Elon Musk: SpaceX, Mars, Tesla Autopilot, Self-Driving, Robotics, and AI
太空与探索历史与文明技术与编程音乐与艺术政治与社会
🤖
AI 智能总结
埃隆·马斯克谈SpaceX、火星、特斯拉自动驾驶与AI
这是 Lex Fridman 与埃隆·马斯克的第三次对话，也是迄今最深入的一次。马斯克详细讨论了 SpaceX 的火星计划、特斯拉 Autopilot 的技术路线、人形机器人 Optimus 的愿景，以及他对 AI 未来的深刻担忧。
SpaceX火星特斯拉自动驾驶Optimus机器人AI安全Twitter
埃隆·马斯克是 SpaceX、特斯拉、xAI、Neuralink 和 The Boring Company 的创始人/CEO，同时拥有 X（原 Twitter）。他是当今世界最具影响力的科技企业家，致力于推动人类成为多星球物种。
📌 核心观点
- SpaceX 与火星殖民：马斯克详细阐述了为什么人类必须成为多星球物种——地球上的任何单一灾难（小行星撞击、核战争、超级病毒）都可能终结人类文明。火星是最近的可行选择，Starship 是实现这一目标的关键。
- 特斯拉 Autopilot 的纯视觉路线：马斯克坚持认为自动驾驶不需要激光雷达，因为人类只用眼睛就能驾驶，AI 也应该能做到。他认为激光雷达是「拐杖」，会阻碍真正的视觉 AI 的发展。
- Optimus 人形机器人：马斯克认为人形机器人将比电动汽车更重要，因为它可以做任何人类能做的工作。他预测 Optimus 将成为特斯拉最有价值的产品，最终可能价值数万亿美元。
- 对 AI 的担忧：马斯克是 AI 存在性风险的早期警告者，他离开 OpenAI 董事会部分原因是对其发展方向的担忧。他创立 xAI 是为了开发「最大程度追求真理」的 AI，作为对 OpenAI 的替代。
- 关于 Twitter/X 收购：马斯克解释了他收购 Twitter 的动机——他认为 Twitter 是现代公共广场，其内容审核政策威胁到了言论自由，他希望将其变成一个更公平的平台。
✨ 金句摘录
马斯克：人类必须成为多星球物种——地球上的任何单一灾难都可能终结文明，火星是我们的备份。
马斯克：激光雷达是拐杖——人类只用眼睛就能驾驶，AI 也应该能做到，这是正确的技术路线。
马斯克：Optimus 将比电动汽车更重要——一个能做任何人类工作的机器人，其价值是无限的。
📋 章节目录
暂无章节信息
🔑 关键词
donhumangotspacemarshistorygoingengineharddoingsureneuralpossiblerocketdonecarhighsaidhumanstesla
💬 精彩语录
"So I think you could actually get a much lower cost, like in optimizing cost per ton to orbit, cost per ton to Mars."
所以我认为你实际上可以获得更低的成本，比如优化每吨进入轨道的成本，每吨进入火星的成本。
— Elon Musk （2:00:05.080）
"I think in a place that is not subject to extreme natural disasters, I think nuclear power is a great way to generate electricity."
我认为在一个没有遭受极端自然灾害的地方，我认为核电是一种很好的发电方式。
— Elon Musk （2:01:34.080）
"because I just went down the total rathole of studying all of the fighters and bombers of World War II"
因为我刚刚深入研究了二战中的所有战斗机和轰炸机
— Elon Musk （1:47:58.080）
"Every time I see anything by him, I just have a stupid smile on my face because he's so excited about stuff."
每当我看到他的任何东西时，我的脸上都会露出愚蠢的微笑，因为他对东西非常兴奋。
— Elon Musk （1:58:45.080）
"And at least the impression I had was that the Soviet Union, Russia and Ukraine primarily were actually in the process of switching to Methalox."
至少我的印象是，苏联、俄罗斯和乌克兰实际上主要正在转向使用甲氧草。
— Elon Musk （1:59:35.080）
🎙️ 完整对话（2767 条）
Lex Fridman (00:00.000)
The following is a conversation with Elon Musk,
以下是与埃隆·马斯克的对话，
Lex Fridman (00:02.540)
his third time on this, The Lex Friedman Podcast.
他第三次参加莱克斯·弗里德曼播客。
Lex Fridman (00:06.980)
Yeah, make yourself comfortable.
是啊，让自己舒服点。
Lex Fridman (00:08.440)
Boo.
嘘。
Lex Fridman (00:09.280)
No, wow, okay.
不，哇，好吧。
Lex Fridman (00:10.120)
You don't do the headphone thing?
你不做耳机的事吗？
Lex Fridman (00:12.760)
No.
不。
Elon Musk (00:13.600)
Okay.
好的。
Lex Fridman (00:14.420)
I mean, how close do I need to get up to the same place?
我的意思是，我需要多近才能到达同一个地方？
Elon Musk (00:15.840)
The closer you are, the sexier you sound.
你们离得越近，听起来就越性感。
Lex Fridman (00:17.560)
Hey, babe.
嘿，宝贝。
Lex Fridman (00:18.400)
What's up?
这是怎么回事？
Lex Fridman (00:19.220)
Can't get enough of your love, baby.
无法得到足够的爱，宝贝。
Elon Musk (00:20.880)
I'm gonna clip that out
我要把它剪掉
Lex Fridman (00:25.080)
any time somebody messages me about it.
任何时候有人给我发消息。
Elon Musk (00:26.720)
If you want my body and you think I'm sexy,
如果你想要我的身体并且你认为我很性感
Lex Fridman (00:30.120)
come right out and tell me so.
快出来告诉我吧。
Elon Musk (00:33.040)
Do, do, do, do, do.
做，做，做，做，做。
Lex Fridman (00:36.000)
So good.
超好的。
Lex Fridman (00:38.840)
So good.
超好的。
Lex Fridman (00:39.680)
Okay, serious mode activate.
Elon Musk (00:41.880)
All right.
Lex Fridman (00:43.240)
Serious mode.
Elon Musk (00:44.360)
Come on, you're Russian.
Lex Fridman (00:45.200)
You can be serious.
Elon Musk (00:46.020)
Yeah, I know.
Lex Fridman (00:46.860)
Everyone's serious all the time in Russia.
Elon Musk (00:47.700)
Yeah, yeah, we'll get there.
Lex Fridman (00:50.600)
We'll get there.
Elon Musk (00:51.920)
Yeah.
Lex Fridman (00:52.760)
It's gotten soft.
Elon Musk (00:53.600)
Allow me to say that the SpaceX launch
Lex Fridman (00:57.760)
of human beings to orbit on May 30th, 2020
Elon Musk (01:02.160)
was seen by many as the first step
Lex Fridman (01:03.920)
in a new era of human space exploration.
Elon Musk (01:07.440)
These human space flight missions were a beacon of hope
Lex Fridman (01:10.620)
to me and to millions over the past two years
Elon Musk (01:12.740)
as our world has been going through
Lex Fridman (01:14.680)
one of the most difficult periods in recent human history.
Elon Musk (01:18.080)
We saw, we see the rise of division, fear, cynicism,
Lex Fridman (01:21.720)
and the loss of common humanity
Elon Musk (01:24.620)
right when it is needed most.
Lex Fridman (01:26.380)
So first, Elon, let me say thank you
Elon Musk (01:29.080)
for giving the world hope and reason
Lex Fridman (01:30.620)
to be excited about the future.
Elon Musk (01:32.520)
Oh, it's kind of easy to say.
Lex Fridman (01:34.200)
I do want to do that.
Elon Musk (01:35.480)
Humanity has obviously a lot of issues
Lex Fridman (01:38.320)
and people at times do bad things,
Lex Fridman (01:42.620)
but despite all that, I love humanity
Lex Fridman (01:47.260)
and I think we should make sure we do everything we can
Elon Musk (01:52.560)
to have a good future and an exciting future
Lex Fridman (01:54.640)
and one where that maximizes the happiness of the people.
Elon Musk (01:58.320)
Let me ask about Crew Dragon Demo 2.
Lex Fridman (02:00.960)
So that first flight with humans on board,
Lex Fridman (02:04.360)
how did you feel leading up to that launch?
Lex Fridman (02:06.160)
Were you scared?
