with his values.
Lex Fridman (1:07:58.220)
So that's awesome.
Lex Fridman (1:07:59.420)
And he read my book and mentions it sometimes,
Lex Fridman (1:08:02.220)
so it makes me even more excited.
Lex Fridman (1:08:04.820)
And the thing that he really made the centerpiece
Erik Brynjolfsson (1:08:07.660)
of his campaign was UBI.
Lex Fridman (1:08:09.940)
And I was originally kind of a fan of it.
Lex Fridman (1:08:13.260)
And then as I studied it more, I became less of a fan,
Lex Fridman (1:08:15.980)
although I'm beginning to come back a little bit.
Lex Fridman (1:08:17.420)
So let me tell you a little bit of my evolution.
Lex Fridman (1:08:19.300)
As an economist, we have, by looking at the problem
Erik Brynjolfsson (1:08:23.060)
of people not having enough income and the simplest thing
Lex Fridman (1:08:25.180)
is, well, why don't we write them a check?
Erik Brynjolfsson (1:08:26.860)
Problem solved.
Lex Fridman (1:08:28.040)
But then I talked to my sociologist friends
Lex Fridman (1:08:30.460)
and they really convinced me that just writing a check
Lex Fridman (1:08:34.420)
doesn't really get at the core values.
Erik Brynjolfsson (1:08:36.940)
Voltaire once said that work solves three great ills,
Lex Fridman (1:08:40.660)
boredom, vice, and need.
Lex Fridman (1:08:43.380)
And you can deal with the need thing by writing a check,
Lex Fridman (1:08:46.680)
but people need a sense of meaning,
Erik Brynjolfsson (1:08:49.300)
they need something to do.
Lex Fridman (1:08:50.820)
And when, say, steel workers or coal miners lost their jobs
Lex Fridman (1:08:57.980)
and were just given checks, alcoholism, depression, divorce,
Lex Fridman (1:09:03.820)
all those social indicators, drug use, all went way up.
Erik Brynjolfsson (1:09:06.540)
People just weren't happy
Lex Fridman (1:09:08.020)
just sitting around collecting a check.
Erik Brynjolfsson (1:09:11.380)
Maybe it's part of the way they were raised.
Lex Fridman (1:09:13.220)
Maybe it's something innate in people
Erik Brynjolfsson (1:09:14.740)
that they need to feel wanted and needed.
Lex Fridman (1:09:17.220)
So it's not as simple as just writing people a check.
Erik Brynjolfsson (1:09:19.540)
You need to also give them a way to have a sense of purpose.
Lex Fridman (1:09:23.980)
And that was important to me.
Lex Fridman (1:09:25.380)
And the second thing is that, as I mentioned earlier,
Lex Fridman (1:09:28.740)
we are far from the end of work.
Erik Brynjolfsson (1:09:31.160)
I don't buy the idea that there's just like
Lex Fridman (1:09:32.800)
not enough work to be done.
Erik Brynjolfsson (1:09:34.140)
I see like our cities need to be cleaned up.
Lex Fridman (1:09:37.100)
And robots can't do most of that.
Erik Brynjolfsson (1:09:39.580)
We need to have better childcare.
Lex Fridman (1:09:40.780)
We need better healthcare.
Erik Brynjolfsson (1:09:41.640)
We need to take care of people who are mentally ill or older.
Lex Fridman (1:09:44.940)
We need to repair our roads.
Erik Brynjolfsson (1:09:46.500)
There's so much work that require at least partly,
Lex Fridman (1:09:49.940)
maybe entirely a human component.
Lex Fridman (1:09:52.300)
So rather than like write all these people off,
Lex Fridman (1:09:54.660)
let's find a way to repurpose them and keep them engaged.
Erik Brynjolfsson (1:09:58.240)
Now that said, I would like to see more buying power
Lex Fridman (1:10:04.640)
from people who are sort of at the bottom end
Erik Brynjolfsson (1:10:06.400)
of the spectrum.
Lex Fridman (1:10:07.320)
The economy has been designed and evolved in a way
Erik Brynjolfsson (1:10:12.540)
that's I think very unfair to a lot of hardworking people.
Lex Fridman (1:10:15.600)
I see super hardworking people who aren't really seeing
Erik Brynjolfsson (1:10:18.100)
their wages grow over the past 20, 30 years,
Lex Fridman (1:10:20.720)
while some other people who have been super smart
Lex Fridman (1:10:24.080)
and or super lucky have made billions
Lex Fridman (1:10:29.480)
or hundreds of billions.
Lex Fridman (1:10:30.920)
And I don't think they need those hundreds of billions
Lex Fridman (1:10:33.800)
to have the right incentives to invent things.
Erik Brynjolfsson (1:10:35.740)
I think if you talk to almost any of them as I have,
Lex Fridman (1:10:39.440)
they don't think that they need an extra $10 billion
Erik Brynjolfsson (1:10:42.440)
to do what they're doing.
Lex Fridman (1:10:43.560)
Most of them probably would love to do it for only a billion
Erik Brynjolfsson (1:10:48.120)
or maybe for nothing.
Lex Fridman (1:10:49.360)
For nothing, many of them, yeah.
Erik Brynjolfsson (1:10:50.800)
I mean, an interesting point to make is,
Lex Fridman (1:10:54.200)
do we think that Bill Gates would have founded Microsoft
Lex Fridman (1:10:56.640)
if tax rates were 70%?
Lex Fridman (1:10:58.720)
Well, we know he would have because they were tax rates
Erik Brynjolfsson (1:11:01.380)
of 70% when he founded it.
Lex Fridman (1:11:03.680)
So I don't think that's as big a deterrent
Lex Fridman (1:11:06.200)
and we could provide more buying power to people.
Lex Fridman (1:11:09.100)
My own favorite tool is the Earned Income Tax Credit,
Erik Brynjolfsson (1:11:12.800)
which is basically a way of supplementing income
Lex Fridman (1:11:16.240)
of people who have jobs and giving employers
Erik Brynjolfsson (1:11:18.160)
an incentive to hire even more people.
Lex Fridman (1:11:20.300)
The minimum wage can discourage employment,
Lex Fridman (1:11:22.400)
but the Earned Income Tax Credit encourages employment
Lex Fridman (1:11:25.160)
by supplementing people's wages.
Erik Brynjolfsson (1:11:27.960)
If the employer can only afford to pay them $10 for a task,
Lex Fridman (1:11:32.680)
the rest of us kick in another five or $10
Lex Fridman (1:11:35.200)
and bring their wages up to 15 or 20 total.
Lex Fridman (1:11:37.640)
And then they have more buying power.
Lex Fridman (1:11:39.360)
Then entrepreneurs are thinking, how can we cater to them?
Lex Fridman (1:11:42.320)
How can we make products for them?
Lex Fridman (1:11:44.080)
And it becomes a self reinforcing system
Lex Fridman (1:11:47.220)
where people are better off.
Erik Brynjolfsson (1:11:49.840)
Ian Drang and I had a good discussion
Lex Fridman (1:11:51.840)
where he suggested instead of a universal basic income,
Erik Brynjolfsson (1:11:55.940)
he suggested, or instead of an unconditional basic income,
Lex Fridman (1:11:59.080)
how about a conditional basic income
Erik Brynjolfsson (1:12:00.600)
where the condition is you learn some new skills,
Lex Fridman (1:12:03.040)
we need to reskill our workforce.
Lex Fridman (1:12:05.040)
So let's make it easier for people to find ways
Lex Fridman (1:12:09.120)
to get those skills and get rewarded for doing them.
Lex Fridman (1:12:11.280)
And that's kind of a neat idea as well.
Lex Fridman (1:12:13.080)
That's really interesting.
Erik Brynjolfsson (1:12:13.900)
So, I mean, one of the questions,
Lex Fridman (1:12:16.160)
one of the dreams of UBI is that you provide
Erik Brynjolfsson (1:12:19.680)
some little safety net while you retrain,
Lex Fridman (1:12:24.280)
while you learn a new skill.
Lex Fridman (1:12:26.040)
But like, I think, I guess you're speaking
Lex Fridman (1:12:28.360)
to the intuition that that doesn't always,
Erik Brynjolfsson (1:12:31.280)
like there needs to be some incentive to reskill,
Lex Fridman (1:12:33.760)
to train, to learn a new thing.
Erik Brynjolfsson (1:12:35.280)
I think it helps.
Lex Fridman (1:12:36.120)
I mean, there are lots of self motivated people,
Lex Fridman (1:12:37.960)
but there are also people that maybe need a little guidance
Lex Fridman (1:12:40.600)
or help and I think it's a really hard question
Erik Brynjolfsson (1:12:44.960)
for someone who is losing a job in one area to know
Lex Fridman (1:12:48.280)
what is the new area I should be learning skills in.
Lex Fridman (1:12:50.600)
And we could provide a much better set of tools
Lex Fridman (1:12:52.600)
and platforms that maps it.
Erik Brynjolfsson (1:12:54.480)
Okay, here's a set of skills you already have.
Lex Fridman (1:12:56.400)
Here's something that's in demand.
Erik Brynjolfsson (1:12:58.120)
Let's create a path for you to go from where you are
Lex Fridman (1:13:00.440)
to where you need to be.
Lex Fridman (1:13:03.120)
So I'm a total, how do I put it nicely about myself?
Lex Fridman (1:13:07.080)
I'm totally clueless about the economy.
Erik Brynjolfsson (1:13:09.640)
It's not totally true, but pretty good approximation.
Lex Fridman (1:13:12.760)
If you were to try to fix our tax system
Erik Brynjolfsson (1:13:20.480)
and, or maybe from another side,
Lex Fridman (1:13:23.240)
if there's fundamental problems in taxation
Erik Brynjolfsson (1:13:26.680)
or some fundamental problems about our economy,
Lex Fridman (1:13:29.720)
what would you try to fix?
Lex Fridman (1:13:31.320)
What would you try to speak to?
Lex Fridman (1:13:33.440)
You know, I definitely think our whole tax system,
Erik Brynjolfsson (1:13:36.320)
our political and economic system has gotten more
Lex Fridman (1:13:40.080)
and more screwed up over the past 20, 30 years.
Erik Brynjolfsson (1:13:43.520)
I don't think it's that hard to make headway
Lex Fridman (1:13:46.520)
in improving it.
Erik Brynjolfsson (1:13:47.360)
I don't think we need to totally reinvent stuff.
Lex Fridman (1:13:49.880)
A lot of it is what I've been elsewhere with Andy
Lex Fridman (1:13:52.400)
and others called economics 101.
Lex Fridman (1:13:54.680)
You know, there's just some basic principles
Erik Brynjolfsson (1:13:56.400)
that have worked really well in the 20th century
Lex Fridman (1:14:00.640)
that we sort of forgot, you know,
Erik Brynjolfsson (1:14:01.880)
in terms of investing in education,
Lex Fridman (1:14:03.960)
investing in infrastructure, welcoming immigrants,
Erik Brynjolfsson (1:14:07.560)
having a tax system that was more progressive and fair.
Lex Fridman (1:14:13.280)
At one point, tax rates were on top incomes
Erik Brynjolfsson (1:14:16.560)
were significantly higher.
Lex Fridman (1:14:18.080)
And they've come down a lot to the point where
Erik Brynjolfsson (1:14:19.880)
in many cases they're lower now
Lex Fridman (1:14:21.440)
than they are for poorer people.
Erik Brynjolfsson (1:14:24.760)
So, and we could do things like earned income tax credit
Lex Fridman (1:14:27.960)
to get a little more wonky.
Erik Brynjolfsson (1:14:29.240)
I'd like to see more Pigouvian taxes.
Lex Fridman (1:14:31.440)
What that means is you tax things that are bad
Erik Brynjolfsson (1:14:35.720)
instead of things that are good.
Lex Fridman (1:14:36.960)
So right now we tax labor, we tax capital
Lex Fridman (1:14:40.640)
and which is unfortunate
Lex Fridman (1:14:42.200)
because one of the basic principles of economics
Erik Brynjolfsson (1:14:44.080)
if you tax something, you tend to get less of it.
Lex Fridman (1:14:46.400)
So, you know, right now there's still work to be done
Lex Fridman (1:14:48.800)
and still capital to be invested in.
Lex Fridman (1:14:51.220)
But instead we should be taxing things like pollution
Lex Fridman (1:14:54.600)
and congestion.
Lex Fridman (1:14:57.200)
And if we did that, we would have less pollution.
Lex Fridman (1:15:00.000)
So a carbon tax is, you know,
Lex Fridman (1:15:02.120)
almost every economist would say it's a no brainer
Erik Brynjolfsson (1:15:04.120)
whether they're Republican or Democrat,
Lex Fridman (1:15:07.560)
Greg Mankiw who is head of George Bush's
Erik Brynjolfsson (1:15:09.680)
Council of Economic Advisers or Dick Schmollensie
Lex Fridman (1:15:13.000)
who is another Republican economist agree.
Lex Fridman (1:15:16.080)
And of course a lot of Democratic economists agree as well.
Lex Fridman (1:15:21.600)
If we taxed carbon,
Erik Brynjolfsson (1:15:22.800)
we could raise hundreds of billions of dollars.
Lex Fridman (1:15:26.040)
We could take that money and redistribute it
Erik Brynjolfsson (1:15:28.600)
through an earned income tax credit or other things
Lex Fridman (1:15:31.200)
so that overall our tax system would become more progressive.
Erik Brynjolfsson (1:15:35.280)
We could tax congestion.
Lex Fridman (1:15:36.960)
One of the things that kills me as an economist
Erik Brynjolfsson (1:15:39.040)
is every time I sit in a traffic jam,
Lex Fridman (1:15:41.080)
I know that it's completely unnecessary.
Erik Brynjolfsson (1:15:43.280)
This is complete wasted time.
Lex Fridman (1:15:44.840)
You just visualize the cost and productivity.
Erik Brynjolfsson (1:15:47.560)
Exactly, because they are taking costs for me
Lex Fridman (1:15:51.260)
and all the people around me.
Lex Fridman (1:15:52.700)
And if they charged a congestion tax,
Lex Fridman (1:15:54.840)
they would take that same amount of money
Lex Fridman (1:15:57.080)
and people would, it would streamline the roads.
Lex Fridman (1:15:59.720)
Like when you're in Singapore, the traffic just flows
Erik Brynjolfsson (1:16:01.640)
because they have a congestion tax.
Lex Fridman (1:16:02.640)
They listened to economists.
Erik Brynjolfsson (1:16:03.640)
They invited me and others to go talk to them.
Lex Fridman (1:16:06.480)
And then I'd still be paying,
Erik Brynjolfsson (1:16:09.240)
I'd be paying a congestion tax instead of paying in my time,
Lex Fridman (1:16:11.740)
but that money would now be available for healthcare,
Erik Brynjolfsson (1:16:14.240)
be available for infrastructure,
Lex Fridman (1:16:15.520)
or be available just to give to people
Lex Fridman (1:16:16.880)
so they could buy food or whatever.
Lex Fridman (1:16:18.660)
So it's just, it saddens me when you sit,
Erik Brynjolfsson (1:16:22.280)
when you're sitting in a traffic jam,
Lex Fridman (1:16:23.320)
it's like taxing me and then taking that money
Lex Fridman (1:16:25.060)
and dumping it in the ocean, just like destroying it.
Lex Fridman (1:16:27.820)
So there are a lot of things like that
Erik Brynjolfsson (1:16:29.500)
that economists, and I'm not,
Lex Fridman (1:16:32.520)
I'm not like doing anything radical here.
Erik Brynjolfsson (1:16:33.940)
Most, you know, good economists would,
Lex Fridman (1:16:36.680)
I probably agree with me point by point on these things.
Lex Fridman (1:16:39.440)
And we could do those things
Lex Fridman (1:16:41.000)
and our whole economy would become much more efficient.
Erik Brynjolfsson (1:16:43.760)
It'd become fairer, invest in R&D and research,
Lex Fridman (1:16:47.000)
which is close to a free lunch is what we have.
Erik Brynjolfsson (1:16:50.060)
My erstwhile MIT colleague, Bob Solla,
Lex Fridman (1:16:53.160)
got the Nobel Prize, not yesterday, but 30 years ago,
Erik Brynjolfsson (1:16:57.360)
for describing that most improvements
Lex Fridman (1:17:00.560)
in living standards come from tech progress.