Lex Fridman (02:07.280)
Were you excited?
Lex Fridman (02:08.260)
What was going through your mind?
Lex Fridman (02:09.640)
So much was at stake.
Lex Fridman (02:11.380)
Yeah, no, that was extremely stressful, no question.
Elon Musk (02:14.700)
We obviously could not let them down in any way.
Lex Fridman (02:19.140)
So extremely stressful, I'd say, to say the least.
Elon Musk (02:25.940)
I was confident that at the time that we launched
Lex Fridman (02:28.380)
that no one could think of anything at all to do
Elon Musk (02:34.140)
that would improve the probability of success.
Lex Fridman (02:36.500)
And we racked our brains to think of any possible way
Elon Musk (02:40.100)
to improve the probability of success.
Lex Fridman (02:41.420)
We could not think of anything more, nor could NASA.
Lex Fridman (02:44.820)
And so that's just the best that we could do.
Lex Fridman (02:48.060)
So then we went ahead and launched.
Elon Musk (02:51.860)
Now, I'm not a religious person,
Lex Fridman (02:54.340)
but I nonetheless got on my knees and prayed for that mission.
Lex Fridman (03:00.340)
Were you able to sleep?
Lex Fridman (03:01.940)
No.
Lex Fridman (03:02.780)
How did it feel when it was a success,
Lex Fridman (03:10.820)
first when the launch was a success,
Lex Fridman (03:12.780)
and when they returned back home or back to Earth?
Lex Fridman (03:16.940)
It was a great relief.
Elon Musk (03:20.380)
Yeah, for high stress situations,
Lex Fridman (03:23.480)
I find it's not so much elation as relief.
Lex Fridman (03:26.060)
And I think once, as we got more comfortable
Lex Fridman (03:33.440)
and proved out the systems,
Elon Musk (03:34.600)
because we really got to make sure everything works,
Lex Fridman (03:41.280)
it was definitely a lot more enjoyable
Elon Musk (03:43.200)
with the subsequent astronaut missions.
Lex Fridman (03:46.520)
And I thought the Inspiration mission
Elon Musk (03:50.160)
was actually a very inspiring Inspiration 4 mission.
Lex Fridman (03:53.960)
I'd encourage people to watch the Inspiration documentary
Elon Musk (03:57.040)
on Netflix, it's actually really good.
Lex Fridman (04:00.280)
And it really isn't, I was actually inspired by that.
Lex Fridman (04:03.360)
And so that one I felt I was kind of able
Lex Fridman (04:07.760)
to enjoy the actual mission
Lex Fridman (04:09.320)
and not just be super stressed all the time.
Lex Fridman (04:10.880)
So for people that somehow don't know,
Elon Musk (04:13.760)
it's the all civilian, first time all civilian
Lex Fridman (04:17.240)
out to space, out to orbit.
Elon Musk (04:19.400)
Yeah, and it was I think the highest orbit
Lex Fridman (04:22.480)
that in like, I don't know, 30 or 40 years or something.
Elon Musk (04:26.120)
The only one that was higher was the one shuttle,
Lex Fridman (04:29.480)
sorry, a Hubble servicing mission.
Lex Fridman (04:32.160)
And then before that, it would have been Apollo in 72.
Lex Fridman (04:37.840)
It's pretty wild.
Lex Fridman (04:39.180)
So it's cool, it's good.
Lex Fridman (04:40.520)
I think as a species, we want to be continuing
Elon Musk (04:46.620)
to do better and reach higher ground.
Lex Fridman (04:50.000)
And like, I think it would be tragic, extremely tragic
Elon Musk (04:52.880)
if Apollo was the high watermark for humanity,
Lex Fridman (04:57.000)
and that's as far as we ever got.
Lex Fridman (05:00.080)
And it's concerning that here we are 49 years
Lex Fridman (05:06.360)
after the last mission to the moon.
Lex Fridman (05:09.060)
And so almost half a century, and we've not been back.
Lex Fridman (05:14.060)
And that's worrying.
Elon Musk (05:16.260)
It's like, does that mean we've peaked as a civilization
Lex Fridman (05:20.280)
or what?
Lex Fridman (05:21.120)
So like, I think we got to get back to the moon
Lex Fridman (05:24.480)
and build a base there, a science base.
Elon Musk (05:27.160)
I think we could learn a lot about the nature
Lex Fridman (05:28.880)
of the universe if we have a proper science base
Elon Musk (05:31.120)
on the moon, like we have a science base in Antarctica
Lex Fridman (05:35.520)
and many other parts of the world.
Lex Fridman (05:38.080)
And so that's like, I think the next big thing,
Lex Fridman (05:41.880)
we've got to have like a serious moon base
Lex Fridman (05:45.680)
and then get people to Mars and get out there
Lex Fridman (05:49.960)
and be a spacefaring civilization.
Elon Musk (05:52.580)
I'll ask you about some of those details,
Lex Fridman (05:54.500)
but since you're so busy with the hard engineering
Elon Musk (05:57.520)
challenges of everything that's involved,
Lex Fridman (06:00.440)
are you still able to marvel at the magic of it all,
Elon Musk (06:03.000)
of space travel, of every time the rocket goes up,
Lex Fridman (06:05.880)
especially when it's a crewed mission?
Elon Musk (06:08.120)
Or are you just so overwhelmed with all the challenges
Lex Fridman (06:11.440)
that you have to solve?
Lex Fridman (06:12.560)
And actually, sort of to add to that,
Lex Fridman (06:16.280)
the reason I wanted to ask this question of May 30th,
Elon Musk (06:19.240)
it's been some time, so you can look back
Lex Fridman (06:21.840)
and think about the impact already.
Elon Musk (06:23.520)
It's already, at the time it was an engineering problem
Lex Fridman (06:26.360)
maybe, now it's becoming a historic moment.
Elon Musk (06:29.320)
Like it's a moment that, how many moments
Lex Fridman (06:31.640)
will be remembered about the 21st century?
Elon Musk (06:33.720)
To me, that or something like that,
Lex Fridman (06:37.300)
maybe Inspiration4, one of those would be remembered
Elon Musk (06:39.780)
as the early steps of a new age of space exploration.
Lex Fridman (06:44.200)
Yeah, I mean, during the launches itself,
Lex Fridman (06:46.640)
so I mean, I think maybe some people will know,
Lex Fridman (06:49.120)
but a lot of people don't know,
Elon Musk (06:50.200)
is like I'm actually the chief engineer of SpaceX,
Lex Fridman (06:52.120)
so I've signed off on pretty much all the design decisions.
Lex Fridman (06:59.040)
And so if there's something that goes wrong
Lex Fridman (07:03.120)
with that vehicle, it's fundamentally my fault, so.
Lex Fridman (07:09.960)
So I'm really just thinking about all the things that,
Lex Fridman (07:13.800)
like, so when I see the rocket,
Elon Musk (07:15.600)
I see all the things that could go wrong
Lex Fridman (07:17.320)
and the things that could be better
Lex Fridman (07:18.520)
and the same with the Dragon spacecraft.
Lex Fridman (07:21.480)
It's like, other people will see,
Elon Musk (07:23.440)
oh, this is a spacecraft or a rocket
Lex Fridman (07:25.600)
and this looks really cool.
Elon Musk (07:27.040)
I'm like, I've like a readout of like,
Lex Fridman (07:29.600)
these are the risks, these are the problems,
Elon Musk (07:32.600)
that's what I see.
Lex Fridman (07:33.840)
Like, choo choo choo choo choo choo choo.
Lex Fridman (07:36.320)
So it's not what other people see
Lex Fridman (07:38.840)
when they see the product, you know.
Lex Fridman (07:40.400)
So let me ask you then to analyze Starship
Lex Fridman (07:43.040)
in that same way.
Elon Musk (07:44.360)
I know you have, you'll talk about,
Lex Fridman (07:46.520)
in more detail about Starship in the near future, perhaps.
Elon Musk (07:49.560)
Yeah, we can talk about it now if you want.
Lex Fridman (07:51.800)
But just in that same way, like you said,
Elon Musk (07:54.120)
you see, when you see a rocket,
Lex Fridman (07:57.620)
you see sort of a list of risks.
Elon Musk (07:59.160)
In that same way, you said that Starship
Lex Fridman (08:01.040)
is a really hard problem.
Lex Fridman (08:03.240)
So there's many ways I can ask this,
Lex Fridman (08:05.480)
but if you magically could solve one problem perfectly,
Elon Musk (08:09.560)
one engineering problem perfectly,
Lex Fridman (08:11.740)
which one would it be?
Lex Fridman (08:13.000)
On Starship?
Lex Fridman (08:14.120)
On, sorry, on Starship.
Lex Fridman (08:15.540)
So is it maybe related to the efficiency,
Lex Fridman (08:18.040)
the engine, the weight of the different components,
Elon Musk (08:21.360)
the complexity of various things,
Lex Fridman (08:22.680)
maybe the controls of the crazy thing it has to do to land?
Elon Musk (08:26.720)
No, it's actually by far the biggest thing
Lex Fridman (08:30.080)
absorbing my time is engine production.
Elon Musk (08:35.080)
Not the design of the engine,
Lex Fridman (08:36.480)
but I've often said prototypes are easy, production is hard.
Lex Fridman (08:45.360)
So we have the most advanced rocket engine
Lex Fridman (08:48.680)
that's ever been designed.
Elon Musk (08:52.720)
Cause I'd say currently the best rocket engine ever
Lex Fridman (08:55.260)
is probably the RD180 or RD170,
Elon Musk (09:00.320)
that's the Russian engine basically.
Lex Fridman (09:03.640)
And still, I think an engine should only count
Elon Musk (09:06.760)
if it's gotten something to orbit.
Lex Fridman (09:09.000)
So our engine has not gotten anything to orbit yet,
Lex Fridman (09:12.720)
but it is, it's the first engine
Lex Fridman (09:14.600)
that's actually better than the Russian RD engines,
Elon Musk (09:20.240)
which were amazing design.
Lex Fridman (09:22.680)
So you're talking about Raptor engine.
Lex Fridman (09:24.120)
What makes it amazing?
Lex Fridman (09:25.520)
What are the different aspects of it that make it,
Elon Musk (09:29.280)
like what are you the most excited about
Lex Fridman (09:31.920)
if the whole thing works in terms of efficiency,
Lex Fridman (09:35.080)
all those kinds of things?
Lex Fridman (09:37.120)
Well, it's, the Raptor is a full flow
Elon Musk (09:42.000)
staged combustion engine
Lex Fridman (09:46.360)
and it's operating at a very high chamber pressure.
Lex Fridman (09:50.840)
So one of the key figures of merit, perhaps the key figure
Lex Fridman (09:54.800)
of merit is what is the chamber pressure
Lex Fridman (09:58.680)
at which the rocket engine can operate?
Lex Fridman (10:00.800)
That's the combustion chamber pressure.
Lex Fridman (10:03.040)
So Raptor is designed to operate at 300 bar,
Lex Fridman (10:06.960)
possibly maybe higher, that's 300 atmospheres.
Lex Fridman (10:10.320)
So the record right now for operational engine
Lex Fridman (10:15.880)
is the RD engine that I mentioned, the Russian RD,
Elon Musk (10:17.880)
which is I believe around 267 bar.
Lex Fridman (10:22.560)
And the difficulty of the chamber pressure
Elon Musk (10:25.840)
is increases on a nonlinear basis.
Lex Fridman (10:27.860)
So 10% more chamber pressure is more like 50% more difficult.
Lex Fridman (10:36.960)
But that chamber pressure is,
Lex Fridman (10:38.640)
that is what allows you to get a very high power density
Elon Musk (10:44.400)
for the engine.
Lex Fridman (10:45.220)
So enabling a very high thrust to weight ratio
Lex Fridman (10:53.300)
and a very high specific impulse.
Lex Fridman (10:57.020)
So specific impulse is like a measure of the efficiency
Elon Musk (10:59.820)
of a rocket engine.
Lex Fridman (11:01.700)
It's really the effect of exhaust velocity
Elon Musk (11:07.820)
of the gas coming out of the engine.
Lex Fridman (11:10.120)
So with a very high chamber pressure,
Elon Musk (11:17.160)
you can have a compact engine
Lex Fridman (11:21.600)
that nonetheless has a high expansion ratio,
Elon Musk (11:24.000)
which is the ratio between the exit nozzle and the throat.
Lex Fridman (11:31.800)
So you see a rocket engine's got sort of like a hourglass
Elon Musk (11:35.640)
shape, it's like a chamber and then it necks down
Lex Fridman (11:38.120)
and there's a nozzle and the ratio of the exit diameter
Elon Musk (11:41.880)
to the throat is the expansion ratio.
Lex Fridman (11:45.960)
So why is it such a hard engine to manufacture at scale?
Elon Musk (11:51.400)
It's very complex.
Lex Fridman (11:53.120)
So a lot of, what is complexity mean here?
Elon Musk (11:55.200)
There's a lot of components involved.
Lex Fridman (11:56.760)
There's a lot of components and a lot of unique materials
Elon Musk (12:02.400)
that, so we had to invent several alloys
Lex Fridman (12:07.040)
that don't exist in order to make this engine work.
Elon Musk (12:11.520)
It's a materials problem too.
Lex Fridman (12:14.560)
It's a materials problem and in a staged combustion,
Elon Musk (12:19.760)
a full flow staged combustion,
Lex Fridman (12:21.200)
there are many feedback loops in the system.
Lex Fridman (12:24.160)
So basically you've got propellants and hot gas
Lex Fridman (12:31.820)
flowing simultaneously to so many different places
Elon Musk (12:36.820)
on the engine and they all have a recursive effect
Lex Fridman (12:42.740)
on each other.
Lex Fridman (12:43.700)
So you change one thing here, it has a recursive effect
Lex Fridman (12:45.780)
here, it changes something over there
Lex Fridman (12:47.420)
and it's quite hard to control.
Lex Fridman (12:52.740)
Like there's a reason no one's made this before.
Lex Fridman (12:58.700)
And the reason we're doing a staged combustion full flow
Lex Fridman (13:03.700)
is because it has the highest possible efficiency.
Lex Fridman (13:12.740)
So in order to make a fully reusable rocket,
Lex Fridman (13:19.380)
which that's the really the holy grail of orbital rocketry.
Elon Musk (13:25.340)
You have to have, everything's gotta be the best.
Lex Fridman (13:28.420)
It's gotta be the best engine, the best airframe,
Elon Musk (13:30.900)
the best heat shield, extremely light avionics,
Lex Fridman (13:38.100)
very clever control mechanisms.
Elon Musk (13:40.620)
You've got to shed mass in any possible way that you can.
Lex Fridman (13:45.100)
For example, we are, instead of putting landing legs
Elon Musk (13:47.620)
on the booster and ship, we are gonna catch them
Lex Fridman (13:49.580)
with a tower to save the weight of the landing legs.
Lex Fridman (13:53.220)
So that's like, I mean, we're talking about catching
Lex Fridman (13:58.220)
the largest flying object ever made with,
Elon Musk (14:03.340)
on a giant tower with chopstick arms.
Lex Fridman (14:06.660)
It's like Karate Kid with the fly, but much bigger.
Elon Musk (14:12.060)
I mean, pulling something like that home.
Lex Fridman (14:12.900)
This probably won't work the first time.
Elon Musk (14:17.620)
Anyway, so this is bananas, this is banana stuff.
Lex Fridman (14:19.820)
So you mentioned that you doubt, well, not you doubt,
Lex Fridman (14:23.100)
but there's days or moments when you doubt
Lex Fridman (14:26.460)
that this is even possible.
Elon Musk (14:28.340)
It's so difficult.
Lex Fridman (14:30.120)
The possible part is, well, at this point,
Elon Musk (14:35.420)
I think we will get Starship to work.
Lex Fridman (14:41.380)
There's a question of timing.
Lex Fridman (14:42.860)
How long will it take us to do this?
Lex Fridman (14:45.460)
How long will it take us to actually achieve
Lex Fridman (14:47.700)
full and rapid reusability?
Lex Fridman (14:50.500)
Cause it will take probably many launches
Elon Musk (14:52.680)
before we are able to have full and rapid reusability.
Lex Fridman (14:57.660)
But I can say that the physics pencils out.