Lex Fridman (1:17:02.880)
And Paul Romer later got a Nobel Prize
Lex Fridman (1:17:04.560)
for noting that investments in R&D and human capital
Erik Brynjolfsson (1:17:08.040)
can speed the rate of tech progress.
Lex Fridman (1:17:11.040)
So if we do that, then we'll be healthier and wealthier.
Erik Brynjolfsson (1:17:14.680)
Yeah, from an economics perspective,
Lex Fridman (1:17:16.200)
I remember taking an undergrad econ,
Erik Brynjolfsson (1:17:18.440)
you mentioned econ 101.
Lex Fridman (1:17:20.380)
It seemed from all the plots I saw
Erik Brynjolfsson (1:17:23.660)
that R&D is an obvious, as close to free lunch as we have,
Lex Fridman (1:17:29.040)
it seemed like obvious that we should do more research.
Erik Brynjolfsson (1:17:32.340)
It is.
Lex Fridman (1:17:33.180)
Like what, what, like, there's no.
Erik Brynjolfsson (1:17:36.620)
Well, we should do basic research.
Lex Fridman (1:17:38.000)
I mean, so let me just be clear.
Erik Brynjolfsson (1:17:39.440)
It'd be great if everybody did more research
Lex Fridman (1:17:41.420)
and I would make this issue
Erik Brynjolfsson (1:17:42.260)
between applied development versus basic research.
Lex Fridman (1:17:46.080)
So applied development, like, you know,
Lex Fridman (1:17:48.120)
how do we get this self driving car, you know,
Lex Fridman (1:17:52.120)
feature to work better in the Tesla?
Erik Brynjolfsson (1:17:53.960)
That's great for private companies
Lex Fridman (1:17:55.240)
because they can capture the value from that.
Erik Brynjolfsson (1:17:57.080)
If they make a better self driving car system,
Lex Fridman (1:17:59.700)
they can sell cars that are more valuable
Lex Fridman (1:18:02.240)
and then make money.
Lex Fridman (1:18:03.080)
So there's an incentive that there's not a big problem there
Lex Fridman (1:18:05.720)
and smart companies, Amazon, Tesla,
Lex Fridman (1:18:08.200)
and others are investing in it.
Erik Brynjolfsson (1:18:09.440)
The problem is with basic research,
Lex Fridman (1:18:11.260)
like coming up with core basic ideas,
Erik Brynjolfsson (1:18:14.420)
whether it's in nuclear fusion
Lex Fridman (1:18:16.120)
or artificial intelligence or biotech.
Erik Brynjolfsson (1:18:19.000)
There, if someone invents something,
Lex Fridman (1:18:21.640)
it's very hard for them to capture the benefits from it.
Erik Brynjolfsson (1:18:23.920)
It's shared by everybody, which is great in a way,
Lex Fridman (1:18:26.740)
but it means that they're not gonna have the incentives
Erik Brynjolfsson (1:18:28.640)
to put as much effort into it.
Lex Fridman (1:18:30.680)
There you need, it's a classic public good.
Erik Brynjolfsson (1:18:32.960)
There you need the government to be involved in it.
Lex Fridman (1:18:35.120)
And the US government used to be investing much more in R&D,
Lex Fridman (1:18:39.360)
but we have slashed that part of the government
Lex Fridman (1:18:42.940)
really foolishly and we're all poorer,
Erik Brynjolfsson (1:18:46.900)
significantly poorer as a result.
Lex Fridman (1:18:48.440)
Growth rates are down.
Erik Brynjolfsson (1:18:50.000)
We're not having the kind of scientific progress
Lex Fridman (1:18:51.680)
we used to have.
Erik Brynjolfsson (1:18:53.260)
It's been sort of a short term eating the seed corn,
Lex Fridman (1:18:57.800)
whatever metaphor you wanna use
Erik Brynjolfsson (1:19:00.120)
where people grab some money, put it in their pockets today,
Lex Fridman (1:19:03.320)
but five, 10, 20 years later,
Erik Brynjolfsson (1:19:07.120)
they're a lot poorer than they otherwise would have been.
Lex Fridman (1:19:10.140)
So we're living through a pandemic right now,
Erik Brynjolfsson (1:19:12.320)
globally in the United States.
Lex Fridman (1:19:16.580)
From an economics perspective,
Lex Fridman (1:19:18.840)
how do you think this pandemic will change the world?
Lex Fridman (1:19:23.040)
It's been remarkable.
Lex Fridman (1:19:24.640)
And it's horrible how many people have suffered,
Lex Fridman (1:19:27.760)
the amount of death, the economic destruction.
Erik Brynjolfsson (1:19:31.240)
It's also striking just the amount of change in work
Lex Fridman (1:19:34.300)
that I've seen.
Erik Brynjolfsson (1:19:35.840)
In the last 20 weeks, I've seen more change
Lex Fridman (1:19:38.440)
than there were in the previous 20 years.
Erik Brynjolfsson (1:19:41.200)
There's been nothing like it
Lex Fridman (1:19:42.400)
since probably the World War II mobilization
Erik Brynjolfsson (1:19:44.700)
in terms of reorganizing our economy.
Lex Fridman (1:19:47.040)
The most obvious one is the shift to remote work.
Lex Fridman (1:19:50.200)
And I and many other people stopped going into the office
Lex Fridman (1:19:54.280)
and teaching my students in person.
Erik Brynjolfsson (1:19:56.160)
I did a study on this with a bunch of colleagues
Lex Fridman (1:19:57.760)
at MIT and elsewhere.
Lex Fridman (1:19:59.180)
And what we found was that before the pandemic,
Lex Fridman (1:20:02.440)
in the beginning of 2020, about one in six,
Erik Brynjolfsson (1:20:05.400)
a little over 15% of Americans were working remotely.
Lex Fridman (1:20:09.840)
When the pandemic hit, that grew steadily and hit 50%,
Erik Brynjolfsson (1:20:13.560)
roughly half of Americans working at home.
Lex Fridman (1:20:16.080)
So a complete transformation.
Lex Fridman (1:20:17.840)
And of course, it wasn't even,
Lex Fridman (1:20:19.160)
it wasn't like everybody did it.
Erik Brynjolfsson (1:20:20.520)
If you're an information worker, professional,
Lex Fridman (1:20:22.760)
if you work mainly with data,
Erik Brynjolfsson (1:20:24.400)
then you're much more likely to work at home.
Lex Fridman (1:20:26.880)
If you're a manufacturing worker,
Erik Brynjolfsson (1:20:28.800)
working with other people or physical things,
Lex Fridman (1:20:32.320)
then it wasn't so easy to work at home.
Lex Fridman (1:20:34.520)
And instead, those people were much more likely
Lex Fridman (1:20:36.480)
to become laid off or unemployed.
Lex Fridman (1:20:39.280)
So it's been something that's had very disparate effects
Lex Fridman (1:20:41.840)
on different parts of the workforce.
Lex Fridman (1:20:44.520)
Do you think it's gonna be sticky in a sense
Lex Fridman (1:20:46.720)
that after vaccine comes out and the economy reopens,
Lex Fridman (1:20:51.060)
do you think remote work will continue?
Lex Fridman (1:20:55.180)
That's a great question.
Erik Brynjolfsson (1:20:57.080)
My hypothesis is yes, a lot of it will.
Lex Fridman (1:20:59.360)
Of course, some of it will go back,
Lex Fridman (1:21:00.800)
but a surprising amount of it will stay.
Lex Fridman (1:21:03.480)
I personally, for instance, I moved my seminars,
Erik Brynjolfsson (1:21:06.620)
my academic seminars to Zoom,
Lex Fridman (1:21:08.840)
and I was surprised how well it worked.
Lex Fridman (1:21:10.800)
So it works?
Lex Fridman (1:21:11.640)
Yeah, I mean, obviously we were able to reach
Erik Brynjolfsson (1:21:13.600)
a much broader audience.
Lex Fridman (1:21:14.760)
So we have people tuning in from Europe
Lex Fridman (1:21:16.600)
and other countries,
Lex Fridman (1:21:18.520)
just all over the United States for that matter.
Erik Brynjolfsson (1:21:20.320)
I also actually found that it would,
Lex Fridman (1:21:21.760)
in many ways, is more egalitarian.
Erik Brynjolfsson (1:21:23.520)
We use the chat feature and other tools,
Lex Fridman (1:21:25.920)
and grad students and others who might've been
Erik Brynjolfsson (1:21:27.600)
a little shy about speaking up,
Lex Fridman (1:21:29.400)
we now kind of have more of ability for lots of voices.
Lex Fridman (1:21:32.680)
And they're answering each other's questions,
Lex Fridman (1:21:34.360)
so you kind of get parallel.
Erik Brynjolfsson (1:21:35.960)
Like if someone had some question about some of the data
Lex Fridman (1:21:39.040)
or a reference or whatever,
Erik Brynjolfsson (1:21:40.660)
then someone else in the chat would answer it.
Lex Fridman (1:21:42.480)
And the whole thing just became like a higher bandwidth,
Erik Brynjolfsson (1:21:44.480)
higher quality thing.
Lex Fridman (1:21:46.600)
So I thought that was kind of interesting.
Erik Brynjolfsson (1:21:48.440)
I think a lot of people are discovering that these tools
Lex Fridman (1:21:51.280)
that thanks to technologists have been developed
Erik Brynjolfsson (1:21:54.480)
over the past decade,
Lex Fridman (1:21:56.440)
they're a lot more powerful than we thought.
Erik Brynjolfsson (1:21:57.920)
I mean, all the terrible things we've seen with COVID
Lex Fridman (1:22:00.120)
and the real failure of many of our institutions
Erik Brynjolfsson (1:22:03.400)
that I thought would work better.
Lex Fridman (1:22:04.960)
One area that's been a bright spot is our technologies.
Erik Brynjolfsson (1:22:09.420)
Bandwidth has held up pretty well,
Lex Fridman (1:22:11.840)
and all of our email and other tools
Erik Brynjolfsson (1:22:14.200)
have just scaled up kind of gracefully.
Lex Fridman (1:22:18.000)
So that's been a plus.
Erik Brynjolfsson (1:22:20.280)
Economists call this question
Lex Fridman (1:22:21.680)
of whether it'll go back a hysteresis.
Erik Brynjolfsson (1:22:23.920)
The question is like when you boil an egg
Lex Fridman (1:22:25.880)
after it gets cold again, it stays hard.
Lex Fridman (1:22:29.020)
And I think that we're gonna have a fair amount
Lex Fridman (1:22:30.860)
of hysteresis in the economy.
Erik Brynjolfsson (1:22:32.160)
We're gonna move to this new,
Lex Fridman (1:22:33.440)
we have moved to a new remote work system,
Lex Fridman (1:22:35.520)
and it's not gonna snap all the way back
Lex Fridman (1:22:37.260)
to where it was before.
Erik Brynjolfsson (1:22:38.720)
One of the things that worries me is that the people
Lex Fridman (1:22:44.160)
with lots of followers on Twitter and people with voices,
Erik Brynjolfsson (1:22:51.380)
people that can, voices that can be magnified by reporters
Lex Fridman (1:22:56.380)
and all that kind of stuff are the people
Erik Brynjolfsson (1:22:57.900)
that fall into this category
Lex Fridman (1:22:59.240)
that we were referring to just now
Erik Brynjolfsson (1:23:01.600)
where they can still function
Lex Fridman (1:23:03.000)
and be successful with remote work.
Lex Fridman (1:23:06.240)
And then there is a kind of quiet suffering
Lex Fridman (1:23:11.240)
of what feels like millions of people
Erik Brynjolfsson (1:23:14.800)
whose jobs are disturbed profoundly by this pandemic,
Lex Fridman (1:23:21.200)
but they don't have many followers on Twitter.
Lex Fridman (1:23:26.320)
What do we, and again, I apologize,
Lex Fridman (1:23:31.840)
but I've been reading the rise and fall of the Third Reich
Lex Fridman (1:23:35.840)
and there's a connection to the depression
Lex Fridman (1:23:38.080)
on the American side.
Erik Brynjolfsson (1:23:39.580)
There's a deep, complicated connection
Lex Fridman (1:23:42.320)
to how suffering can turn into forces
Erik Brynjolfsson (1:23:46.400)
that potentially change the world in destructive ways.
Lex Fridman (1:23:51.960)
So like it's something I worry about is like,
Lex Fridman (1:23:53.840)
what is this suffering going to materialize itself
Lex Fridman (1:23:56.600)
in five, 10 years?
Lex Fridman (1:23:58.080)
Is that something you worry about, think about?
Lex Fridman (1:24:01.020)
It's like the center of what I worry about.
Lex Fridman (1:24:03.320)
And let me break it down to two parts.
Lex Fridman (1:24:05.400)
There's a moral and ethical aspect to it.
Erik Brynjolfsson (1:24:07.280)
We need to relieve this suffering.
Lex Fridman (1:24:09.340)
I mean, I'm sure the values of, I think most Americans,
Erik Brynjolfsson (1:24:13.280)
we like to see shared prosperity
Lex Fridman (1:24:15.000)
or most people on the planet.
Lex Fridman (1:24:16.620)
And we would like to see people not falling behind
Lex Fridman (1:24:20.220)
and they have fallen behind, not just due to COVID,
Lex Fridman (1:24:23.080)
but in the previous couple of decades,
Lex Fridman (1:24:25.760)
median income has barely moved,
Erik Brynjolfsson (1:24:27.920)
depending on how you measure it.
Lex Fridman (1:24:29.900)
And the incomes of the top 1% have skyrocketed.
Lex Fridman (1:24:33.360)
And part of that is due to the ways technology has been used.
Lex Fridman (1:24:36.460)
Part of this been due to, frankly, our political system
Erik Brynjolfsson (1:24:38.840)
has continually shifted more wealth into those people
Lex Fridman (1:24:43.680)
who have the powerful interest.
Lex Fridman (1:24:45.120)
So there's just, I think, a moral imperative
Lex Fridman (1:24:48.720)
to do a better job.
Lex Fridman (1:24:49.800)
And ultimately, we're all gonna be wealthier
Lex Fridman (1:24:51.900)
if more people can contribute,
Erik Brynjolfsson (1:24:53.320)
more people have the wherewithal.
Lex Fridman (1:24:55.040)
But the second thing is that there's a real political risk.
Erik Brynjolfsson (1:24:58.640)
I'm not a political scientist,
Lex Fridman (1:24:59.960)
but you don't have to be one, I think,
Erik Brynjolfsson (1:25:02.560)
to see how a lot of people are really upset
Lex Fridman (1:25:05.660)
with they're getting a raw deal
Lex Fridman (1:25:07.380)
and they want to smash the system in different ways,
Lex Fridman (1:25:13.680)
in 2016 and 2018.
Lex Fridman (1:25:15.960)
And now I think there are a lot of people
Lex Fridman (1:25:18.280)
who are looking at the political system
Lex Fridman (1:25:19.600)
and they feel like it's not working for them
Lex Fridman (1:25:21.120)
and they just wanna do something radical.
Erik Brynjolfsson (1:25:24.720)
Unfortunately, demagogues have harnessed that
Lex Fridman (1:25:28.140)
in a way that is pretty destructive to the country.
Lex Fridman (1:25:33.140)
And an analogy I see is what happened with trade.
Lex Fridman (1:25:37.240)
Almost every economist thinks that free trade
Erik Brynjolfsson (1:25:39.440)
is a good thing, that when two people voluntarily exchange
Lex Fridman (1:25:42.440)
almost by definition, they're both better off
Erik Brynjolfsson (1:25:44.940)
if it's voluntary.
Lex Fridman (1:25:47.320)
And so generally, trade is a good thing.
Lex Fridman (1:25:49.800)
But they also recognize that trade can lead
Lex Fridman (1:25:52.480)
to uneven effects, that there can be winners and losers
Erik Brynjolfsson (1:25:56.260)
in some of the people who didn't have the skills
Lex Fridman (1:25:59.280)
to compete with somebody else or didn't have other assets.
Lex Fridman (1:26:02.880)
And so trade can shift prices
Lex Fridman (1:26:04.920)
in ways that are averse to some people.
Lex Fridman (1:26:08.460)
So there's a formula that economists have,
Lex Fridman (1:26:11.340)
which is that you have free trade,
Lex Fridman (1:26:13.440)
but then you compensate the people who are hurt
Lex Fridman (1:26:15.920)
and free trade makes the pie bigger.