Elon Musk (15:01.540)
Like we're not, like at this point,
Lex Fridman (15:06.620)
I'd say we're confident that, like let's say,
Elon Musk (15:10.100)
I'm very confident success is in the set
Lex Fridman (15:12.620)
of all possible outcomes.
Elon Musk (15:13.820)
For a while there, I was not convinced
Lex Fridman (15:18.340)
that success was in the set of possible outcomes,
Elon Musk (15:20.820)
which is very important actually.
Lex Fridman (15:23.820)
But so we're saying there's a chance.
Elon Musk (15:29.700)
I'm saying there's a chance, exactly.
Lex Fridman (15:33.260)
Just not sure how long it will take.
Elon Musk (15:38.260)
We have a very talented team.
Lex Fridman (15:39.580)
They're working night and day to make it happen.
Lex Fridman (15:43.420)
And like I said, the critical thing to achieve
Lex Fridman (15:48.260)
for the revolution in space flight
Lex Fridman (15:49.540)
and for humanity to be a spacefaring civilization
Lex Fridman (15:52.020)
is to have a fully and rapidly reusable rocket,
Elon Musk (15:54.380)
orbital rocket.
Lex Fridman (15:56.500)
There's not even been any orbital rocket
Elon Musk (15:58.820)
that's been fully reusable ever.
Lex Fridman (16:00.100)
And this has always been the holy grail of rocketry.
Lex Fridman (16:06.020)
And many smart people, very smart people,
Lex Fridman (16:09.700)
have tried to do this before and they've not succeeded.
Elon Musk (16:12.700)
So, cause it's such a hard problem.
Lex Fridman (16:16.100)
What's your source of belief in situations like this?
Elon Musk (16:21.220)
When the engineering problem is so difficult,
Lex Fridman (16:23.580)
there's a lot of experts, many of whom you admire,
Elon Musk (16:27.500)
who have failed in the past.
Lex Fridman (16:29.220)
And a lot of people, a lot of experts,
Elon Musk (16:37.220)
maybe journalists, all the kind of,
Lex Fridman (16:38.860)
the public in general have a lot of doubt
Elon Musk (16:40.900)
about whether it's possible.
Lex Fridman (16:42.820)
And you yourself know that even if it's a non null set,
Elon Musk (16:47.340)
non empty set of success, it's still unlikely
Lex Fridman (16:50.580)
or very difficult.
Elon Musk (16:52.140)
Like where do you go to, both personally,
Lex Fridman (16:55.780)
intellectually as an engineer, as a team,
Elon Musk (16:58.920)
like for source of strength needed
Lex Fridman (17:00.860)
to sort of persevere through this.
Lex Fridman (17:02.560)
And to keep going with the project, take it to completion.
Lex Fridman (17:18.560)
A source of strength.
Elon Musk (17:19.800)
That's really not how I think about things.
Lex Fridman (17:23.600)
I mean, for me, it's simply this,
Elon Musk (17:25.000)
this is something that is important to get done
Lex Fridman (17:28.080)
and we should just keep doing it or die trying.
Lex Fridman (17:32.480)
And I don't need a source of strength.
Lex Fridman (17:35.920)
So quitting is not even like...
Elon Musk (17:39.000)
That's not, it's not in my nature.
Lex Fridman (17:41.520)
And I don't care about optimism or pessimism.
Elon Musk (17:46.240)
Fuck that, we're gonna get it done.
Lex Fridman (17:47.880)
Gonna get it done.
Lex Fridman (17:51.720)
Can you then zoom back in to specific problems
Lex Fridman (17:55.160)
with Starship or any engineering problems you work on?
Lex Fridman (17:58.300)
Can you try to introspect your particular
Lex Fridman (18:00.900)
biological neural network, your thinking process
Lex Fridman (18:03.520)
and describe how you think through problems,
Lex Fridman (18:06.120)
the different engineering and design problems?
Lex Fridman (18:07.960)
Is there like a systematic process?
Lex Fridman (18:10.080)
You've spoken about first principles thinking,
Lex Fridman (18:11.880)
but is there a kind of process to it?
Lex Fridman (18:14.140)
Well, yeah, like saying like physics is a law
Lex Fridman (18:19.200)
and everything else is a recommendation.
Lex Fridman (18:21.640)
Like I've met a lot of people that can break the law,
Lex Fridman (18:23.200)
but I haven't met anyone who could break physics.
Lex Fridman (18:25.700)
So the first, for any kind of technology problem,
Elon Musk (18:32.520)
you have to sort of just make sure
Lex Fridman (18:34.560)
you're not violating physics.
Lex Fridman (18:39.960)
And first principles analysis, I think,
Lex Fridman (18:45.120)
is something that can be applied to really any walk of life,
Elon Musk (18:49.200)
anything really.
Lex Fridman (18:50.040)
It's really just saying, let's boil something down
Elon Musk (18:54.680)
to the most fundamental principles.
Lex Fridman (18:58.440)
The things that we are most confident are true
Elon Musk (19:00.840)
at a foundational level.
Lex Fridman (19:02.480)
And that sets your axiomatic base,
Lex Fridman (19:05.040)
and then you reason up from there,
Lex Fridman (19:07.000)
and then you cross check your conclusion
Elon Musk (19:09.040)
against the axiomatic truths.
Lex Fridman (19:13.680)
So some basics in physics would be like,
Elon Musk (19:18.120)
are you violating conservation of energy or momentum
Lex Fridman (19:20.280)
or something like that?
Elon Musk (19:21.400)
Then it's not gonna work.
Lex Fridman (19:29.560)
So that's just to establish, is it possible?
Lex Fridman (19:32.960)
And another good physics tool
Lex Fridman (19:34.760)
is thinking about things in the limit.
Elon Musk (19:36.480)
If you take a particular thing
Lex Fridman (19:38.360)
and you scale it to a very large number
Lex Fridman (19:41.680)
or to a very small number, how do things change?
Lex Fridman (19:45.960)
Both like in number of things you manufacture
Lex Fridman (19:48.920)
or something like that, and then in time?
Lex Fridman (19:51.640)
Yeah, like let's say take an example of like manufacturing,
Elon Musk (19:55.920)
which I think is just a very underrated problem.
Lex Fridman (1:00:02.700)
So, I'm not saying that it's like the ideal system
Elon Musk (1:00:06.340)
for a currency, but I think it actually is
Lex Fridman (1:00:09.500)
just fundamentally better than anything else I've seen
Elon Musk (1:00:13.460)
just by accident, so.
Lex Fridman (1:00:16.420)
I like how you said around 2008.
Elon Musk (1:00:19.740)
So, you're not, you know, some people suggested
Lex Fridman (1:00:23.420)
you might be Satoshi Nakamoto.
Elon Musk (1:00:24.820)
You've previously said you're not.
Lex Fridman (1:00:26.340)
Let me ask.
Elon Musk (1:00:27.180)
You're not for sure.
Lex Fridman (1:00:28.780)
Would you tell us if you were?
Elon Musk (1:00:30.140)
Yes.
Lex Fridman (1:00:30.980)
Okay.
Lex Fridman (1:00:33.460)
Do you think it's a feature or a bug
Lex Fridman (1:00:34.780)
that he's anonymous or she or they?
Elon Musk (1:00:38.980)
It's an interesting kind of quirk of human history
Lex Fridman (1:00:41.700)
that there is a particular technology
Elon Musk (1:00:43.580)
that is a completely anonymous inventor
Lex Fridman (1:00:46.140)
or creator.
Elon Musk (1:01:03.380)
Well, I mean, you can look at the evolution of ideas
Lex Fridman (1:01:10.060)
before the launch of Bitcoin
Lex Fridman (1:01:11.860)
and see who wrote, you know, about those ideas.
Lex Fridman (1:01:19.540)
And then, like, I don't know exactly,
Elon Musk (1:01:21.900)
obviously I don't know who created Bitcoin
Lex Fridman (1:01:24.180)
for practical purposes,
Lex Fridman (1:01:25.180)
but the evolution of ideas is pretty clear for that.
Lex Fridman (1:01:28.980)
And like, it seems as though like Nick Szabo
Elon Musk (1:01:31.940)
is probably more than anyone else responsible
Lex Fridman (1:01:35.260)
for the evolution of those ideas.