Lex Fridman (1:26:18.400)
And since the pie is bigger,
Lex Fridman (1:26:19.460)
it's possible for everyone to be better off.
Erik Brynjolfsson (1:26:21.920)
You can make the winners better off,
Lex Fridman (1:26:23.200)
but you can also compensate those who don't win.
Lex Fridman (1:26:25.440)
And so they end up being better off as well.
Lex Fridman (1:26:28.460)
What happened was that we didn't fulfill that promise.
Erik Brynjolfsson (1:26:33.160)
We did have some more increased free trade
Lex Fridman (1:26:36.040)
in the 80s and 90s, but we didn't compensate the people
Erik Brynjolfsson (1:26:39.480)
who were hurt.
Lex Fridman (1:26:40.640)
And so they felt like the people in power
Erik Brynjolfsson (1:26:43.800)
reneged on the bargain, and I think they did.
Lex Fridman (1:26:45.900)
And so then there's a backlash against trade.
Lex Fridman (1:26:48.760)
And now both political parties,
Lex Fridman (1:26:50.840)
but especially Trump and company,
Erik Brynjolfsson (1:26:53.640)
have really pushed back against free trade.
Lex Fridman (1:26:58.200)
Ultimately, that's bad for the country.
Erik Brynjolfsson (1:27:00.680)
Ultimately, that's bad for living standards.
Lex Fridman (1:27:02.720)
But in a way I can understand
Erik Brynjolfsson (1:27:04.400)
that people felt they were betrayed.
Lex Fridman (1:27:07.080)
Technology has a lot of similar characteristics.
Erik Brynjolfsson (1:27:10.680)
Technology can make us all better off.
Lex Fridman (1:27:14.920)
It makes the pie bigger.
Erik Brynjolfsson (1:27:16.120)
It creates wealth and health, but it can also be uneven.
Lex Fridman (1:27:18.920)
Not everyone automatically benefits.
Erik Brynjolfsson (1:27:21.280)
It's possible for some people,
Lex Fridman (1:27:22.880)
even a majority of people to get left behind
Erik Brynjolfsson (1:27:25.080)
while a small group benefits.
Lex Fridman (1:27:28.200)
What most economists would say,
Erik Brynjolfsson (1:27:29.560)
well, let's make the pie bigger,
Lex Fridman (1:27:30.880)
but let's make sure we adjust the system
Lex Fridman (1:27:33.000)
so we compensate the people who are hurt.
Lex Fridman (1:27:35.200)
And since the pie is bigger,
Erik Brynjolfsson (1:27:36.920)
we can make the rich richer,
Lex Fridman (1:27:38.000)
we can make the middle class richer,
Erik Brynjolfsson (1:27:39.200)
we can make the poor richer.
Lex Fridman (1:27:40.980)
Mathematically, everyone could be better off.
Lex Fridman (1:27:43.640)
But again, we're not doing that.
Lex Fridman (1:27:45.400)
And again, people are saying this isn't working for us.
Lex Fridman (1:27:48.940)
And again, instead of fixing the distribution,
Lex Fridman (1:27:52.540)
a lot of people are beginning to say,
Erik Brynjolfsson (1:27:54.280)
hey, technology sucks, we've got to stop it.
Lex Fridman (1:27:57.280)
Let's throw rocks at the Google bus.
Erik Brynjolfsson (1:27:59.040)
Let's blow it up.
Lex Fridman (1:27:59.980)
Let's blow it up.
Lex Fridman (1:28:01.240)
And there were the Luddites almost exactly 200 years ago
Lex Fridman (1:28:04.760)
who smashed the looms and the spinning machines
Erik Brynjolfsson (1:28:08.040)
because they felt like those machines weren't helping them.
Lex Fridman (1:28:11.320)
We have a real imperative,
Erik Brynjolfsson (1:28:12.720)
not just to do the morally right thing,
Lex Fridman (1:28:14.700)
but to do the thing that is gonna save the country,
Erik Brynjolfsson (1:28:17.520)
which is make sure that we create
Lex Fridman (1:28:19.440)
not just prosperity, but shared prosperity.
Lex Fridman (1:28:22.680)
So you've been at MIT for over 30 years, I think.
Lex Fridman (1:28:27.600)
Don't tell anyone how old I am.
Erik Brynjolfsson (1:28:28.440)
Yeah, no, that's true, that's true.
Lex Fridman (1:28:30.280)
And you're now moved to Stanford.
Erik Brynjolfsson (1:28:34.000)
I'm gonna try not to say anything
Lex Fridman (1:28:37.240)
about how great MIT is.
Lex Fridman (1:28:39.760)
What's that move been like?
Lex Fridman (1:28:41.520)
What, it's East Coast to West Coast?
Erik Brynjolfsson (1:28:44.960)
Well, MIT is great.
Lex Fridman (1:28:46.160)
MIT has been very good to me.
Erik Brynjolfsson (1:28:48.080)
It continues to be very good to me.
Lex Fridman (1:28:49.560)
It's an amazing place.
Erik Brynjolfsson (1:28:51.440)
I continue to have so many amazing friends
Lex Fridman (1:28:53.200)
and colleagues there.
Erik Brynjolfsson (1:28:54.600)
I'm very fortunate to have been able
Lex Fridman (1:28:56.120)
to spend a lot of time at MIT.
Erik Brynjolfsson (1:28:58.480)
Stanford's also amazing.
Lex Fridman (1:29:00.200)
And part of what attracted me out here
Erik Brynjolfsson (1:29:01.980)
was not just the weather, but also Silicon Valley,
Lex Fridman (1:29:04.960)
let's face it, is really more of the epicenter
Erik Brynjolfsson (1:29:07.360)
of the technological revolution.
Lex Fridman (1:29:09.000)
And I wanna be close to the people
Erik Brynjolfsson (1:29:10.400)
who are inventing AI and elsewhere.
Lex Fridman (1:29:12.320)
A lot of it is being invested at MIT for that matter
Erik Brynjolfsson (1:29:14.920)
in Europe and China and elsewhere, in Nia.
Lex Fridman (1:29:18.940)
But being a little closer to some of the key technologists
Erik Brynjolfsson (1:29:23.800)
was something that was important to me.
Lex Fridman (1:29:25.920)
And it may be shallow,
Lex Fridman (1:29:28.600)
but I also do enjoy the good weather.
Lex Fridman (1:29:30.180)
And I felt a little ripped off
Erik Brynjolfsson (1:29:33.120)
when I came here a couple of months ago.
Lex Fridman (1:29:35.040)
And immediately there are the fires
Lex Fridman (1:29:36.640)
and my eyes were burning, the sky was orange
Lex Fridman (1:29:39.840)
and there's the heat waves.
Lex Fridman (1:29:41.320)
And so it wasn't exactly what I've been promised,
Lex Fridman (1:29:44.460)
but fingers crossed it'll get back to better.
Lex Fridman (1:29:47.960)
But maybe on a brief aside,
Lex Fridman (1:29:50.720)
there's been some criticism of academia
Lex Fridman (1:29:52.720)
and universities and different avenues.
Lex Fridman (1:29:55.760)
And I, as a person who's gotten to enjoy universities
Erik Brynjolfsson (1:30:00.760)
from the pure playground of ideas that it can be,
Lex Fridman (1:30:06.380)
always kind of try to find the words
Erik Brynjolfsson (1:30:08.840)
to tell people that these are magical places.
Lex Fridman (1:30:13.160)
Is there something that you can speak to
Lex Fridman (1:30:17.000)
that is beautiful or powerful about universities?
Lex Fridman (1:30:22.440)
Well, sure.
Erik Brynjolfsson (1:30:23.280)
I mean, first off, I mean,
Lex Fridman (1:30:24.500)
economists have this concept called revealed preference.
Erik Brynjolfsson (1:30:26.660)
You can ask people what they say
Lex Fridman (1:30:28.300)
or you can watch what they do.
Lex Fridman (1:30:29.940)
And so obviously by reveal preferences, I love academia.
Lex Fridman (1:30:33.960)
I could be doing lots of other things,
Lex Fridman (1:30:35.540)
but it's something I enjoy a lot.
Lex Fridman (1:30:37.600)
And I think the word magical is exactly right.
Erik Brynjolfsson (1:30:39.640)
At least it is for me.
Lex Fridman (1:30:41.480)
I do what I love, you know,
Erik Brynjolfsson (1:30:43.120)
hopefully my Dean won't be listening,
Lex Fridman (1:30:44.320)
but I would do this for free.
Erik Brynjolfsson (1:30:45.640)
You know, it's just what I like to do.
Lex Fridman (1:30:49.060)
I like to do research.
Erik Brynjolfsson (1:30:50.160)
I love to have conversations like this with you
Lex Fridman (1:30:51.840)
and with my students, with my fellow colleagues.
Erik Brynjolfsson (1:30:53.740)
I love being around the smartest people I can find
Lex Fridman (1:30:55.760)
and learning something from them
Lex Fridman (1:30:57.220)
and having them challenge me.
Lex Fridman (1:30:58.640)
And that just gives me joy.
Lex Fridman (1:31:02.480)
And every day I find something new and exciting to work on.
Lex Fridman (1:31:05.500)
And a university environment is really filled
Erik Brynjolfsson (1:31:08.040)
with other people who feel that way.
Lex Fridman (1:31:09.820)
And so I feel very fortunate to be part of it.
Lex Fridman (1:31:12.960)
And I'm lucky that I'm in a society
Lex Fridman (1:31:14.840)
where I can actually get paid for it
Lex Fridman (1:31:16.200)
and put food on the table
Lex Fridman (1:31:17.240)
while doing the stuff that I really love.
Lex Fridman (1:31:19.260)
And I hope someday everybody can have jobs
Lex Fridman (1:31:21.560)
that are like that.
Lex Fridman (1:31:22.800)
And I appreciate that it's not necessarily easy
Lex Fridman (1:31:25.340)
for everybody to have a job that they both love
Lex Fridman (1:31:27.400)
and also they get paid for.
Lex Fridman (1:31:30.660)
So there are things that don't go well in academia,
Lex Fridman (1:31:34.000)
but by and large, I think it's a kind of, you know,
Lex Fridman (1:31:36.000)
kinder, gentler version of a lot of the world.
Erik Brynjolfsson (1:31:37.960)
You know, we sort of cut each other a little slack
Lex Fridman (1:31:41.280)
on things like, you know, on just a lot of things.
Erik Brynjolfsson (1:31:45.800)
You know, of course there's harsh debates
Lex Fridman (1:31:48.320)
and discussions about things
Lex Fridman (1:31:49.900)
and some petty politics here and there.
Lex Fridman (1:31:52.060)
I personally, I try to stay away
Erik Brynjolfsson (1:31:53.520)
from most of that sort of politics.
Lex Fridman (1:31:55.600)
It's not my thing.
Lex Fridman (1:31:56.560)
And so it doesn't affect me most of the time,
Lex Fridman (1:31:58.320)
sometimes a little bit, maybe.
Erik Brynjolfsson (1:32:00.480)
But, you know, being able to pull together something,
Lex Fridman (1:32:03.200)
we have the digital economy lab.
Erik Brynjolfsson (1:32:04.860)
We've got all these brilliant grad students
Lex Fridman (1:32:07.480)
and undergraduates and postdocs
Erik Brynjolfsson (1:32:09.280)
that are just doing stuff that I learned from.
Lex Fridman (1:32:12.320)
And every one of them has some aspect
Erik Brynjolfsson (1:32:14.760)
of what they're doing that's just,
Lex Fridman (1:32:16.640)
I couldn't even understand.
Erik Brynjolfsson (1:32:17.600)
It's like way, way more brilliant.
Lex Fridman (1:32:19.340)
And that's really, to me, actually I really enjoy that,
Erik Brynjolfsson (1:32:23.040)
being in a room with lots of other smart people.
Lex Fridman (1:32:25.120)
And Stanford has made it very easy to attract,
Erik Brynjolfsson (1:32:29.440)
you know, those people.
Lex Fridman (1:32:31.260)
I just, you know, say I'm gonna do a seminar, whatever,
Lex Fridman (1:32:33.680)
and the people come, they come and wanna work with me.
Lex Fridman (1:32:36.820)
We get funding, we get data sets,
Lex Fridman (1:32:38.880)
and it's come together real nicely.
Lex Fridman (1:32:41.440)
And the rest is just fun.
Erik Brynjolfsson (1:32:44.220)
It's fun, yeah.
Lex Fridman (1:32:45.840)
And we feel like we're working on important problems,
Erik Brynjolfsson (1:32:47.480)
you know, and we're doing things that, you know,
Lex Fridman (1:32:50.320)
I think are first order in terms of what's important
Erik Brynjolfsson (1:32:53.680)
in the world, and that's very satisfying to me.
Lex Fridman (1:32:56.320)
Maybe a bit of a fun question.
Lex Fridman (1:32:58.080)
What three books, technical, fiction, philosophical,
Lex Fridman (1:33:02.180)
you've enjoyed, had a big, big impact in your life?
Erik Brynjolfsson (1:33:07.380)
Well, I guess I go back to like my teen years,
Lex Fridman (1:33:09.980)
and, you know, I read Sid Arthur,
Erik Brynjolfsson (1:33:12.300)
which is a philosophical book,
Lex Fridman (1:33:13.420)
and kind of helps keep me centered.
Erik Brynjolfsson (1:33:15.260)
By Herman Hesse.
Lex Fridman (1:33:16.180)
Yeah, by Herman Hesse, exactly.
Erik Brynjolfsson (1:33:17.340)
Don't get too wrapped up in material things
Lex Fridman (1:33:20.380)
or other things, and just sort of, you know,
Erik Brynjolfsson (1:33:21.980)
try to find peace on things.
Lex Fridman (1:33:24.780)
A book that actually influenced me a lot
Erik Brynjolfsson (1:33:26.340)
in terms of my career was called
Lex Fridman (1:33:27.620)
The Worldly Philosophers by Robert Halbrenner.
Erik Brynjolfsson (1:33:30.460)
It's actually about economists.
Lex Fridman (1:33:31.660)
It goes through a series of different,
Erik Brynjolfsson (1:33:33.500)
it's written in a very lively form,
Lex Fridman (1:33:34.900)
and it probably sounds boring,
Lex Fridman (1:33:36.220)
but it did describe whether it's Adam Smith
Lex Fridman (1:33:38.820)
or Karl Marx or John Maynard Keynes,
Lex Fridman (1:33:40.820)
and each of them sort of what their key insights were,
Lex Fridman (1:33:43.340)
but also kind of their personalities,
Lex Fridman (1:33:45.340)
and I think that's one of the reasons
Lex Fridman (1:33:46.520)
I became an economist was just understanding
Lex Fridman (1:33:50.600)
how they grapple with the big questions of the world.
Lex Fridman (1:33:53.100)
So would you recommend it as a good whirlwind overview
Lex Fridman (1:33:56.340)
of the history of economics?
Lex Fridman (1:33:57.540)
Yeah, yeah, I think that's exactly right.
Erik Brynjolfsson (1:33:59.060)
It kind of takes you through the different things,
Lex Fridman (1:34:00.940)
and so you can understand how they reach,
Erik Brynjolfsson (1:34:04.020)
thinking some of the strengths and weaknesses.
Lex Fridman (1:34:06.380)
I mean, it probably is a little out of date now.
Erik Brynjolfsson (1:34:07.900)
It needs to be updated a bit,
Lex Fridman (1:34:08.980)
but you could at least look through
Erik Brynjolfsson (1:34:10.380)
the first couple hundred years of economics,
Lex Fridman (1:34:12.940)
which is not a bad place to start.
Erik Brynjolfsson (1:34:15.020)
More recently, I mean, a book I really enjoyed
Lex Fridman (1:34:17.580)
is by my friend and colleague, Max Tegmark,
Erik Brynjolfsson (1:34:20.260)
called Life 3.0.
Lex Fridman (1:34:21.340)
You should have him on your podcast if you haven't already.
Erik Brynjolfsson (1:34:23.260)
He was episode number one.
Lex Fridman (1:34:25.460)
Oh my God.
Lex Fridman (1:34:26.500)
And he's back, he'll be back, he'll be back soon.
Lex Fridman (1:34:30.220)
Yeah, no, he's terrific.
Erik Brynjolfsson (1:34:31.460)
I love the way his brain works,
Lex Fridman (1:34:33.460)
and he makes you think about profound things.
Erik Brynjolfsson (1:34:35.780)
He's got such a joyful approach to life,
Lex Fridman (1:34:38.540)
and so that's been a great book,
Lex Fridman (1:34:41.060)
and I learn a lot from it, I think everybody,
Lex Fridman (1:34:43.180)
but he explains it in a way, even though he's so brilliant,
Erik Brynjolfsson (1:34:45.580)
that everyone can understand, that I can understand.
Lex Fridman (1:34:50.020)
That's three, but let me mention maybe one or two others.
Erik Brynjolfsson (1:34:52.920)
I mean, I recently read More From Less
Lex Fridman (1:34:55.340)
by my sometimes coauthor, Andrew McAfee.
Erik Brynjolfsson (1:34:58.620)
It made me optimistic about how we can continue
Lex Fridman (1:35:01.940)
to have rising living standards
Erik Brynjolfsson (1:35:04.580)
while living more lightly on the planet.
Lex Fridman (1:35:06.140)
In fact, because of higher living standards,
Erik Brynjolfsson (1:35:07.860)
because of technology,
Lex Fridman (1:35:09.140)
because of digitization that I mentioned,
Erik Brynjolfsson (1:35:11.500)
we don't have to have as big an impact on the planet,
Lex Fridman (1:35:13.580)
and that's a great story to tell,
Lex Fridman (1:35:15.740)
and he documents it very carefully.
Lex Fridman (1:35:19.740)
You know, a personal kind of self help book
Erik Brynjolfsson (1:35:21.380)
that I found kind of useful, People, is Atomic Habits.
Lex Fridman (1:35:24.140)
I think it's, what's his name, James Clear.
Erik Brynjolfsson (1:35:26.180)
Yeah, James Clear.
Lex Fridman (1:35:27.540)
He's just, yeah, it's a good name,
Erik Brynjolfsson (1:35:29.100)
because he writes very clearly,
Lex Fridman (1:35:30.460)
and you know, most of the sentences I read in that book,
Erik Brynjolfsson (1:35:33.620)
I was like, yeah, I know that,
Lex Fridman (1:35:34.500)
but it just really helps to have somebody like remind you
Lex Fridman (1:35:37.220)
and tell you and kind of just reinforce it, and it's helpful.
Lex Fridman (1:35:40.860)
So build habits in your life that you hope to have,
Erik Brynjolfsson (1:35:45.020)
that have a positive impact,
Lex Fridman (1:35:46.140)
and don't have to make it big things.
Erik Brynjolfsson (1:35:48.100)
It could be just tiny little.
Lex Fridman (1:35:49.220)
Exactly, I mean, the word atomic,
Erik Brynjolfsson (1:35:50.720)
it's a little bit of a pun, I think he says.
Lex Fridman (1:35:52.540)
You know, one, atomic means they're really small.
Erik Brynjolfsson (1:35:54.020)
You take these little things, but also like atomic power,
Lex Fridman (1:35:56.860)
can have like, you know, big impact.
Erik Brynjolfsson (1:35:59.460)
That's funny, yeah.
Lex Fridman (1:36:01.460)
The biggest ridiculous question,
Erik Brynjolfsson (1:36:04.180)
especially to ask an economist, but also a human being,
Lex Fridman (1:36:06.860)
what's the meaning of life?
Erik Brynjolfsson (1:36:08.260)
I hope you've gotten the answer to that from somebody else.
Lex Fridman (1:36:11.460)
I think we're all still working on that one, but what is it?
Erik Brynjolfsson (1:36:14.740)
You know, I actually learned a lot from my son, Luke,
Lex Fridman (1:36:18.120)
and he's 19 now, but he's always loved philosophy,
Lex Fridman (1:36:22.100)
and he reads way more sophisticated philosophy than I do.
Lex Fridman (1:36:24.900)
I went and took him to Oxford,
Lex Fridman (1:36:25.860)
and he spent the whole time like pulling
Lex Fridman (1:36:27.060)
all these obscure books down and reading them.
Lex Fridman (1:36:29.020)
And a couple of years ago, we had this argument,
Lex Fridman (1:36:32.600)
and he was trying to convince me that hedonism
Erik Brynjolfsson (1:36:34.500)
was the ultimate, you know, meaning of life,
Lex Fridman (1:36:37.480)
just pleasure seeking, and...
Lex Fridman (1:36:40.380)
Well, how old was he at the time?
Lex Fridman (1:36:41.580)
17, so...
Erik Brynjolfsson (1:36:42.420)
Okay.
Lex Fridman (1:36:43.260)
But he made a really good like intellectual argument
Erik Brynjolfsson (1:36:46.700)
for it too, and you know,
Lex Fridman (1:36:47.540)
but you know, it just didn't strike me as right.
Lex Fridman (1:36:50.180)
And I think that, you know, while I am kind of a utilitarian,
Lex Fridman (1:36:54.540)
like, you know, I do think we should do the grace,
Erik Brynjolfsson (1:36:55.940)
good for the grace number, that's just too shallow.
Lex Fridman (1:36:58.740)
And I think I've convinced myself that real happiness
Erik Brynjolfsson (1:37:02.820)
doesn't come from seeking pleasure.
Lex Fridman (1:37:04.260)
It's kind of a little, it's ironic.
Erik Brynjolfsson (1:37:05.700)
Like if you really focus on being happy,
Lex Fridman (1:37:07.700)
I think it doesn't work.
Erik Brynjolfsson (1:37:09.740)
You gotta like be doing something bigger.
Lex Fridman (1:37:12.420)
I think the analogy I sometimes use is, you know,
Erik Brynjolfsson (1:37:14.900)
when you look at a dim star in the sky,
Lex Fridman (1:37:17.580)
if you look right at it, it kind of disappears,
Lex Fridman (1:37:19.460)
but you have to look a little to the side,
Lex Fridman (1:37:20.740)
and then the parts of your retina
Erik Brynjolfsson (1:37:23.180)
that are better at absorbing light,
Lex Fridman (1:37:24.940)
you know, can pick it up better.
Erik Brynjolfsson (1:37:26.340)
It's the same thing with happiness.
Lex Fridman (1:37:27.420)
I think you need to sort of find something, other goal,
Erik Brynjolfsson (1:37:32.500)
something, some meaning in life,
Lex Fridman (1:37:33.980)
and that ultimately makes you happier
Erik Brynjolfsson (1:37:36.180)
than if you go squarely at just pleasure.
Lex Fridman (1:37:39.060)
And so for me, you know, the kind of research I do
Erik Brynjolfsson (1:37:42.260)
that I think is trying to change the world,
Lex Fridman (1:37:44.220)
make the world a better place,
Lex Fridman (1:37:46.140)
and I'm not like an evolutionary psychologist,
Lex Fridman (1:37:47.980)
but my guess is that our brains are wired,
Erik Brynjolfsson (1:37:50.860)
not just for pleasure, but we're social animals,
Lex Fridman (1:37:53.860)
and we're wired to like help others.
Lex Fridman (1:37:57.220)
And ultimately, you know,
Lex Fridman (1:37:58.860)
that's something that's really deeply rooted in our psyche.
Lex Fridman (1:38:02.060)
And if we do help others, if we do,
Lex Fridman (1:38:04.500)
or at least feel like we're helping others,
Erik Brynjolfsson (1:38:06.660)
you know, our reward systems kick in,
Lex Fridman (1:38:08.220)
and we end up being more deeply satisfied
Erik Brynjolfsson (1:38:10.460)
than if we just do something selfish and shallow.
Lex Fridman (1:38:13.620)
Beautifully put.
Erik Brynjolfsson (1:38:14.460)
I don't think there's a better way to end it, Eric.
Lex Fridman (1:38:16.980)
You were one of the people when I first showed up at MIT,
Erik Brynjolfsson (1:38:20.500)
that made me proud to be at MIT.
Lex Fridman (1:38:22.420)
So it's so sad that you're now at Stanford,
Lex Fridman (1:38:24.540)
but I'm sure you'll do wonderful things at Stanford as well.
Lex Fridman (1:38:28.980)
I can't wait till future books,
Lex Fridman (1:38:30.900)
and people should definitely read your other books.
Lex Fridman (1:38:32.260)
Well, thank you so much.
Lex Fridman (1:38:33.180)
And I think we're all part of the invisible college,
Lex Fridman (1:38:35.580)
as we call it.
Erik Brynjolfsson (1:38:36.420)
You know, we're all part of this intellectual
Lex Fridman (1:38:38.700)
and human community where we all can learn from each other.
Erik Brynjolfsson (1:38:41.660)
It doesn't really matter physically
Lex Fridman (1:38:43.100)
where we are so much anymore.
Erik Brynjolfsson (1:38:44.860)
Beautiful.
Lex Fridman (1:38:45.700)
Thanks for talking today.
Erik Brynjolfsson (1:38:46.540)
My pleasure.
Lex Fridman (1:38:48.060)
Thanks for listening to this conversation
Erik Brynjolfsson (1:38:49.460)
with Eric Brynjolfsson.
Lex Fridman (1:38:50.860)
And thank you to our sponsors.
Erik Brynjolfsson (1:38:52.620)
Vincero Watches, the maker of classy,
Lex Fridman (1:38:55.060)
well performing watches.
Erik Brynjolfsson (1:38:56.860)
Fort Sigmatic, the maker of delicious mushroom coffee.
Lex Fridman (1:39:00.060)
ExpressVPN, the VPN I've used for many years
Erik Brynjolfsson (1:39:03.140)
to protect my privacy on the internet.
Lex Fridman (1:39:05.260)
And CashApp, the app I use to send money to friends.
Erik Brynjolfsson (1:39:09.140)
Please check out these sponsors in the description
Lex Fridman (1:39:11.180)
to get a discount and to support this podcast.
Erik Brynjolfsson (1:39:14.900)
If you enjoy this thing, subscribe on YouTube.
Lex Fridman (1:39:17.280)
Review it with five stars on Apple Podcast,
Erik Brynjolfsson (1:39:19.500)
follow on Spotify, support on Patreon,
Lex Fridman (1:39:22.100)
or connect with me on Twitter at Lex Friedman.
Lex Fridman (1:39:25.380)
And now, let me leave you with some words
Lex Fridman (1:39:27.700)
from Albert Einstein.
Erik Brynjolfsson (1:39:29.980)
It has become appallingly obvious
Lex Fridman (1:39:32.860)
that our technology has exceeded our humanity.
Erik Brynjolfsson (1:39:36.600)
Thank you for listening and hope to see you next time.
Lex Fridman (20:01.960)
a public, like anyone from the public
Erik Brynjolfsson (20:05.480)
in the Phoenix, Arizona to, you know,
Lex Fridman (20:12.160)
you can get a ride in a Waymo car
Erik Brynjolfsson (20:14.360)
with no person, no driver.
Lex Fridman (20:16.120)
Oh, they've taken away the safety driver?
Erik Brynjolfsson (20:17.640)
Oh yeah, for a while now there's been no safety driver.
Lex Fridman (20:21.080)
Okay, because I mean, I've been following that one
Erik Brynjolfsson (20:22.760)
in particular, but I thought it was kind of funny
Lex Fridman (20:24.840)
about a year ago when they had the safety driver
Lex Fridman (20:26.960)
and then they added a second safety driver
Lex Fridman (20:28.400)
because the first safety driver would fall asleep.
Erik Brynjolfsson (20:30.880)
It's like, I'm not sure they're going
Lex Fridman (20:32.120)
in the right direction with that.
Erik Brynjolfsson (20:33.400)
No, they've Waymo in particular
Lex Fridman (20:38.200)
done a really good job of that.
Erik Brynjolfsson (20:39.480)
They actually have a very interesting infrastructure
Lex Fridman (20:44.360)
of remote like observation.
Lex Fridman (20:47.480)
So they're not controlling the vehicles remotely,
Lex Fridman (20:49.840)
but they're able to, it's like a customer service.
Erik Brynjolfsson (20:52.440)
They can anytime tune into the car.
Lex Fridman (20:55.160)
I bet they can probably remotely control it as well,
Lex Fridman (20:58.160)
but that's officially not the function that they use.
Lex Fridman (21:00.920)
Yeah, I can see that being really,
Erik Brynjolfsson (21:02.760)
because I think the thing that's proven harder
Lex Fridman (21:06.280)
than maybe some of the early people expected
Erik Brynjolfsson (21:08.040)
was there's a long tail of weird exceptions.
Lex Fridman (21:10.840)
So you can deal with 90, 99, 99.99% of the cases,
Lex Fridman (21:15.440)
but then there's something that just never been seen before
Lex Fridman (21:17.720)
in the training data.
Lex Fridman (21:18.840)
And humans more or less can work around that.
Lex Fridman (21:21.080)
Although let me be clear and note,
Erik Brynjolfsson (21:22.840)
there are about 30,000 human fatalities
Lex Fridman (21:25.640)
just in the United States and maybe a million worldwide.
Lex Fridman (21:28.360)
So they're far from perfect.
Lex Fridman (21:30.000)
But I think people have higher expectations of machines.
Erik Brynjolfsson (21:33.440)
They wouldn't tolerate that level of death
Lex Fridman (21:36.280)
and damage from a machine.
Lex Fridman (21:40.000)
And so we have to do a lot better
Lex Fridman (21:41.800)
at dealing with those edge cases.
Lex Fridman (21:43.480)
And also the tricky thing that if I have a criticism
Lex Fridman (21:46.960)
for the Waymo folks, there's such a huge focus on safety
Erik Brynjolfsson (21:51.800)
where people don't talk enough about creating products
Lex Fridman (21:55.160)
that people, that customers love,
Erik Brynjolfsson (21:57.240)
that human beings love using.
Lex Fridman (22:00.680)
It's very easy to create a thing that's safe
Erik Brynjolfsson (22:03.320)
at the extremes, but then nobody wants to get into it.
Lex Fridman (22:06.880)
Yeah, well, back to Elon, I think one of,
Erik Brynjolfsson (22:09.640)
part of his genius was with the electric cars.
Lex Fridman (22:11.440)
Before he came along, electric cars were all kind of
Erik Brynjolfsson (22:13.960)
underpowered, really light,
Lex Fridman (22:15.680)
and there were sort of wimpy cars that weren't fun.
Lex Fridman (22:20.680)
And the first thing he did was he made a roadster
Lex Fridman (22:23.640)
that went zero to 60 faster than just about any other car
Lex Fridman (22:27.440)
and went the other end.
Lex Fridman (22:28.480)
And I think that was a really wise marketing move
Erik Brynjolfsson (22:30.800)
as well as a wise technology move.
Lex Fridman (22:33.160)
Yeah, it's difficult to figure out
Lex Fridman (22:34.840)
what the right marketing move is for AI systems.
Lex Fridman (22:37.960)
That's always been, I think it requires guts and risk taking
Erik Brynjolfsson (22:42.960)
which is what Elon practices.
Lex Fridman (22:46.320)
I mean, to the chagrin of perhaps investors or whatever,
Lex Fridman (22:50.480)
but it also requires rethinking what you're doing.
Lex Fridman (22:54.320)
I think way too many people are unimaginative,
Erik Brynjolfsson (22:57.520)
intellectually lazy, and when they take AI,
Lex Fridman (22:59.760)
they basically say, what are we doing now?
Lex Fridman (23:01.640)
How can we make a machine do the same thing?
Lex Fridman (23:04.000)
Maybe we'll save some costs, we'll have less labor.
Lex Fridman (23:06.720)
And yeah, it's not necessarily the worst thing
Lex Fridman (23:08.560)
in the world to do, but it's really not leading
Erik Brynjolfsson (23:10.640)
to a quantum change in the way you do things.
Lex Fridman (23:12.840)
When Jeff Bezos said, hey, we're gonna use the internet
Erik Brynjolfsson (23:16.640)
to change how bookstores work and we're gonna use technology,
Lex Fridman (23:19.320)
he didn't go and say, okay, let's put a robot cashier
Erik Brynjolfsson (23:22.680)
where the human cashier is and leave everything else alone.
Lex Fridman (23:25.640)
That would have been a very lame way to automate a bookstore.
Erik Brynjolfsson (23:28.360)
He's like went from soup to nuts and let's just rethink it.