Elon Musk (1:01:37.100)
So, he claims not to be Satoshi Nakamoto,
Lex Fridman (1:01:41.100)
but I'm not sure that's neither here nor there,
Lex Fridman (1:01:44.460)
but he seems to be the one more responsible
Lex Fridman (1:01:47.700)
for the ideas behind Bitcoin than anyone else.
Elon Musk (1:01:50.340)
So, it's not perhaps like singular figures
Lex Fridman (1:01:52.820)
aren't even as important as the figures involved
Elon Musk (1:01:55.740)
in the evolution of ideas, the Leto thing, so.
Lex Fridman (1:01:58.100)
Yeah.
Elon Musk (1:01:58.940)
Yeah, it's, you know, perhaps it's sad
Lex Fridman (1:02:02.260)
to think about history,
Lex Fridman (1:02:03.100)
but maybe most names will be forgotten anyway.
Lex Fridman (1:02:06.340)
What is a name anyway?
Elon Musk (1:02:07.460)
It's a name attached to an idea.
Lex Fridman (1:02:11.260)
What does it even mean, really?
Elon Musk (1:02:13.700)
I think Shakespeare had a thing about roses and stuff,
Lex Fridman (1:02:16.260)
whatever he said.
Elon Musk (1:02:17.220)
A rose by any other name, it smells sweet.
Lex Fridman (1:02:22.420)
I got Elon to quote Shakespeare.
Elon Musk (1:02:24.340)
I feel like I accomplished something today.
Lex Fridman (1:02:26.900)
Shall I compare thee to a summer's day?
Lex Fridman (1:02:28.860)
What?
Lex Fridman (1:02:30.820)
I'm gonna clip that out.
Elon Musk (1:02:31.820)
I said it to people.
Lex Fridman (1:02:33.980)
Not more temperate and more fair.
Elon Musk (1:02:39.060)
Autopilot.
Lex Fridman (1:02:40.620)
Tesla autopilot.
Elon Musk (1:02:46.140)
Tesla autopilot has been through an incredible journey
Lex Fridman (1:02:48.500)
over the past six years,
Elon Musk (1:02:50.540)
or perhaps even longer in the minds of,
Lex Fridman (1:02:52.780)
in your mind and the minds of many involved.
Elon Musk (1:02:57.020)
Yeah, I think that's where we first like connected really
Lex Fridman (1:02:58.980)
was the autopilot stuff, autonomy and.
Elon Musk (1:03:01.900)
The whole journey was incredible to me to watch.
Lex Fridman (1:03:05.220)
I was, because I knew, well, part of it is I was at MIT
Lex Fridman (1:03:10.340)
and I knew the difficulty of computer vision.
Lex Fridman (1:03:13.180)
And I knew the whole, I had a lot of colleagues and friends
Elon Musk (1:03:15.780)
about the DARPA challenge and knew how difficult it is.
Lex Fridman (1:03:18.420)
And so there was a natural skepticism
Elon Musk (1:03:20.140)
when I first drove a Tesla with the initial system
Lex Fridman (1:03:23.700)
based on Mobileye.
Elon Musk (1:03:25.140)
I thought there's no way, so first when I got in,
Lex Fridman (1:03:28.900)
I thought there's no way this car could maintain,
Elon Musk (1:03:32.580)
like stay in the lane and create a comfortable experience.
Lex Fridman (1:03:35.900)
So my intuition initially was that the lane keeping problem
Elon Musk (1:03:39.500)
is way too difficult to solve.
Lex Fridman (1:03:41.780)
Oh, lane keeping, yeah, that's relatively easy.
Elon Musk (1:03:43.900)
Well, like, but not this, but solve in the way
Lex Fridman (1:03:47.820)
that we just, we talked about previous is prototype
Elon Musk (1:03:50.860)
versus a thing that actually creates a pleasant experience
Lex Fridman (1:03:54.380)
over hundreds of thousands of miles or millions.
Elon Musk (1:03:57.460)
Yeah, so we had to wrap a lot of code
Lex Fridman (1:04:00.460)
around the Mobileye thing.
Elon Musk (1:04:01.740)
It doesn't just work by itself.
Lex Fridman (1:04:04.380)
I mean, that's part of the story
Elon Musk (1:04:06.420)
of how you approach things sometimes.
Lex Fridman (1:04:07.980)
Sometimes you do things from scratch.
Elon Musk (1:04:09.680)
Sometimes at first you kind of see what's out there
Lex Fridman (1:04:12.340)
and then you decide to do from scratch.
Elon Musk (1:04:14.340)
That was one of the boldest decisions I've seen
Lex Fridman (1:04:17.180)
is both in the hardware and the software
Elon Musk (1:04:18.820)
to decide to eventually go from scratch.
Lex Fridman (1:04:21.020)
I thought, again, I was skeptical
Elon Musk (1:04:22.700)
of whether that's going to be able to work out
Lex Fridman (1:04:24.500)
because it's such a difficult problem.
Lex Fridman (1:04:26.860)
And so it was an incredible journey
Lex Fridman (1:04:28.900)
what I see now with everything,
Elon Musk (1:04:31.460)
the hardware, the compute, the sensors,
Lex Fridman (1:04:33.220)
the things I maybe care and love about most
Elon Musk (1:04:37.300)
is the stuff that Andre Karpathy is leading
Lex Fridman (1:04:40.060)
with the data set selection,
Elon Musk (1:04:41.740)
the whole data engine process,
Lex Fridman (1:04:43.100)
the neural network architectures,
Elon Musk (1:04:45.020)
the way that's in the real world,
Lex Fridman (1:04:47.300)
that network is tested, validated,
Elon Musk (1:04:49.380)
all the different test sets,
Lex Fridman (1:04:52.340)
versus the ImageNet model of computer vision,
Elon Musk (1:04:54.740)
like what's in academia is like real world
Lex Fridman (1:04:58.340)
artificial intelligence.
Lex Fridman (1:05:01.340)
And Andre's awesome and obviously plays an important role,
Lex Fridman (1:05:04.220)
but we have a lot of really talented people driving things.
Lex Fridman (1:05:09.540)
And Ashok is actually the head of autopilot engineering.
Lex Fridman (1:05:14.820)
Andre's the director of AI.
Elon Musk (1:05:16.380)
AI stuff, yeah, yeah.
Lex Fridman (1:05:17.700)
So yeah, I'm aware that there's an incredible team
Elon Musk (1:05:20.580)
of just a lot going on.
Lex Fridman (1:05:22.060)
Yeah, obviously people will give me too much credit
Lex Fridman (1:05:26.060)
and they'll give Andre too much credit, so.
Lex Fridman (1:05:28.700)
And people should realize how much is going on
Elon Musk (1:05:31.700)
under the hood.
Lex Fridman (1:05:32.540)
Yeah, it's just a lot of really talented people.
Elon Musk (1:05:36.540)
The Tesla Autopilot AI team is extremely talented.
Lex Fridman (1:05:40.020)
It's like some of the smartest people in the world.
Lex Fridman (1:05:43.700)
So yeah, we're getting it done.
Lex Fridman (1:05:45.060)
What are some insights you've gained
Elon Musk (1:05:47.660)
over those five, six years of autopilot
Lex Fridman (1:05:51.300)
about the problem of autonomous driving?
Lex Fridman (1:05:54.260)
So you leaped in having some sort of
Lex Fridman (1:05:58.580)
first principles kinds of intuitions,
Lex Fridman (1:06:00.820)
but nobody knows how difficult the problem, like the problem.
Lex Fridman (1:06:05.340)
I thought the self driving problem would be hard,
Lex Fridman (1:06:07.140)
but it was harder than I thought.
Lex Fridman (1:06:08.980)
It's not like I thought it would be easy.
Elon Musk (1:06:09.980)
I thought it would be very hard,
Lex Fridman (1:06:10.800)
but it was actually way harder than even that.
Elon Musk (1:06:14.220)
So, I mean, what it comes down to at the end of the day
Lex Fridman (1:06:17.060)
is to solve self driving,
Elon Musk (1:06:18.760)
you basically need to recreate what humans do to drive,
Lex Fridman (1:06:28.680)
which is humans drive with optical sensors,
Elon Musk (1:06:31.480)
eyes and biological neural nets.
Lex Fridman (1:06:34.880)
And so in order to,
Elon Musk (1:06:36.420)
that's how the entire road system is designed to work
Lex Fridman (1:06:39.120)
with basically passive optical and neural nets,
Elon Musk (1:06:45.360)
biologically.