Lex Fridman (23:31.400)
We get rid of the physical bookstore.
Erik Brynjolfsson (23:33.040)
We have a warehouse, we have delivery,
Lex Fridman (23:34.520)
we have people order on a screen
Lex Fridman (23:36.360)
and everything was reinvented.
Lex Fridman (23:38.480)
And that's been the story
Erik Brynjolfsson (23:39.720)
of these general purpose technologies all through history.
Lex Fridman (23:43.560)
And in my books, I write about like electricity
Lex Fridman (23:46.320)
and how for 30 years, there was almost no productivity gain
Lex Fridman (23:50.360)
from the electrification of factories a century ago.
Erik Brynjolfsson (23:53.040)
Now it's not because electricity
Lex Fridman (23:54.160)
is a wimpy useless technology.
Erik Brynjolfsson (23:55.800)
We all know how awesome electricity is.
Lex Fridman (23:57.560)
It's cause at first,
Erik Brynjolfsson (23:58.400)
they really didn't rethink the factories.
Lex Fridman (24:00.560)
It was only after they reinvented them
Lex Fridman (24:02.160)
and we describe how in the book,
Lex Fridman (24:04.040)
then you suddenly got a doubling and tripling
Erik Brynjolfsson (24:05.920)
of productivity growth.
Lex Fridman (24:07.560)
But it's the combination of the technology
Erik Brynjolfsson (24:09.920)
with the new business models, new business organization.
Lex Fridman (24:12.960)
That just takes a long time
Lex Fridman (24:14.000)
and it takes more creativity than most people have.
Lex Fridman (24:16.920)
Can you maybe linger on electricity?
Erik Brynjolfsson (24:19.000)
Cause that's a fun one.
Lex Fridman (24:20.080)
Yeah, well, sure, I'll tell you what happened.
Erik Brynjolfsson (24:22.480)
Before electricity, there were basically steam engines
Lex Fridman (24:25.760)
or sometimes water wheels and to power the machinery,
Erik Brynjolfsson (24:28.400)
you had to have pulleys and crankshafts
Lex Fridman (24:30.560)
and you really can't make them too long
Erik Brynjolfsson (24:32.120)
cause they'll break the torsion.
Lex Fridman (24:34.040)
So all the equipment was kind of clustered
Erik Brynjolfsson (24:35.960)
around this one giant steam engine.
Lex Fridman (24:37.960)
You can't make small steam engines either
Erik Brynjolfsson (24:39.480)
cause of thermodynamics.
Lex Fridman (24:40.800)
So you have one giant steam engine,
Erik Brynjolfsson (24:42.360)
all the equipment clustered around it, multi story.
Lex Fridman (24:44.320)
They have it vertical to minimize the distance
Erik Brynjolfsson (24:46.080)
as well as horizontal.
Lex Fridman (24:47.760)
And then when they did electricity,
Erik Brynjolfsson (24:48.960)
they took out the steam engine.
Lex Fridman (24:50.080)
They got the biggest electric motor
Erik Brynjolfsson (24:51.360)
they could buy from General Electric or someone like that.
Lex Fridman (24:54.200)
And nothing much else changed.
Erik Brynjolfsson (24:57.920)
It took until a generation of managers retired
Lex Fridman (25:00.760)
or died three years later,
Erik Brynjolfsson (25:03.200)
that people started thinking,
Lex Fridman (25:04.440)
wait, we don't have to do it that way.
Erik Brynjolfsson (25:05.840)
You can make electric motors, big, small, medium.
Lex Fridman (25:09.480)
You can put one with each piece of equipment.
Erik Brynjolfsson (25:11.480)
There's this big debate
Lex Fridman (25:12.320)
if you read the management literature
Erik Brynjolfsson (25:13.360)
between what they call a group drive versus unit drive
Lex Fridman (25:16.160)
where every machine would have its own motor.
Erik Brynjolfsson (25:18.960)
Well, once they did that, once they went to unit drive,
Lex Fridman (25:21.040)
those guys won the debate.
Erik Brynjolfsson (25:23.040)
Then you started having a new kind of factory
Lex Fridman (25:25.080)
which is sometimes spread out over acres, single story
Lex Fridman (25:29.400)
and each piece of equipment has its own motor.
Lex Fridman (25:31.360)
And most importantly, they weren't laid out based on
Erik Brynjolfsson (25:33.360)
who needed the most power.
Lex Fridman (25:35.000)
They were laid out based on
Lex Fridman (25:37.600)
what is the workflow of materials?
Lex Fridman (25:40.000)
Assembly line, let's have it go from this machine
Erik Brynjolfsson (25:41.720)
to that machine, to that machine.
Lex Fridman (25:43.240)
Once they rethought the factory that way,
Erik Brynjolfsson (25:46.040)
huge increases in productivity.
Lex Fridman (25:47.680)
It was just staggering.
Erik Brynjolfsson (25:48.520)
People like Paul David have documented this
Lex Fridman (25:50.080)
in their research papers.
Lex Fridman (25:51.760)
And I think that that is a lesson you see over and over.
Lex Fridman (25:55.920)
It happened when the steam engine changed manual production.
Erik Brynjolfsson (25:58.560)
It's happened with the computerization.
Lex Fridman (26:00.240)
People like Michael Hammer said, don't automate, obliterate.
Erik Brynjolfsson (26:03.600)
In each case, the big gains only came once
Lex Fridman (26:08.440)
smart entrepreneurs and managers
Erik Brynjolfsson (26:10.400)
basically reinvented their industries.
Lex Fridman (26:13.040)
I mean, one other interesting point about all that
Erik Brynjolfsson (26:14.680)
is that during that reinvention period,
Lex Fridman (26:18.920)
you often actually not only don't see productivity growth,
Erik Brynjolfsson (26:22.200)
you can actually see a slipping back.
Lex Fridman (26:24.320)
Measured productivity actually falls.
Erik Brynjolfsson (26:26.560)
I just wrote a paper with Chad Severson and Daniel Rock
Lex Fridman (26:29.040)
called the productivity J curve,
Erik Brynjolfsson (26:31.520)
which basically shows that in a lot of these cases,
Lex Fridman (26:33.840)
you have a downward dip before it goes up.
Lex Fridman (26:36.520)
And that downward dip is when everyone's trying
Lex Fridman (26:38.320)
to like reinvent things.
Lex Fridman (26:40.400)
And you could say that they're creating knowledge
Lex Fridman (26:43.120)
and intangible assets,
Lex Fridman (26:44.640)
but that doesn't show up on anyone's balance sheet.
Lex Fridman (26:46.760)
It doesn't show up in GDP.
Lex Fridman (26:48.320)
So it's as if they're doing nothing.
Lex Fridman (26:50.080)
Like take self driving cars, we were just talking about it.
Erik Brynjolfsson (26:52.480)
There have been hundreds of billions of dollars
Lex Fridman (26:55.040)
spent developing self driving cars.
Lex Fridman (26:57.880)
And basically no chauffeur has lost his job, no taxi driver.
Lex Fridman (27:02.360)
I guess I got to check out the ones that.
Erik Brynjolfsson (27:03.200)
It's a big J curve.
Lex Fridman (27:04.440)
Yeah, so there's a bunch of spending
Lex Fridman (27:06.080)
and no real consumer benefit.
Lex Fridman (27:08.120)
Now they're doing that in the belief,
Erik Brynjolfsson (27:11.440)
I think the justified belief
Lex Fridman (27:13.240)
that they will get the upward part of the J curve
Lex Fridman (27:15.160)
and there will be some big returns,
Lex Fridman (27:16.920)
but in the short run, you're not seeing it.
Erik Brynjolfsson (27:19.320)
That's happening with a lot of other AI technologies,
Lex Fridman (27:21.840)
just as it happened
Erik Brynjolfsson (27:22.680)
with earlier general purpose technologies.
Lex Fridman (27:25.040)
And it's one of the reasons
Erik Brynjolfsson (27:25.880)
we're having relatively low productivity growth lately.
Lex Fridman (27:29.280)
As an economist, one of the things that disappoints me
Erik Brynjolfsson (27:31.400)
is that as eye popping as these technologies are,
Lex Fridman (27:34.440)
you and I are both excited
Erik Brynjolfsson (27:35.360)
about some of the things they can do.
Lex Fridman (27:36.880)
The economic productivity statistics are kind of dismal.
Erik Brynjolfsson (27:40.280)
We actually, believe it or not,
Lex Fridman (27:42.200)
have had lower productivity growth
Erik Brynjolfsson (27:44.560)
in the past about 15 years
Lex Fridman (27:47.000)
than we did in the previous 15 years,
Erik Brynjolfsson (27:48.840)
in the 90s and early 2000s.
Lex Fridman (27:51.360)
And so that's not what you would have expected
Erik Brynjolfsson (27:53.200)
if these technologies were that much better.
Lex Fridman (27:55.520)
But I think we're in kind of a long J curve there.
Erik Brynjolfsson (27:59.400)
Personally, I'm optimistic.
Lex Fridman (28:00.560)
We'll start seeing the upward tick,
Erik Brynjolfsson (28:02.120)
maybe as soon as next year.
Lex Fridman (28:04.520)
But the past decade has been a bit disappointing
Erik Brynjolfsson (28:08.520)
if you thought there's a one to one relationship
Lex Fridman (28:10.000)
between cool technology and higher productivity.
Erik Brynjolfsson (28:12.760)
Well, what would you place your biggest hope
Lex Fridman (28:15.240)
for productivity increases on?
Erik Brynjolfsson (28:17.240)
Because you kind of said at a high level AI,
Lex Fridman (28:19.880)
but if I were to think about
Lex Fridman (28:22.840)
what has been so revolutionary in the last 10 years,
Lex Fridman (28:28.200)
I would 15 years and thinking about the internet,
Erik Brynjolfsson (28:32.240)
I would say things like,
Lex Fridman (28:35.800)
hopefully I'm not saying anything ridiculous,
Lex Fridman (28:37.160)
but everything from Wikipedia to Twitter.
Lex Fridman (28:41.040)
So like these kind of websites,
Erik Brynjolfsson (28:43.600)
not so much AI,
Lex Fridman (28:46.080)
but like I would expect to see some kind
Erik Brynjolfsson (28:48.160)
of big productivity increases
Lex Fridman (28:50.520)
from just the connectivity between people
Lex Fridman (28:54.160)
and the access to more information.
Lex Fridman (28:58.040)
Yeah, well, so that's another area
Erik Brynjolfsson (29:00.040)
I've done quite a bit of research on actually,
Lex Fridman (29:01.840)
is these free goods like Wikipedia, Facebook, Twitter, Zoom.
Erik Brynjolfsson (29:06.840)
We're actually doing this in person,
Lex Fridman (29:08.120)
but almost everything else I do these days is online.
Erik Brynjolfsson (29:12.400)
The interesting thing about all those
Lex Fridman (29:13.760)
is most of them have a price of zero.
Lex Fridman (29:18.880)
What do you pay for Wikipedia?
Lex Fridman (29:19.960)
Maybe like a little bit for the electrons
Erik Brynjolfsson (29:21.680)
to come to your house.
Lex Fridman (29:22.960)
Basically zero, right?
Erik Brynjolfsson (29:25.600)
Take a small pause and say, I donate to Wikipedia.
Lex Fridman (29:28.040)
Often you should too.
Erik Brynjolfsson (29:28.880)
It's good for you, yeah.
Lex Fridman (29:30.080)
So, but what does that do mean for GDP?
Erik Brynjolfsson (29:32.480)
GDP is based on the price and quantity
Lex Fridman (29:36.120)
of all the goods, things bought and sold.
Erik Brynjolfsson (29:37.760)
If something has zero price,
Lex Fridman (29:39.560)
you know how much it contributes to GDP?
Erik Brynjolfsson (29:42.280)
To a first approximation, zero.
Lex Fridman (29:44.520)
So these digital goods that we're getting more and more of,
Erik Brynjolfsson (29:47.560)
we're spending more and more hours a day
Lex Fridman (29:50.240)
consuming stuff off of screens,
Erik Brynjolfsson (29:52.080)
little screens, big screens,
Lex Fridman (29:54.520)
that doesn't get priced into GDP.
Erik Brynjolfsson (29:56.160)
It's like they don't exist.
Lex Fridman (29:58.640)
That doesn't mean they don't create value.
Erik Brynjolfsson (30:00.000)
I get a lot of value from watching cat videos
Lex Fridman (30:03.360)
and reading Wikipedia articles and listening to podcasts,
Erik Brynjolfsson (30:06.160)
even if I don't pay for them.
Lex Fridman (30:08.440)
So we've got a mismatch there.
Erik Brynjolfsson (30:10.440)
Now, in fairness, economists,
Lex Fridman (30:12.520)
since Simon Kuznets invented GDP and productivity,
Erik Brynjolfsson (30:15.320)
all those statistics back in the 1930s,
Lex Fridman (30:17.760)
he recognized, he in fact said,
Erik Brynjolfsson (30:19.680)
this is not a measure of wellbeing.
Lex Fridman (30:21.520)
This is not a measure of welfare.
Erik Brynjolfsson (30:23.200)
It's a measure of production.
Lex Fridman (30:25.120)
But almost everybody has kind of forgotten
Erik Brynjolfsson (30:28.960)
that he said that and they just use it.
Lex Fridman (30:31.000)
It's like, how well off are we?
Lex Fridman (30:32.200)
What was GDP last year?
Lex Fridman (30:33.240)
It was 2.3% growth or whatever.
Erik Brynjolfsson (30:35.800)
That is how much physical production,
Lex Fridman (30:39.440)
but it's not the value we're getting.
Erik Brynjolfsson (30:42.360)
We need a new set of statistics
Lex Fridman (30:43.840)
and I'm working with some colleagues.
Erik Brynjolfsson (30:45.280)
Avi Collis and others to develop something
Lex Fridman (30:48.360)
we call GDP dash B.
Erik Brynjolfsson (30:50.440)
GDP B measures the benefits you get, not the cost.
Lex Fridman (30:55.440)
If you get benefit from Zoom or Wikipedia or Facebook,
Erik Brynjolfsson (31:00.360)
then that gets counted in GDP B,
Lex Fridman (31:02.520)
even if you pay zero for it.
Erik Brynjolfsson (31:04.560)
So, you know, back to your original point,
Lex Fridman (31:07.360)
I think there is a lot of gain over the past decade
Erik Brynjolfsson (31:10.480)
in these digital goods that doesn't show up in GDP,
Lex Fridman (31:15.280)
doesn't show up in productivity.
Erik Brynjolfsson (31:16.440)
By the way, productivity is just defined
Lex Fridman (31:17.920)
as GDP divided by hours worked.
Lex Fridman (31:20.080)
So if you mismeasure GDP,
Lex Fridman (31:22.080)
you mismeasure productivity by the exact same amount.
Erik Brynjolfsson (31:25.360)
That's something we need to fix.
Lex Fridman (31:26.480)
I'm working with the statistical agencies
Erik Brynjolfsson (31:28.440)
to come up with a new set of metrics.
Lex Fridman (31:30.200)
And, you know, over the coming years,
Erik Brynjolfsson (31:32.200)
I think we'll see, we're not gonna do away with GDP.
Lex Fridman (31:34.480)
It's very useful, but we'll see a parallel set of accounts
Erik Brynjolfsson (31:37.240)
that measure the benefits.
Lex Fridman (31:38.400)
How difficult is it to get that B in the GDP B?
Erik Brynjolfsson (31:41.080)
It's pretty hard.
Lex Fridman (31:41.920)
I mean, one of the reasons it hasn't been done before
Erik Brynjolfsson (31:44.360)
is that, you know, you can measure it,
Lex Fridman (31:46.720)
the cash register, what people pay for stuff,
Lex Fridman (31:49.000)
but how do you measure what they would have paid,
Lex Fridman (31:51.040)
like what the value is?
Lex Fridman (31:52.040)
That's a lot harder, you know?
Lex Fridman (31:54.040)
How much is Wikipedia worth to you?
Erik Brynjolfsson (31:56.040)
That's what we have to answer.
Lex Fridman (31:57.480)
And to do that, what we do is we can use online experiments.
Erik Brynjolfsson (32:00.720)
We do massive online choice experiments.
Lex Fridman (32:03.120)
We ask hundreds of thousands, now millions of people
Erik Brynjolfsson (32:05.720)
to do lots of sort of A, B tests.
Lex Fridman (32:07.760)
How much would I have to pay you
Lex Fridman (32:09.080)
to give up Wikipedia for a month?
Lex Fridman (32:10.840)
How much would I have to pay you to stop using your phone?
Lex Fridman (32:14.120)
And in some cases, it's hypothetical.
Lex Fridman (32:15.960)
In other cases, we actually enforce it,
Erik Brynjolfsson (32:17.520)
which is kind of expensive.
Lex Fridman (32:18.920)
Like we pay somebody $30 to stop using Facebook
Lex Fridman (32:22.440)
and we see if they'll do it.
Lex Fridman (32:23.440)
And some people will give it up for $10.
Erik Brynjolfsson (32:26.280)
Some people won't give it up even if you give them $100.
Lex Fridman (32:28.880)
And then you get a whole demand curve.
Erik Brynjolfsson (32:31.080)
You get to see what all the different prices are
Lex Fridman (32:33.600)
and how much value different people get.
Lex Fridman (32:36.000)
And not surprisingly,
Lex Fridman (32:36.880)
different people have different values.
Erik Brynjolfsson (32:38.240)
We find that women tend to value Facebook more than men.
Lex Fridman (32:41.520)
Old people tend to value it a little bit more
Erik Brynjolfsson (32:43.200)
than young people.
Lex Fridman (32:44.040)
That was interesting.
Erik Brynjolfsson (32:44.880)
I think young people maybe know about other networks
Lex Fridman (32:46.600)
that I don't know the name of that are better than Facebook.
Lex Fridman (32:50.280)
And so you get to see these patterns,
Lex Fridman (32:53.480)
but every person's individual.
Lex Fridman (32:55.440)
And then if you add up all those numbers,
Lex Fridman (32:57.280)
you start getting an estimate of the value.
Erik Brynjolfsson (33:00.040)
Okay, first of all, that's brilliant.
Lex Fridman (33:01.240)
Is this a work that will soon eventually be published?
Erik Brynjolfsson (33:05.720)
Yeah, well, there's a version of it
Lex Fridman (33:07.040)
in the Proceedings of the National Academy of Sciences
Erik Brynjolfsson (33:09.520)
about I think we call it massive online choice experiments.
Lex Fridman (33:11.880)
I should remember the title, but it's on my website.
Lex Fridman (33:14.920)
So yeah, we have some more papers coming out on it,
Lex Fridman (33:17.520)
but the first one is already out.
Erik Brynjolfsson (33:20.160)
You know, it's kind of a fascinating mystery
Lex Fridman (33:22.320)
that Twitter, Facebook,
Erik Brynjolfsson (33:24.360)
like all these social networks are free.
Lex Fridman (33:26.800)
And it seems like almost none of them except for YouTube
Erik Brynjolfsson (33:31.440)
have experimented with removing ads for money.
Lex Fridman (33:35.200)
Can you like, do you understand that
Lex Fridman (33:37.160)
from both economics and the product perspective?
Lex Fridman (33:39.800)
Yeah, it's something that, you know,
Lex Fridman (33:41.000)
so I teach a course on digital business models.
Lex Fridman (33:43.240)
So I used to at MIT, at Stanford, I'm not quite sure.
Erik Brynjolfsson (33:45.800)
I'm not teaching until next spring.
Lex Fridman (33:47.360)
I'm still thinking what my course is gonna be.
Lex Fridman (33:50.040)
But there are a lot of different business models.
Lex Fridman (33:52.200)
And when you have something that has zero marginal cost,
Erik Brynjolfsson (33:54.880)
there's a lot of forces,
Lex Fridman (33:56.400)
especially if there's any kind of competition
Erik Brynjolfsson (33:57.880)
that push prices down to zero.
Lex Fridman (33:59.960)
But you can have ad supported systems,
Erik Brynjolfsson (34:03.360)
you can bundle things together.
Lex Fridman (34:05.520)
You can have volunteer, you mentioned Wikipedia,
Erik Brynjolfsson (34:07.360)
there's donations.
Lex Fridman (34:08.760)
And I think economists underestimate
Erik Brynjolfsson (34:11.120)
the power of volunteerism and donations.
Lex Fridman (34:14.560)
Your national public radio.
Erik Brynjolfsson (34:16.040)
Actually, how do you, this podcast, how is this,
Lex Fridman (34:18.560)
what's the revenue model?
Erik Brynjolfsson (34:19.480)
There's sponsors at the beginning.
Lex Fridman (34:22.240)
And then, and people, the funny thing is,
Erik Brynjolfsson (34:24.640)
I tell people they can, it's very,
Lex Fridman (34:26.640)
I tell them the timestamp.
Lex Fridman (34:27.880)
So if you wanna skip the sponsors, you're free.
Lex Fridman (34:30.960)
But it's funny that a bunch of people,
Lex Fridman (34:33.560)
so I read the advertisement
Lex Fridman (34:36.200)
and then a bunch of people enjoy reading it.
Lex Fridman (34:38.400)
And it's.
Lex Fridman (34:39.240)
Well, they may learn something from it.
Lex Fridman (34:40.080)
And also from the advertiser's perspective,
Lex Fridman (34:42.920)
those are people who are actually interested.
Erik Brynjolfsson (34:45.400)
I mean, the example I sometimes get is like,
Lex Fridman (34:46.960)
I bought a car recently and all of a sudden,
Erik Brynjolfsson (34:49.840)
all the car ads were like interesting to me.
Lex Fridman (34:52.400)
Exactly.
Lex Fridman (34:53.240)
And then like, now that I have the car,
Lex Fridman (34:54.360)
like I sort of zone out on, but that's fine.
Erik Brynjolfsson (34:56.280)
The car companies, they don't really wanna be advertising
Lex Fridman (34:58.720)
to me if I'm not gonna buy their product.
Lex Fridman (35:01.320)
So there are a lot of these different revenue models
Lex Fridman (35:03.560)
and it's a little complicated,
Lex Fridman (35:06.880)
but the economic theory has to do
Lex Fridman (35:08.000)
with what the shape of the demand curve is,
Erik Brynjolfsson (35:09.480)
when it's better to monetize it with charging people
Lex Fridman (35:13.160)
versus when you're better off doing advertising.
Erik Brynjolfsson (35:15.640)
I mean, in short, when the demand curve
Lex Fridman (35:18.280)
is relatively flat and wide,
Erik Brynjolfsson (35:20.600)
like generic news and things like that,
Lex Fridman (35:22.760)
then you tend to do better with advertising.
Erik Brynjolfsson (35:25.920)
If it's a good that's only useful to a small number
Lex Fridman (35:28.840)
of people, but they're willing to pay a lot,
Erik Brynjolfsson (35:30.320)
they have a very high value for it,
Lex Fridman (35:32.720)
then advertising isn't gonna work as well
Lex Fridman (35:34.560)
and you're better off charging for it.
Lex Fridman (35:36.080)
Both of them have some inefficiencies.
Lex Fridman (35:38.080)
And then when you get into targeting
Lex Fridman (35:39.480)
and you get into these other revenue models,
Erik Brynjolfsson (35:40.600)
it gets more complicated,
Lex Fridman (35:41.960)
but there's some economic theory on it.
Erik Brynjolfsson (35:45.320)
I also think to be frank,
Lex Fridman (35:47.560)
there's just a lot of experimentation that's needed
Erik Brynjolfsson (35:49.560)
because sometimes things are a little counterintuitive,
Lex Fridman (35:53.200)
especially when you get into what are called
Erik Brynjolfsson (35:55.160)
two sided networks or platform effects,
Lex Fridman (35:57.640)
where you may grow the market on one side
Lex Fridman (36:01.840)
and harvest the revenue on the other side.
Lex Fridman (36:04.120)
Facebook tries to get more and more users
Lex Fridman (36:06.080)
and then they harvest the revenue from advertising.
Lex Fridman (36:08.960)
So that's another way of kind of thinking about it.
Lex Fridman (36:12.040)
Is it strange to you that they haven't experimented?
Lex Fridman (36:14.400)
Well, they are experimenting.
Lex Fridman (36:15.360)
So they are doing some experiments
Lex Fridman (36:17.600)
about what the willingness is for people to pay.
Erik Brynjolfsson (36:22.040)
I think that when they do the math,
Lex Fridman (36:23.560)
it's gonna work out that they still are better off
Erik Brynjolfsson (36:26.400)
with an advertising driven model, but...
Lex Fridman (36:29.440)
What about a mix?
Lex Fridman (36:30.400)
Like this is what YouTube is, right?
Lex Fridman (36:32.400)
It's you allow the person to decide,
Erik Brynjolfsson (36:36.360)
the customer to decide exactly which model they prefer.
Lex Fridman (36:39.360)
No, that can work really well.
Lex Fridman (36:40.920)
And newspapers, of course,
Lex Fridman (36:41.760)
have known this for a long time.
Erik Brynjolfsson (36:42.760)
The Wall Street Journal, the New York Times,
Lex Fridman (36:44.560)
they have subscription revenue.
Erik Brynjolfsson (36:45.840)
They also have advertising revenue.
Lex Fridman (36:48.080)
And that can definitely work.
Erik Brynjolfsson (36:52.200)
Online, it's a lot easier to have a dial
Lex Fridman (36:54.080)
that's much more personalized
Lex Fridman (36:55.240)
and everybody can kind of roll their own mix.
Lex Fridman (36:57.720)
And I could imagine having a little slider
Erik Brynjolfsson (37:00.320)
about how much advertising you want or are willing to take.
Lex Fridman (37:05.040)
And if it's done right and it's incentive compatible,
Erik Brynjolfsson (37:07.400)
it could be a win win where both the content provider
Lex Fridman (37:10.960)
and the consumer are better off
Erik Brynjolfsson (37:12.560)
than they would have been before.
Lex Fridman (37:14.480)
Yeah, the done right part is a really good point.
Erik Brynjolfsson (37:17.960)
Like with the Jeff Bezos
Lex Fridman (37:19.600)
and the single click purchase on Amazon,
Erik Brynjolfsson (37:22.000)
the frictionless effort there,
Lex Fridman (37:23.880)
if I could just rant for a second
Erik Brynjolfsson (37:25.760)
about the Wall Street Journal,
Lex Fridman (37:27.240)
all the newspapers you mentioned,
Erik Brynjolfsson (37:29.280)
is I have to click so many times to subscribe to them
Lex Fridman (37:34.800)
that I literally don't subscribe
Erik Brynjolfsson (37:37.400)
just because of the number of times I have to click.
Lex Fridman (37:39.520)
I'm totally with you.
Erik Brynjolfsson (37:40.360)
I don't understand why so many companies make it so hard.
Lex Fridman (37:44.560)
I mean, another example is when you buy a new iPhone
Erik Brynjolfsson (37:47.240)
or a new computer, whatever,
Lex Fridman (37:48.900)
I feel like, okay, I'm gonna lose an afternoon
Erik Brynjolfsson (37:51.440)
just like loading up and getting all my stuff back.
Lex Fridman (37:53.800)
And for a lot of us,
Erik Brynjolfsson (37:56.080)
that's more of a deterrent than the price.
Lex Fridman (37:58.600)
And if they could make it painless,
Erik Brynjolfsson (38:01.800)
we'd give them a lot more money.
Lex Fridman (38:03.680)
So I'm hoping somebody listening is working
Erik Brynjolfsson (38:06.440)
on making it more painless for us to buy your products.
Lex Fridman (38:10.000)
If we could just like linger a little bit
Erik Brynjolfsson (38:12.280)
on the social network thing,
Lex Fridman (38:13.680)
because there's this Netflix social dilemma.
Erik Brynjolfsson (38:18.200)
Yeah, no, I saw that.
Lex Fridman (38:19.280)
And Tristan Harris and company, yeah.
Lex Fridman (38:24.000)
And people's data,
Lex Fridman (38:29.500)
it's really sensitive and social networks
Erik Brynjolfsson (38:31.560)
are at the core arguably of many of societal like tension
Lex Fridman (38:37.440)
and some of the most important things happening in society.
Lex Fridman (38:39.640)
So it feels like it's important to get this right,
Lex Fridman (38:42.040)
both from a business model perspective
Lex Fridman (38:43.960)
and just like a trust perspective.
Lex Fridman (38:46.340)
I still gotta, I mean, it just still feels like,
Erik Brynjolfsson (38:49.840)
I know there's experimentation going on.
Lex Fridman (38:52.140)
It still feels like everyone is afraid
Erik Brynjolfsson (38:54.740)
to try different business models, like really try.
Lex Fridman (38:57.520)
Well, I'm worried that people are afraid
Erik Brynjolfsson (38:59.600)
to try different business models.
Lex Fridman (39:01.220)
I'm also worried that some of the business models
Erik Brynjolfsson (39:03.480)
may lead them to bad choices.
Lex Fridman (39:06.280)
And Danny Kahneman talks about system one and system two,
Erik Brynjolfsson (39:10.980)
sort of like a reptilian brain
Lex Fridman (39:12.280)
that reacts quickly to what we see,
Erik Brynjolfsson (39:14.360)
see something interesting, we click on it,
Lex Fridman (39:16.160)
we retweet it versus our system two,
Erik Brynjolfsson (39:20.800)
our frontal cortex that's supposed to be more careful
Lex Fridman (39:24.080)
and rational that really doesn't make
Erik Brynjolfsson (39:26.240)
as many decisions as it should.
Lex Fridman (39:28.840)
I think there's a tendency for a lot of these social networks
Erik Brynjolfsson (39:32.680)
to really exploit system one, our quick instant reaction,
Lex Fridman (39:37.680)
make it so we just click on stuff and pass it on
Lex Fridman (39:40.960)
and not really think carefully about it.
Lex Fridman (39:42.320)
And that system, it tends to be driven
Erik Brynjolfsson (39:45.160)
by sex, violence, disgust, anger, fear,
Lex Fridman (39:51.320)
these relatively primitive kinds of emotions.
Erik Brynjolfsson (39:53.800)
Maybe they're important for a lot of purposes,
Lex Fridman (39:55.960)
but they're not a great way to organize a society.
Lex Fridman (39:58.920)
And most importantly, when you think about this huge,
Lex Fridman (40:01.920)
amazing information infrastructure we've had
Erik Brynjolfsson (40:04.320)
that's connected billions of brains across the globe,
Lex Fridman (40:08.000)
not just so we can all access information,
Lex Fridman (40:09.640)
but we can all contribute to it and share it.
Lex Fridman (40:12.640)
Arguably the most important thing
Erik Brynjolfsson (40:14.100)
that that network should do is favor truth over falsehoods.
Lex Fridman (40:19.360)
And the way it's been designed,
Erik Brynjolfsson (40:21.640)
not necessarily intentionally, is exactly the opposite.
Lex Fridman (40:24.660)
My MIT colleagues are all, and Deb Roy and others at MIT,
Erik Brynjolfsson (40:29.440)
did a terrific paper in the cover of Science.
Lex Fridman (40:31.760)
And they documented what we all feared,
Erik Brynjolfsson (40:33.460)
which is that lies spread faster than truth
Lex Fridman (40:37.740)
on social networks.
Erik Brynjolfsson (40:39.760)
They looked at a bunch of tweets and retweets,
Lex Fridman (40:42.760)
and they found that false information
Erik Brynjolfsson (40:44.560)
was more likely to spread further, faster, to more people.
Lex Fridman (40:48.960)
And why was that?
Erik Brynjolfsson (40:49.920)
It's not because people like lies.
Lex Fridman (40:53.360)
It's because people like things that are shocking,
Lex Fridman (40:55.880)
amazing, can you believe this?
Lex Fridman (40:57.840)
Something that is not mundane,
Erik Brynjolfsson (41:00.280)
not something that everybody else already knew.
Lex Fridman (41:02.460)
And what are the most unbelievable things?
Erik Brynjolfsson (41:05.400)
Well, lies.
Lex Fridman (41:07.320)
And so if you wanna find something unbelievable,
Erik Brynjolfsson (41:09.820)
it's a lot easier to do that
Lex Fridman (41:10.660)
if you're not constrained by the truth.
Lex Fridman (41:12.440)
So they found that the emotional valence
Lex Fridman (41:15.640)
of false information was just much higher.