Lex Fridman (1:06:46.200)
And now that we need to,
Lex Fridman (1:06:47.920)
so for actually for full self driving to work,
Lex Fridman (1:06:50.080)
we have to recreate that in digital form.
Lex Fridman (1:06:52.960)
So we have to, that means cameras with advanced neural nets
Lex Fridman (1:07:01.520)
in silicon form, and then it will obviously solve
Elon Musk (1:07:06.400)
for full self driving.
Lex Fridman (1:07:08.000)
That's the only way.
Elon Musk (1:07:09.000)
I don't think there's any other way.
Lex Fridman (1:07:10.320)
But the question is what aspects of human nature
Lex Fridman (1:07:12.920)
do you have to encode into the machine, right?
Lex Fridman (1:07:15.560)
Do you have to solve the perception problem, like detect?
Lex Fridman (1:07:18.720)
And then you first realize
Lex Fridman (1:07:21.320)
what is the perception problem for driving,
Elon Musk (1:07:23.080)
like all the kinds of things you have to be able to see,
Lex Fridman (1:07:25.400)
like what do we even look at when we drive?
Elon Musk (1:07:27.900)
There's, I just recently heard Andre talked about MIT
Lex Fridman (1:07:32.440)
about car doors.
Elon Musk (1:07:33.720)
I think it was the world's greatest talk of all time
Lex Fridman (1:07:36.080)
about car doors, the fine details of car doors.
Lex Fridman (1:07:41.380)
Like what is even an open car door, man?
Lex Fridman (1:07:44.440)
So like the ontology of that,
Elon Musk (1:07:46.880)
that's a perception problem.
Lex Fridman (1:07:48.000)
We humans solve that perception problem,
Lex Fridman (1:07:49.880)
and Tesla has to solve that problem.
Lex Fridman (1:07:51.640)
And then there's the control and the planning
Elon Musk (1:07:53.380)
coupled with the perception.
Lex Fridman (1:07:54.980)
You have to figure out like what's involved in driving,
Elon Musk (1:07:58.280)
like especially in all the different edge cases.
Lex Fridman (1:08:02.320)
And then, I mean, maybe you can comment on this,
Lex Fridman (1:08:06.540)
how much game theoretic kind of stuff needs to be involved
Lex Fridman (1:08:10.920)
at a four way stop sign.
Elon Musk (1:08:12.700)
As humans, when we drive, our actions affect the world.
Lex Fridman (1:08:18.060)
Like it changes how others behave.
Elon Musk (1:08:20.780)
Most autonomous driving, if you,
Lex Fridman (1:08:23.340)
you're usually just responding to the scene
Elon Musk (1:08:27.420)
as opposed to like really asserting yourself in the scene.
Lex Fridman (1:08:31.360)
Do you think?
Elon Musk (1:08:33.140)
I think these sort of control logic conundrums
Lex Fridman (1:08:37.580)
are not the hard part.
Elon Musk (1:08:39.340)
The, you know, let's see.
Lex Fridman (1:08:45.540)
What do you think is the hard part
Lex Fridman (1:08:46.860)
in this whole beautiful, complex problem?
Lex Fridman (1:08:50.580)
So it's a lot of freaking software, man.
Elon Musk (1:08:52.980)
A lot of smart lines of code.
Lex Fridman (1:08:57.300)
For sure, in order to have,
Elon Musk (1:09:01.340)
create an accurate vector space.
Lex Fridman (1:09:03.980)
So like you're coming from image space,
Elon Musk (1:09:08.280)
which is like this flow of photons,
Lex Fridman (1:09:12.540)
you're going to the camera, cameras,
Lex Fridman (1:09:14.380)
and then you have this massive bitstream
Lex Fridman (1:09:21.220)
in image space, and then you have to effectively compress
Elon Musk (1:09:29.580)
a massive bitstream corresponding to photons
Lex Fridman (1:09:34.580)
that knocked off an electron in a camera sensor
Lex Fridman (1:09:38.500)
and turn that bitstream into a vector space.
Lex Fridman (1:09:44.860)
By vector space, I mean like, you know,
Elon Musk (1:09:47.860)
you've got cars and humans and lane lines and curves
Lex Fridman (1:09:55.060)
and traffic lights and that kind of thing.
Elon Musk (1:09:59.380)
Once you've got all of that in your head,
Lex Fridman (1:10:03.180)
once you have an accurate vector space,
Elon Musk (1:10:08.500)
the control problem is similar to that of a video game,
Lex Fridman (1:10:11.680)
like a Grand Theft Auto of Cyberpunk,
Elon Musk (1:10:14.100)
if you have accurate vector space.
Lex Fridman (1:10:16.260)
It's the control problem is,
Elon Musk (1:10:18.300)
I wouldn't say it's trivial, it's not trivial,
Lex Fridman (1:10:20.300)
but it's not like some insurmountable thing.
Elon Musk (1:10:29.020)
Having an accurate vector space is very difficult.
Lex Fridman (1:10:32.140)
Yeah, I think we humans don't give enough respect
Elon Musk (1:10:35.540)
to how incredible the human perception system is
Lex Fridman (1:10:37.900)
to mapping the raw photons to the vector space
Elon Musk (1:10:42.460)
representation in our heads.
Lex Fridman (1:10:44.660)
Your brain is doing an incredible amount of processing
Lex Fridman (1:10:48.360)
and giving you an image that is a very cleaned up image.
Lex Fridman (1:10:51.360)
Like when we look around here, we see,
Elon Musk (1:10:53.380)
like you see color in the corners of your eyes,
Lex Fridman (1:10:55.340)
but actually your eyes have very few cones,
Elon Musk (1:10:59.420)
like cone receptors in the peripheral vision.
Lex Fridman (1:11:02.260)
Your eyes are painting color in the peripheral vision.
Elon Musk (1:11:05.660)
You don't realize it,
Lex Fridman (1:11:06.500)
but their eyes are actually painting color
Lex Fridman (1:11:09.060)
and your eyes also have like this blood vessels
Lex Fridman (1:11:12.260)
and all sorts of gnarly things and there's a blind spot,
Lex Fridman (1:11:14.660)
but do you see your blind spot?
Lex Fridman (1:11:16.380)
No, your brain is painting in the missing, the blind spot.
Elon Musk (1:11:21.180)
You're gonna do these things online where you look here
Lex Fridman (1:11:24.980)
and look at this point and then look at this point
Lex Fridman (1:11:27.380)
and if it's in your blind spot,
Lex Fridman (1:11:30.460)
your brain will just fill in the missing bits.
Elon Musk (1:11:33.660)
The peripheral vision is so cool.
Lex Fridman (1:11:35.500)
It makes you realize all the illusions for vision sciences,
Lex Fridman (1:11:38.060)
so it makes you realize just how incredible the brain is.
Lex Fridman (1:11:40.620)
The brain is doing crazy amount of post processing
Elon Musk (1:11:42.660)
on the vision signals for your eyes.
Lex Fridman (1:11:45.820)
It's insane.
Lex Fridman (1:11:49.180)
And then even once you get all those vision signals,
Lex Fridman (1:11:51.940)
your brain is constantly trying to forget
Elon Musk (1:11:56.260)
as much as possible.
Lex Fridman (1:11:57.660)
So human memory is,
Elon Musk (1:11:59.500)
perhaps the weakest thing about the brain is memory.
Lex Fridman (1:12:01.900)
So because memory is so expensive to our brain
Lex Fridman (1:12:05.260)
and so limited,
Lex Fridman (1:12:06.660)
your brain is trying to forget as much as possible
Lex Fridman (1:12:09.740)
and distill the things that you see
Lex Fridman (1:12:12.140)
into the smallest amounts of information possible.
Lex Fridman (1:12:16.500)
So your brain is trying to not just get to a vector space,
Lex Fridman (1:12:19.340)
but get to a vector space that is the smallest possible
Elon Musk (1:12:22.860)
vector space of only relevant objects.
Lex Fridman (1:12:26.620)
And I think you can sort of look inside your brain
Elon Musk (1:12:29.860)
or at least I can,
Lex Fridman (1:12:31.540)
when you drive down the road and try to think about
Lex Fridman (1:12:35.380)
what your brain is actually doing consciously.