Erik Brynjolfsson (41:17.960)
It was more likely to be shocking,
Lex Fridman (41:19.680)
and therefore more likely to be spread.
Erik Brynjolfsson (41:22.880)
Another interesting thing was that
Lex Fridman (41:24.040)
that wasn't necessarily driven by the algorithms.
Erik Brynjolfsson (41:27.580)
I know that there is some evidence,
Lex Fridman (41:29.680)
Zeynep Tufekci and others have pointed out on YouTube,
Erik Brynjolfsson (41:32.400)
some of the algorithms unintentionally were tuned
Lex Fridman (41:34.680)
to amplify more extremist content.
Lex Fridman (41:37.840)
But in the study of Twitter that Sinan and Deb and others did,
Lex Fridman (41:42.420)
they found that even if you took out all the bots
Lex Fridman (41:44.480)
and all the automated tweets,
Lex Fridman (41:47.880)
you still had lies spreading significantly faster.
Erik Brynjolfsson (41:50.760)
It's just the problems with ourselves
Lex Fridman (41:52.560)
that we just can't resist passing on the salacious content.
Lex Fridman (41:58.480)
But I also blame the platforms
Lex Fridman (41:59.920)
because there's different ways you can design a platform.
Erik Brynjolfsson (42:03.160)
You can design a platform in a way
Lex Fridman (42:05.400)
that makes it easy to spread lies
Lex Fridman (42:07.280)
and to retweet and spread things on,
Lex Fridman (42:09.520)
or you can kind of put some friction on that
Lex Fridman (42:11.520)
and try to favor truth.
Lex Fridman (42:13.960)
I had dinner with Jimmy Wales once,
Erik Brynjolfsson (42:15.520)
the guy who helped found Wikipedia.
Lex Fridman (42:19.720)
And he convinced me that, look,
Erik Brynjolfsson (42:22.580)
you can make some design choices,
Lex Fridman (42:24.500)
whether it's at Facebook, at Twitter,
Erik Brynjolfsson (42:26.360)
at Wikipedia, or Reddit, whatever,
Lex Fridman (42:29.240)
and depending on how you make those choices,
Erik Brynjolfsson (42:32.340)
you're more likely or less likely to have false news.
Lex Fridman (42:35.080)
Create a little bit of friction, like you said.
Erik Brynjolfsson (42:37.160)
Yeah.
Lex Fridman (42:38.000)
You know, that's the, and so if I'm...
Erik Brynjolfsson (42:39.560)
It could be friction, it could be speeding the truth,
Lex Fridman (42:41.560)
either way, but, and I don't totally understand...
Erik Brynjolfsson (42:44.400)
Speeding the truth, I love it.
Lex Fridman (42:45.520)
Yeah, yeah.
Erik Brynjolfsson (42:47.040)
Amplifying it and giving it more credit.
Lex Fridman (42:48.900)
And in academia, which is far, far from perfect,
Lex Fridman (42:52.520)
but when someone has an important discovery,
Lex Fridman (42:55.640)
it tends to get more cited
Lex Fridman (42:56.880)
and people kind of look to it more
Lex Fridman (42:58.160)
and sort of, it tends to get amplified a little bit.
Lex Fridman (43:00.760)
So you could try to do that too.
Lex Fridman (43:03.320)
I don't know what the silver bullet is,
Lex Fridman (43:04.680)
but the meta point is that if we spend time
Lex Fridman (43:07.440)
thinking about it, we can amplify truth over falsehoods.
Lex Fridman (43:10.800)
And I'm disappointed in the heads of these social networks
Lex Fridman (43:14.920)
that they haven't been as successful
Erik Brynjolfsson (43:16.680)
or maybe haven't tried as hard to amplify truth.
Lex Fridman (43:19.540)
And part of it, going back to what we said earlier,
Erik Brynjolfsson (43:21.560)
is these revenue models may push them
Lex Fridman (43:25.140)
more towards growing fast, spreading information rapidly,
Erik Brynjolfsson (43:29.880)
getting lots of users,
Lex Fridman (43:31.440)
which isn't the same thing as finding truth.
Erik Brynjolfsson (43:34.560)
Yeah, I mean, implicit in what you're saying now
Lex Fridman (43:38.840)
is a hopeful message that with platforms,
Erik Brynjolfsson (43:42.240)
we can take a step towards a greater
Lex Fridman (43:47.440)
and greater popularity of truth.
Lex Fridman (43:51.120)
But the more cynical view is that
Lex Fridman (43:54.000)
what the last few years have revealed
Erik Brynjolfsson (43:56.800)
is that there's a lot of money to be made
Lex Fridman (43:59.020)
in dismantling even the idea of truth,
Erik Brynjolfsson (44:03.120)
that nothing is true.
Lex Fridman (44:05.000)
And as a thought experiment,
Erik Brynjolfsson (44:07.020)
I've been thinking about if it's possible
Lex Fridman (44:09.320)
that our future will have,
Erik Brynjolfsson (44:11.200)
like the idea of truth is something we won't even have.
Lex Fridman (44:14.360)
Do you think it's possible in the future
Erik Brynjolfsson (44:17.800)
that everything is on the table in terms of truth,
Lex Fridman (44:20.980)
and we're just swimming in this kind of digital economy
Erik Brynjolfsson (44:24.720)
where ideas are just little toys
Lex Fridman (44:29.720)
that are not at all connected to reality?
Erik Brynjolfsson (44:33.080)
Yeah, I think that's definitely possible.
Lex Fridman (44:35.760)
I'm not a technological determinist,
Lex Fridman (44:37.960)
so I don't think that's inevitable.
Lex Fridman (44:40.280)
I don't think it's inevitable that it doesn't happen.
Erik Brynjolfsson (44:42.300)
I mean, the thing that I've come away with
Lex Fridman (44:43.960)
every time I do these studies,
Lex Fridman (44:45.320)
and I emphasize it in my books and elsewhere,
Lex Fridman (44:47.200)
is that technology doesn't shape our destiny,
Erik Brynjolfsson (44:50.040)
we shape our destiny.
Lex Fridman (44:51.680)
So just by us having this conversation,
Erik Brynjolfsson (44:54.640)
I hope that your audience is gonna take it upon themselves
Lex Fridman (44:58.440)
as they design their products,
Lex Fridman (44:59.880)
and they think about, they use products,
Lex Fridman (45:01.280)
as they manage companies,
Lex Fridman (45:02.760)
how can they make conscious decisions
Lex Fridman (45:05.300)
to favor truth over falsehoods,
Erik Brynjolfsson (45:08.840)
favor the better kinds of societies,
Lex Fridman (45:10.880)
and not abdicate and say, well, we just build the tools.
Erik Brynjolfsson (45:13.720)
I think there was a saying that,
Lex Fridman (45:16.940)
was it the German scientist
Lex Fridman (45:18.300)
when they were working on the missiles in late World War II?
Lex Fridman (45:23.000)
They said, well, our job is to make the missiles go up.
Erik Brynjolfsson (45:25.680)
Where they come down, that's someone else's department.
Lex Fridman (45:28.400)
And that's obviously not the, I think it's obvious,
Erik Brynjolfsson (45:31.840)
that's not the right attitude
Lex Fridman (45:32.840)
that technologists should have,
Erik Brynjolfsson (45:33.980)
that engineers should have.
Lex Fridman (45:35.680)
They should be very conscious
Erik Brynjolfsson (45:36.920)
about what the implications are.
Lex Fridman (45:38.800)
And if we think carefully about it,
Erik Brynjolfsson (45:40.600)
we can avoid the kind of world that you just described,
Lex Fridman (45:42.920)
where truth is all relative.
Erik Brynjolfsson (45:45.040)
There are going to be people who benefit from a world
Lex Fridman (45:47.840)
of where people don't check facts,
Lex Fridman (45:51.320)
and where truth is relative,
Lex Fridman (45:52.680)
and popularity or fame or money is orthogonal to truth.
Lex Fridman (45:59.880)
But one of the reasons I suspect
Lex Fridman (46:01.880)
that we've had so much progress over the past few hundred
Erik Brynjolfsson (46:04.540)
years is the invention of the scientific method,
Lex Fridman (46:07.600)
which is a really powerful tool or meta tool
Erik Brynjolfsson (46:10.200)
for finding truth and favoring things that are true
Lex Fridman (46:15.400)
versus things that are false.
Erik Brynjolfsson (46:16.600)
If they don't pass the scientific method,
Lex Fridman (46:18.560)
they're less likely to be true.
Lex Fridman (46:20.640)
And that has, the societies and the people
Lex Fridman (46:25.560)
and the organizations that embrace that
Erik Brynjolfsson (46:27.760)
have done a lot better than the ones who haven't.
Lex Fridman (46:30.520)
And so I'm hoping that people keep that in mind
Lex Fridman (46:32.800)
and continue to try to embrace not just the truth,
Lex Fridman (46:35.460)
but methods that lead to the truth.
Lex Fridman (46:37.640)
So maybe on a more personal question,
Lex Fridman (46:41.400)
if one were to try to build a competitor to Twitter,
Lex Fridman (46:45.480)
what would you advise?
Lex Fridman (46:47.360)
Is there, I mean, the bigger, the meta question,
Lex Fridman (46:53.360)
is that the right way to improve systems?
Lex Fridman (46:55.680)
Yeah, no, I think that the underlying premise
Erik Brynjolfsson (46:59.380)
behind Twitter and all these networks is amazing,
Lex Fridman (47:01.380)
that we can communicate with each other.
Lex Fridman (47:02.800)
And I use it a lot.
Lex Fridman (47:04.000)
There's a subpart of Twitter called Econ Twitter,
Erik Brynjolfsson (47:05.920)
where we economists tweet to each other
Lex Fridman (47:08.640)
and talk about new papers.
Erik Brynjolfsson (47:10.560)
Something came out in the NBER,
Lex Fridman (47:11.960)
the National Bureau of Economic Research,
Lex Fridman (47:13.320)
and we share about it.
Lex Fridman (47:14.160)
People critique it.
Erik Brynjolfsson (47:15.360)
I think it's been a godsend
Lex Fridman (47:16.880)
because it's really sped up the scientific process,
Erik Brynjolfsson (47:20.040)
if you can call economic scientific.
Lex Fridman (47:21.880)
Does it get divisive in that little?
Erik Brynjolfsson (47:23.560)
Sometimes, yeah, sure.
Lex Fridman (47:24.500)
Sometimes it does.
Erik Brynjolfsson (47:25.340)
It can also be done in nasty ways and there's the bad parts.
Lex Fridman (47:28.360)
But the good parts are great
Erik Brynjolfsson (47:29.680)
because you just speed up that clock speed
Lex Fridman (47:31.640)
of learning about things.
Erik Brynjolfsson (47:33.320)
Instead of like in the old, old days,
Lex Fridman (47:35.480)
waiting to read it in a journal,
Erik Brynjolfsson (47:36.800)
or the not so old days when you'd see it posted
Lex Fridman (47:39.520)
on a website and you'd read it.
Erik Brynjolfsson (47:41.600)
Now on Twitter, people will distill it down
Lex Fridman (47:44.000)
and it's a real art to getting to the essence of things.
Lex Fridman (47:47.160)
So that's been great.
Lex Fridman (47:49.080)
But it certainly, we all know that Twitter
Erik Brynjolfsson (47:52.320)
can be a cesspool of misinformation.
Lex Fridman (47:55.560)
And like I just said,
Erik Brynjolfsson (47:57.360)
unfortunately misinformation tends to spread faster
Lex Fridman (48:00.240)
on Twitter than truth.
Lex Fridman (48:02.320)
And there are a lot of people
Lex Fridman (48:03.160)
who are very vulnerable to it.
Erik Brynjolfsson (48:04.200)
I'm sure I've been fooled at times.
Lex Fridman (48:06.000)
There are agents, whether from Russia
Erik Brynjolfsson (48:09.120)
or from political groups or others
Lex Fridman (48:11.680)
that explicitly create efforts at misinformation
Lex Fridman (48:15.640)
and efforts at getting people to hate each other.
Lex Fridman (48:17.900)
Or even more important lately I've discovered
Erik Brynjolfsson (48:19.720)
is nut picking.
Lex Fridman (48:21.200)
You know the idea of nut picking?
Lex Fridman (48:22.320)
No, what's that?
Lex Fridman (48:23.160)
It's a good term.
Erik Brynjolfsson (48:24.320)
Nut picking is when you find like an extreme nut case
Lex Fridman (48:27.800)
on the other side and then you amplify them
Lex Fridman (48:30.700)
and make it seem like that's typical of the other side.
Lex Fridman (48:34.000)
So you're not literally lying.
Erik Brynjolfsson (48:35.480)
You're taking some idiot, you know,
Lex Fridman (48:37.760)
renting on the subway or just, you know,
Erik Brynjolfsson (48:39.920)
whether they're in the KKK or Antifa or whatever,
Lex Fridman (48:42.800)
they're just, and you,
Erik Brynjolfsson (48:44.360)
normally nobody would pay attention to this guy.
Lex Fridman (48:46.040)
Like 12 people would see him and it'd be the end.
Erik Brynjolfsson (48:48.080)
Instead with video or whatever,
Lex Fridman (48:51.120)
you get tens of millions of people say it.
Lex Fridman (48:54.520)
And I've seen this, you know, I look at it,
Lex Fridman (48:56.320)
I'm like, I get angry.
Erik Brynjolfsson (48:57.160)
I'm like, I can't believe that person
Lex Fridman (48:58.280)
did something so terrible.
Erik Brynjolfsson (48:59.720)
Let me tell all my friends about this terrible person.
Lex Fridman (49:02.880)
And it's a great way to generate division.
Erik Brynjolfsson (49:06.640)
I talked to a friend who studied Russian misinformation
Lex Fridman (49:10.520)
campaigns, and they're very clever about literally
Erik Brynjolfsson (49:13.840)
being on both sides of some of these debates.
Lex Fridman (49:15.880)
They would have some people pretend to be part of BLM.
Erik Brynjolfsson (49:18.620)
Some people pretend to be white nationalists
Lex Fridman (49:21.040)
and they would be throwing epithets at each other,
Erik Brynjolfsson (49:22.960)
saying crazy things at each other.
Lex Fridman (49:25.100)
And they're literally playing both sides of it,
Lex Fridman (49:26.600)
but their goal wasn't for one or the other to win.
Lex Fridman (49:28.600)
It was for everybody to get behaving
Lex Fridman (49:30.120)
and distrusting everyone else.
Lex Fridman (49:32.000)
So these tools can definitely be used for that.
Lex Fridman (49:34.500)
And they are being used for that.
Lex Fridman (49:36.580)
It's been super destructive for our democracy
Lex Fridman (49:39.680)
and our society.
Lex Fridman (49:41.080)
And the people who run these platforms,
Erik Brynjolfsson (49:43.540)
I think have a social responsibility,
Lex Fridman (49:46.100)
a moral and ethical, personal responsibility
Erik Brynjolfsson (49:48.680)
to do a better job and to shut that stuff down.
Lex Fridman (49:51.800)
Well, I don't know if you can shut it down,
Lex Fridman (49:52.960)
but to design them in a way that, you know,
Lex Fridman (49:55.800)
as I said earlier, favors truth over falsehoods
Lex Fridman (49:58.620)
and favors positive types of
Lex Fridman (50:03.200)
communication versus destructive ones.
Lex Fridman (50:06.060)
And just like you said, it's also on us.
Lex Fridman (50:09.600)
I try to be all about love and compassion,
Erik Brynjolfsson (50:12.400)
empathy on Twitter.
Lex Fridman (50:13.280)
I mean, one of the things,
Erik Brynjolfsson (50:14.820)
nut picking is a fascinating term.
Lex Fridman (50:16.600)
One of the things that people do,
Erik Brynjolfsson (50:18.940)
that's I think even more dangerous
Lex Fridman (50:21.800)
is nut picking applied to individual statements
Erik Brynjolfsson (50:26.760)
of good people.
Lex Fridman (50:28.440)
So basically worst case analysis in computer science
Erik Brynjolfsson (50:32.180)
is taking sometimes out of context,
Lex Fridman (50:35.360)
but sometimes in context,
Erik Brynjolfsson (50:38.480)
a statement, one statement by a person,
Lex Fridman (50:42.320)
like I've been, because I've been reading
Erik Brynjolfsson (50:43.740)
The Rise and Fall of the Third Reich,
Lex Fridman (50:45.360)
I often talk about Hitler on this podcast with folks
Lex Fridman (50:48.960)
and it is so easy.
Lex Fridman (50:50.640)
That's really dangerous.
Lex Fridman (50:52.060)
But I'm all leaning in, I'm 100%.
Lex Fridman (50:54.560)
Because, well, it's actually a safer place
Erik Brynjolfsson (50:56.960)
than people realize because it's history
Lex Fridman (50:59.200)
and history in long form is actually very fascinating
Erik Brynjolfsson (51:04.120)
to think about and it's,
Lex Fridman (51:06.300)
but I could see how that could be taken
Erik Brynjolfsson (51:09.600)
totally out of context and it's very worrying.
Lex Fridman (51:11.320)
You know, these digital infrastructures,
Erik Brynjolfsson (51:12.800)
not just they disseminate things,
Lex Fridman (51:14.040)
but they're sort of permanent.
Lex Fridman (51:14.880)
So anything you say at some point,
Lex Fridman (51:16.540)
someone can go back and find something you said
Erik Brynjolfsson (51:18.160)
three years ago, perhaps jokingly, perhaps not,
Lex Fridman (51:21.080)
maybe you're just wrong and you made them, you know,
Lex Fridman (51:22.800)
and like that becomes, they can use that to define you
Lex Fridman (51:25.600)
if they have ill intent.
Lex Fridman (51:26.840)
And we all need to be a little more forgiving.
Lex Fridman (51:29.080)
I mean, somewhere in my 20s, I told myself,
Erik Brynjolfsson (51:32.240)
I was going through all my different friends
Lex Fridman (51:33.820)
and I was like, you know, every one of them
Erik Brynjolfsson (51:37.300)
has at least like one nutty opinion.
Lex Fridman (51:39.400)
And I was like, there's like nobody
Erik Brynjolfsson (51:42.040)
who's like completely, except me, of course,
Lex Fridman (51:44.160)
but I'm sure they thought that about me too.
Lex Fridman (51:45.700)
And so you just kind of like learned
Lex Fridman (51:47.760)
to be a little bit tolerant that like, okay,
Erik Brynjolfsson (51:49.420)
there's just, you know.
Lex Fridman (51:51.140)
Yeah, I wonder who the responsibility lays on there.
Erik Brynjolfsson (51:55.240)
Like, I think ultimately it's about leadership.
Lex Fridman (51:59.680)
Like the previous president, Barack Obama,
Erik Brynjolfsson (52:02.760)
has been, I think, quite eloquent
Lex Fridman (52:06.040)
at walking this very difficult line
Erik Brynjolfsson (52:07.680)
of talking about cancel culture, but it's a difficult,
Lex Fridman (52:10.640)
it takes skill.
Erik Brynjolfsson (52:12.160)
Because you say the wrong thing
Lex Fridman (52:13.800)
and you piss off a lot of people.
Lex Fridman (52:15.320)
And so you have to do it well.
Lex Fridman (52:17.440)
But then also the platform of the technology is,
Erik Brynjolfsson (52:21.220)
should slow down, create friction,
Lex Fridman (52:23.600)
and spreading this kind of nut picking in all its forms.
Erik Brynjolfsson (52:26.440)
Absolutely.
Lex Fridman (52:27.280)
No, and your point that we have to like learn over time,
Lex Fridman (52:29.780)
how to manage it.
Lex Fridman (52:30.620)
I mean, we can't put it all on the platform
Lex Fridman (52:31.800)
and say, you guys design it.
Lex Fridman (52:33.240)
Because if we're idiots about using it,
Erik Brynjolfsson (52:35.200)
nobody can design a platform that withstands that.
Lex Fridman (52:38.480)
And every new technology people learn its dangers.
Erik Brynjolfsson (52:41.720)
You know, when someone invented fire,
Lex Fridman (52:43.960)
it's great cooking and everything,
Lex Fridman (52:44.960)
but then somebody burned themself.
Lex Fridman (52:46.160)
And then you had to like learn how to like avoid,
Erik Brynjolfsson (52:48.200)
maybe somebody invented a fire extinguisher later.
Lex Fridman (52:50.640)
So you kind of like figure out ways
Erik Brynjolfsson (52:52.840)
of working around these technologies.
Lex Fridman (52:54.640)
Someone invented seat belts, et cetera.
Lex Fridman (52:57.440)
And that's certainly true
Lex Fridman (52:58.640)
with all the new digital technologies
Erik Brynjolfsson (53:00.620)
that we have to figure out,
Lex Fridman (53:02.320)
not just technologies that protect us,
Lex Fridman (53:05.280)
but ways of using them that emphasize
Lex Fridman (53:08.640)
that are more likely to be successful than dangerous.
Lex Fridman (53:11.520)
So you've written quite a bit
Lex Fridman (53:12.560)
about how artificial intelligence might change our world.
Lex Fridman (53:19.000)
How do you think if we look forward,
Lex Fridman (53:21.240)
again, it's impossible to predict the future,
Lex Fridman (53:23.200)
but if we look at trends from the past
Lex Fridman (53:26.440)
and we tried to predict what's gonna happen
Erik Brynjolfsson (53:28.200)
in the rest of the 21st century,
Lex Fridman (53:29.720)
how do you think AI will change our world?
Erik Brynjolfsson (53:33.080)
That's a big question.
Lex Fridman (53:34.200)
You know, I'm mostly a techno optimist.
Erik Brynjolfsson (53:37.440)
I'm not at the extreme, you know,
Lex Fridman (53:38.660)
the singularity is near end of the spectrum,
Lex Fridman (53:41.080)
but I do think that we're likely in
Lex Fridman (53:44.560)
for some significantly improved living standards,
Erik Brynjolfsson (53:47.480)
some really important progress,
Lex Fridman (53:49.260)
even just the technologies that are already kind of like
Erik Brynjolfsson (53:51.240)
in the can that haven't diffused.
Lex Fridman (53:53.080)
You know, when I talked earlier about the J curve,
Erik Brynjolfsson (53:54.880)
it could take 10, 20, 30 years for an existing technology
Lex Fridman (53:58.760)
to have the kind of profound effects.
Lex Fridman (54:00.780)
And when I look at whether it's, you know,
Lex Fridman (54:03.760)
vision systems, voice recognition, problem solving systems,
Erik Brynjolfsson (54:07.840)
even if nothing new got invented,
Lex Fridman (54:09.400)
we would have a few decades of progress.
Lex Fridman (54:11.800)
So I'm excited about that.
Lex Fridman (54:13.440)
And I think that's gonna lead to us being wealthier,
Erik Brynjolfsson (54:16.840)
healthier, I mean,
Lex Fridman (54:17.800)
the healthcare is probably one of the applications
Erik Brynjolfsson (54:19.520)
that I'm most excited about.
Lex Fridman (54:22.520)
So that's good news.
Erik Brynjolfsson (54:23.760)
I don't think we're gonna have the end of work anytime soon.
Lex Fridman (54:26.760)
There's just too many things that machines still can't do.
Erik Brynjolfsson (54:30.960)
When I look around the world
Lex Fridman (54:32.000)
and think of whether it's childcare or healthcare,
Erik Brynjolfsson (54:34.640)
cleaning the environment, interacting with people,
Lex Fridman (54:37.740)
scientific work, artistic creativity,
Erik Brynjolfsson (54:40.900)
these are things that for now,
Lex Fridman (54:42.560)
machines aren't able to do nearly as well as humans,
Erik Brynjolfsson (54:45.640)
even just something as mundane as, you know,
Lex Fridman (54:47.160)
folding laundry or whatever.
Lex Fridman (54:48.720)
And many of these, I think are gonna be years or decades
Lex Fridman (54:52.920)
before machines catch up.
Erik Brynjolfsson (54:54.720)
You know, I may be surprised on some of them,
Lex Fridman (54:56.120)
but overall, I think there's plenty of work
Erik Brynjolfsson (54:58.760)
for humans to do.
Lex Fridman (54:59.760)
There's plenty of problems in society
Erik Brynjolfsson (55:01.320)
that need the human touch.
Lex Fridman (55:02.560)
So we'll have to repurpose.
Erik Brynjolfsson (55:04.180)
We'll have to, as machines are able to do some tasks,
Lex Fridman (55:07.880)
people are gonna have to reskill and move into other areas.
Lex Fridman (55:11.040)
And that's probably what's gonna be going on
Lex Fridman (55:12.740)
for the next, you know, 10, 20, 30 years or more,
Erik Brynjolfsson (55:16.240)
kind of big restructuring of society.
Lex Fridman (55:18.920)
We'll get wealthier and people will have to do new skills.
Erik Brynjolfsson (55:22.420)
Now, if you turn the dial further, I don't know,
Lex Fridman (55:24.360)
50 or a hundred years into the future,
Erik Brynjolfsson (55:26.960)
then, you know, maybe all bets are off.
Lex Fridman (55:29.640)
Then it's possible that machines will be able to do
Erik Brynjolfsson (55:32.880)
most of what people do.
Lex Fridman (55:34.240)
You know, say one or 200 years, I think it's even likely.
Lex Fridman (55:37.360)
And at that point,
Lex Fridman (55:38.400)
then we're more in the sort of abundance economy.
Erik Brynjolfsson (55:41.040)
Then we're in a world where there's really little
Lex Fridman (55:44.040)
for the humans can do economically better than machines,
Erik Brynjolfsson (55:48.000)
other than be human.
Lex Fridman (55:49.900)
And, you know, that will take a transition as well,
Erik Brynjolfsson (55:53.640)
kind of more of a transition of how we get meaning in life
Lex Fridman (55:56.480)
and what our values are.
Lex Fridman (55:58.220)
But shame on us if we screw that up.
Lex Fridman (56:00.400)
I mean, that should be like great, great news.
Lex Fridman (56:02.720)
And it kind of saddens me that some people see that
Lex Fridman (56:04.520)
as like a big problem.
Erik Brynjolfsson (56:05.540)
I think that would be, should be wonderful
Lex Fridman (56:07.640)
if people have all the health and material things
Erik Brynjolfsson (56:10.420)
that they need and can focus on loving each other
Lex Fridman (56:14.180)
and discussing philosophy and playing
Lex Fridman (56:16.840)
and doing all the other things that don't require work.
Lex Fridman (56:19.440)
Do you think you'd be surprised to see what the 20,
Erik Brynjolfsson (56:23.960)
if we were to travel in time, 100 years into the future,
Lex Fridman (56:27.420)
do you think you'll be able to,
Erik Brynjolfsson (56:29.560)
like if I gave you a month to like talk to people,
Lex Fridman (56:32.300)
no, like let's say a week,
Lex Fridman (56:34.120)
would you be able to understand what the hell's going on?
Lex Fridman (56:37.800)
You mean if I was there for a week?
Erik Brynjolfsson (56:39.200)
Yeah, if you were there for a week.
Lex Fridman (56:40.840)
A hundred years in the future?
Erik Brynjolfsson (56:42.120)
Yeah.
Lex Fridman (56:43.000)
So like, so I'll give you one thought experiment is like,
Erik Brynjolfsson (56:46.600)
isn't it possible that we're all living in virtual reality
Lex Fridman (56:49.640)
by then?
Erik Brynjolfsson (56:50.480)
Yeah, no, I think that's very possible.
Lex Fridman (56:52.620)
I've played around with some of those VR headsets
Lex Fridman (56:54.640)
and they're not great,
Lex Fridman (56:55.480)
but I mean the average person spends many waking hours
Erik Brynjolfsson (57:00.960)
staring at screens right now.
Lex Fridman (57:03.320)
They're kind of low res compared to what they could be
Erik Brynjolfsson (57:05.720)
in 30 or 50 years, but certainly games
Lex Fridman (57:10.680)
and why not any other interactions could be done with VR?
Lex Fridman (57:15.360)
And that would be a pretty different world
Lex Fridman (57:16.320)
and we'd all, in some ways be as rich as we wanted.
Erik Brynjolfsson (57:19.520)
We could have castles and we could be traveling
Lex Fridman (57:21.360)
anywhere we want and it could obviously be multisensory.
Lex Fridman (57:25.960)
So that would be possible and of course there's people,
Lex Fridman (57:30.880)
you've had Elon Musk on and others, there are people,
Erik Brynjolfsson (57:33.360)
Nick Bostrom makes the simulation argument
Lex Fridman (57:35.380)
that maybe we're already there.
Erik Brynjolfsson (57:36.760)
We're already there.
Lex Fridman (57:37.720)
So, but in general, or do you not even think about
Erik Brynjolfsson (57:41.200)
in this kind of way, you're self critically thinking,
Lex Fridman (57:45.080)
how good are you as an economist at predicting
Lex Fridman (57:48.560)
what the future looks like?
Lex Fridman (57:50.340)
Do you have a?
Erik Brynjolfsson (57:51.180)
Well, it starts getting, I mean,
Lex Fridman (57:52.000)
I feel reasonably comfortable the next five, 10, 20 years
Erik Brynjolfsson (57:55.960)
in terms of that path.
Lex Fridman (57:58.720)
When you start getting truly superhuman
Erik Brynjolfsson (58:01.720)
artificial intelligence, kind of by definition,
Lex Fridman (58:06.000)
be able to think of a lot of things
Erik Brynjolfsson (58:07.040)
that I couldn't have thought of and create a world
Lex Fridman (58:09.080)
that I couldn't even imagine.
Lex Fridman (58:10.960)
And so I'm not sure I can predict what that world
Lex Fridman (58:15.240)
is going to be like.
Erik Brynjolfsson (58:16.520)
One thing that AI researchers, AI safety researchers
Lex Fridman (58:19.840)
worry about is what's called the alignment problem.
Erik Brynjolfsson (58:22.540)
When an AI is that powerful,
Lex Fridman (58:25.080)
then they can do all sorts of things.
Lex Fridman (58:27.960)
And you really hope that their values
Lex Fridman (58:30.560)
are aligned with our values.
Lex Fridman (58:32.440)
And it's even tricky to finding what our values are.
Lex Fridman (58:34.480)
I mean, first off, we all have different values.
Lex Fridman (58:37.220)
And secondly, maybe if we were smarter,
Lex Fridman (58:40.440)
we would have better values.
Erik Brynjolfsson (58:41.620)
Like, I like to think that we have better values
Lex Fridman (58:44.200)
than we did in 1860 and, or in the year 200 BC
Erik Brynjolfsson (58:50.320)
on a lot of dimensions,
Lex Fridman (58:51.360)
things that we consider barbaric today.
Lex Fridman (58:53.440)
And it may be that if I thought about it more deeply,
Lex Fridman (58:56.080)
I would also be morally evolved.
Erik Brynjolfsson (58:57.400)
Maybe I'd be a vegetarian or do other things
Lex Fridman (59:00.120)
that right now, whether my future self
Erik Brynjolfsson (59:02.980)
would consider kind of immoral.
Lex Fridman (59:05.240)
So that's a tricky problem,
Erik Brynjolfsson (59:07.740)
getting the AI to do what we want,
Lex Fridman (59:11.120)
assuming it's even a friendly AI.
Erik Brynjolfsson (59:12.960)
I mean, I should probably mention
Lex Fridman (59:14.780)
there's a nontrivial other branch
Lex Fridman (59:17.100)
where we destroy ourselves, right?
Lex Fridman (59:18.720)
I mean, there's a lot of exponentially improving
Erik Brynjolfsson (59:22.040)
technologies that could be ferociously destructive,
Lex Fridman (59:26.640)
whether it's in nanotechnology or biotech
Lex Fridman (59:29.480)
and weaponized viruses, AI and other things that.
Lex Fridman (59:34.280)
nuclear weapons.
Erik Brynjolfsson (59:35.120)
Nuclear weapons, of course.
Lex Fridman (59:36.240)
The old school technology.
Erik Brynjolfsson (59:37.320)
Yeah, good old nuclear weapons that could be devastating
Lex Fridman (59:42.040)
or even existential and new things yet to be invented.
Lex Fridman (59:45.240)
So that's a branch that I think is pretty significant.
Lex Fridman (59:52.200)
And there are those who think that one of the reasons
Lex Fridman (59:54.260)
we haven't been contacted by other civilizations, right?
Lex Fridman (59:57.480)
Is that once you get to a certain level of complexity
🔗 相关节目