Lex Fridman (1:12:38.940)
And it's like you'll see a car,
Elon Musk (1:12:44.540)
because you don't have cameras,
Lex Fridman (1:12:46.860)
I don't have eyes in the back of your head or a side.
Lex Fridman (1:12:48.940)
So you basically have like two cameras on a slow gimbal.
Lex Fridman (1:13:00.460)
And eyesight is not that great.
Elon Musk (1:13:01.700)
Okay, human eyes are like,
Lex Fridman (1:13:04.300)
and people are constantly distracted
Lex Fridman (1:13:05.700)
and thinking about things and texting
Lex Fridman (1:13:07.180)
and doing all sorts of things they shouldn't do in a car,
Elon Musk (1:13:09.260)
changing the radio station.
Lex Fridman (1:13:10.940)
So having arguments is like,
Lex Fridman (1:13:15.060)
so when's the last time you looked right and left
Lex Fridman (1:13:22.540)
and rearward, or even diagonally forward
Lex Fridman (1:13:27.060)
to actually refresh your vector space?
Lex Fridman (1:13:30.140)
So you're glancing around and what your mind is doing
Elon Musk (1:13:32.740)
is trying to distill the relevant vectors,
Lex Fridman (1:13:37.300)
basically objects with a position and motion.
Lex Fridman (1:13:40.220)
And then editing that down to the least amount
Lex Fridman (1:13:48.140)
that's necessary for you to drive.
Elon Musk (1:13:49.940)
It does seem to be able to edit it down
Lex Fridman (1:13:53.260)
or compress it even further into things like concepts.
Lex Fridman (1:13:55.780)
So it's not, it's like it goes beyond,
Lex Fridman (1:13:57.660)
the human mind seems to go sometimes beyond vector space
Elon Musk (1:14:01.260)
to sort of space of concepts to where you'll see a thing.
Lex Fridman (1:14:05.080)
It's no longer represented spatially somehow.
Elon Musk (1:14:07.520)
It's almost like a concept that you should be aware of.
Lex Fridman (1:14:10.060)
Like if this is a school zone,
Elon Musk (1:14:12.300)
you'll remember that as a concept,
Lex Fridman (1:14:14.940)
which is a weird thing to represent,
Lex Fridman (1:14:16.420)
but perhaps for driving,
Lex Fridman (1:14:17.480)
you don't need to fully represent those things.
Elon Musk (1:14:20.460)
Or maybe you get those kind of indirectly.
Lex Fridman (1:14:25.860)
You need to establish vector space
Lex Fridman (1:14:27.740)
and then actually have predictions for those vector spaces.
Lex Fridman (1:14:32.740)
So if you drive past, say a bus and you see that there's people,
Elon Musk (1:14:47.340)
before you drove past the bus,
Lex Fridman (1:14:48.500)
you saw people crossing or some,
Elon Musk (1:14:50.580)
just imagine there's like a large truck
Lex Fridman (1:14:52.820)
or something blocking site.
Lex Fridman (1:14:55.500)
But before you came up to the truck,
Lex Fridman (1:14:57.420)
you saw that there were some kids about to cross the road
Elon Musk (1:15:00.700)
in front of the truck.
Lex Fridman (1:15:01.540)
Now you can no longer see the kids,
Lex Fridman (1:15:03.260)
but you would now know, okay,
Lex Fridman (1:15:06.300)
those kids are probably gonna pass by the truck
Lex Fridman (1:15:09.020)
and cross the road, even though you cannot see them.
Lex Fridman (1:15:12.020)
So you have to have memory,
Elon Musk (1:15:17.340)
you have to need to remember that there were kids there
Lex Fridman (1:15:19.100)
and you need to have some forward prediction
Elon Musk (1:15:21.820)
of what their position will be at the time of relevance.
Lex Fridman (1:15:25.660)
So with occlusions and computer vision,
Elon Musk (1:15:28.540)
when you can't see an object anymore,
Lex Fridman (1:15:30.840)
even when it just walks behind a tree and reappears,
Elon Musk (1:15:33.460)
that's a really, really,
Lex Fridman (1:15:35.100)
I mean, at least in academic literature,
Elon Musk (1:15:37.100)
it's tracking through occlusions, it's very difficult.
Lex Fridman (1:15:40.620)
Yeah, we're doing it.
Elon Musk (1:15:41.940)
I understand this.
Lex Fridman (1:15:42.780)
Yeah.
Lex Fridman (1:15:43.600)
So some of it.
Lex Fridman (1:15:44.440)
It's like object permanence,
Elon Musk (1:15:45.420)
like same thing happens with humans with neural nets.
Lex Fridman (1:15:47.700)
Like when like a toddler grows up,
Elon Musk (1:15:50.100)
like there's a point in time where they develop,
Lex Fridman (1:15:54.420)
they have a sense of object permanence.
Lex Fridman (1:15:56.220)
So before a certain age, if you have a ball or a toy
Lex Fridman (1:15:59.580)
or whatever, and you put it behind your back
Lex Fridman (1:16:01.180)
and you pop it out, if they don't,
Lex Fridman (1:16:03.360)
before they have object permanence,
Elon Musk (1:16:04.660)
it's like a new thing every time.
Lex Fridman (1:16:05.860)
It's like, whoa, this toy went poof, just fared
Lex Fridman (1:16:08.260)
and now it's back again and they can't believe it.
Lex Fridman (1:16:09.980)
And that they can play peekaboo all day long
Elon Musk (1:16:12.140)
because peekaboo is fresh every time.
Lex Fridman (1:16:13.940)
But then we figured out object permanence,
Elon Musk (1:16:18.220)
then they realize, oh no, the object is not gone,
Lex Fridman (1:16:20.360)
it's just behind your back.
Elon Musk (1:16:22.660)
Sometimes I wish we never did figure out object permanence.
Lex Fridman (1:16:26.380)
Yeah, so that's a...
Lex Fridman (1:16:28.260)
So that's an important problem to solve.
Lex Fridman (1:16:31.620)
Yes, so like an important evolution
Elon Musk (1:16:33.960)
of the neural nets in the car is
Lex Fridman (1:16:39.740)
memory across both time and space.
Lex Fridman (1:16:43.860)
So now you can't remember, like you have to say,
Lex Fridman (1:16:47.100)
like how long do you want to remember things for?
Lex Fridman (1:16:48.960)
And there's a cost to remembering things for a long time.
Lex Fridman (1:16:53.260)
So you can like run out of memory
Elon Musk (1:16:55.740)
to try to remember too much for too long.
Lex Fridman (1:16:58.600)
And then you also have things that are stale
Elon Musk (1:17:01.420)
if you remember them for too long.
Lex Fridman (1:17:03.540)
And then you also need things that are remembered over time.
Lex Fridman (1:17:06.880)
So even if you like say have like,
Lex Fridman (1:17:10.640)
for our good sake, five seconds of memory on a time basis,
Lex Fridman (1:17:14.580)
but like let's say you're parked at a light
Lex Fridman (1:17:17.100)
and you saw, use a pedestrian example,
Elon Musk (1:17:20.660)
that people were waiting to cross the road
Lex Fridman (1:17:25.220)
and you can't quite see them because of an occlusion,
Lex Fridman (1:17:28.900)
but they might wait for a minute before the light changes
Lex Fridman (1:17:31.820)
for them to cross the road.
Elon Musk (1:17:33.180)
You still need to remember that that's where they were
Lex Fridman (1:17:36.940)
and that they're probably going
Elon Musk (1:17:38.420)
to cross the road type of thing.
Lex Fridman (1:17:40.580)
So even if that exceeds your time based memory,
Elon Musk (1:17:44.740)
it should not exceed your space memory.
Lex Fridman (1:17:48.140)
And I just think the data engine side of that,
Lex Fridman (1:17:50.500)
so getting the data to learn all of the concepts
Lex Fridman (1:17:53.540)
that you're saying now is an incredible process.
Elon Musk (1:17:56.180)
It's this iterative process of just,
Lex Fridman (1:17:58.380)
it's this hydranet of many.
Elon Musk (1:18:00.740)
Hydranet.
Lex Fridman (1:18:03.380)
We're changing the name to something else.
Elon Musk (1:18:05.420)
Okay, I'm sure it'll be equally as Rick and Morty like.
Lex Fridman (1:18:09.700)
There's a lot of, yeah.
Elon Musk (1:18:11.460)
We've rearchitected the neural net,
Lex Fridman (1:18:14.700)
the neural nets in the cars so many times it's crazy.
Elon Musk (1:18:17.980)
Oh, so every time there's a new major version,
Lex Fridman (1:18:20.020)
you'll rename it to something more ridiculous
Elon Musk (1:18:21.940)
or memorable and beautiful, sorry.
Lex Fridman (1:18:25.060)
Not ridiculous, of course.
Elon Musk (1:18:28.140)
If you see the full array of neural nets
Lex Fridman (1:18:32.500)
that are operating in the cars,
Elon Musk (1:18:34.180)
it kind of boggles the mind.
Lex Fridman (1:18:36.420)
There's so many layers, it's crazy.
Elon Musk (1:18:39.860)
So, yeah.
Lex Fridman (1:18:43.620)
And we started off with simple neural nets
Elon Musk (1:18:48.620)
that were basically image recognition
Lex Fridman (1:18:53.300)
on a single frame from a single camera
Lex Fridman (1:18:56.620)
and then trying to knit those together
Lex Fridman (1:19:00.700)
with C, I should say we're really primarily running C here
Elon Musk (1:19:07.680)
because C++ is too much overhead
Lex Fridman (1:19:10.000)
and we have our own C compiler.
Lex Fridman (1:19:11.660)
So to get maximum performance,
Lex Fridman (1:19:13.580)
we actually wrote our own C compiler
Lex Fridman (1:19:15.780)
and are continuing to optimize our C compiler
Lex Fridman (1:19:18.060)
for maximum efficiency.
Elon Musk (1:19:20.060)
In fact, we've just recently done a new river
Lex Fridman (1:19:23.100)
on our C compiler that'll compile directly
Elon Musk (1:19:25.260)
to our autopilot hardware.
Lex Fridman (1:19:26.980)
If you want to compile the whole thing down
Elon Musk (1:19:28.960)
with your own compiler, like so efficiency here,
Lex Fridman (1:19:32.660)
because there's all kinds of compute,
Elon Musk (1:19:33.940)
there's CPU, GPU, there's like basic types of things
Lex Fridman (1:19:37.340)
and you have to somehow figure out the scheduling
Elon Musk (1:19:39.140)
across all of those things.
Lex Fridman (1:19:40.140)
And so you're compiling the code down that does all, okay.
Lex Fridman (1:19:44.500)
So that's why there's a lot of people involved.
Lex Fridman (1:19:46.900)
There's a lot of hardcore software engineering
Elon Musk (1:19:50.620)
at a very sort of bare metal level
Lex Fridman (1:19:54.700)
because we're trying to do a lot of compute
Elon Musk (1:19:57.280)
that's constrained to our full self driving computer.
Lex Fridman (1:20:03.040)
So we want to try to have the highest frames per second
Elon Musk (1:20:07.740)
possible in a sort of very finite amount of compute
Lex Fridman (1:20:14.580)
and power.
Lex Fridman (1:20:15.420)
So we really put a lot of effort into the efficiency
Lex Fridman (1:20:20.940)
of our compute.
Lex Fridman (1:20:23.820)
And so there's actually a lot of work done
Lex Fridman (1:20:26.060)
by some very talented software engineers at Tesla
Elon Musk (1:20:29.660)
that at a very foundational level
Lex Fridman (1:20:33.140)
to improve the efficiency of compute
Lex Fridman (1:20:35.260)
and how we use the trip accelerators,
Lex Fridman (1:20:38.940)
which are basically, you know,
Elon Musk (1:20:43.180)
doing matrix math dot products,
Lex Fridman (1:20:45.420)
like a bazillion dot products.
Lex Fridman (1:20:47.340)
And it's like, what are neural nets?
Lex Fridman (1:20:49.580)
It's like compute wise, like 99% dot products.
Elon Musk (1:20:54.340)
So, you know.
Lex Fridman (1:20:57.100)
And you want to achieve as many high frame rates
Elon Musk (1:20:59.700)
like a video game.
Lex Fridman (1:21:00.580)
You want full resolution, high frame rate.
Elon Musk (1:21:05.020)
High frame rate, low latency, low jitter.
Lex Fridman (1:21:10.020)
So I think one of the things we're moving towards now
Elon Musk (1:21:18.340)
is no post processing of the image
Lex Fridman (1:21:22.540)
through the image signal processor.
Lex Fridman (1:21:26.700)
So like what happens for cameras is that,
Lex Fridman (1:21:32.580)
almost all cameras is they,
Elon Musk (1:21:35.740)
there's a lot of post processing done
Lex Fridman (1:21:37.700)
in order to make pictures look pretty.
Lex Fridman (1:21:40.260)
And so we don't care about pictures looking pretty.
Lex Fridman (1:21:43.540)
We just want the data.
Lex Fridman (1:21:45.380)
So we're moving just raw photon counts.
Lex Fridman (1:21:48.780)
So the system will, like the image that the computer sees
Elon Musk (1:21:55.060)
is actually much more than what you'd see
Lex Fridman (1:21:57.860)
if you represented it on a camera.
Elon Musk (1:21:59.100)
It's got much more data.
Lex Fridman (1:22:00.780)
And even in a very low light conditions,
Elon Musk (1:22:02.560)
you can see that there's a small photon count difference
Lex Fridman (1:22:05.220)
between this spot here and that spot there,
Elon Musk (1:22:08.780)
which means that,
Lex Fridman (1:22:09.620)
so it can see in the dark incredibly well
Elon Musk (1:22:12.900)
because it can detect these tiny differences
Lex Fridman (1:22:15.100)
in photon counts.
Elon Musk (1:22:16.980)
Like much better than you could possibly imagine.
Lex Fridman (1:22:20.780)
So, and then we also save 13 milliseconds on a latency.
Elon Musk (1:22:27.420)
So.
Lex Fridman (1:22:29.260)
From removing the post processing on the image?
Elon Musk (1:22:31.220)
Yes.
Lex Fridman (1:22:32.060)
It's like,
Elon Musk (1:22:32.900)
it's incredible.
Lex Fridman (1:22:34.300)
Cause we've got eight cameras
Lex Fridman (1:22:35.820)
and then there's roughly, I don't know,
Lex Fridman (1:22:39.780)
one and a half milliseconds or so,
Elon Musk (1:22:41.980)
maybe 1.6 milliseconds of latency for each camera.
Lex Fridman (1:22:46.220)
And so like going to just,
Elon Musk (1:22:53.380)
basically bypassing the image processor
Lex Fridman (1:22:56.300)
gets us back 13 milliseconds of latency,
Elon Musk (1:22:58.240)
which is important.
Lex Fridman (1:22:59.320)
And we track latency all the way from, you know,
Elon Musk (1:23:03.440)
photon hits the camera to, you know,
Lex Fridman (1:23:06.760)
all the steps that it's got to go through to get,
Elon Musk (1:23:08.920)
you know, go through the various neural nets
Lex Fridman (1:23:12.280)
and the C code.
Lex Fridman (1:23:13.320)
And there's a little bit of C++ there as well.
Lex Fridman (1:23:17.800)
Well, I can maybe a lot, but it,
Elon Musk (1:23:20.200)
the core stuff is heavy duty computers all in C.
Lex Fridman (1:23:25.120)
And so we track that latency all the way
Elon Musk (1:23:28.840)
to an output command to the drive unit to accelerate
Lex Fridman (1:23:33.480)
the brakes just to slow down the steering,
Elon Musk (1:23:36.600)
you know, turn left or right.
Lex Fridman (1:23:38.920)
So, cause you go to output a command
Elon Musk (1:23:40.680)
that's got to go to a controller.
Lex Fridman (1:23:41.800)
And like some of these controllers have an update frequency
Elon Musk (1:23:44.120)
that's maybe 10 Hertz or something like that,
Lex Fridman (1:23:46.480)
which is slow.
Elon Musk (1:23:47.320)
That's like now you lose a hundred milliseconds potentially.
Lex Fridman (1:23:50.240)
So, so then we want to update the,
Elon Musk (1:23:55.040)
the drivers on the like say steering and braking control
Lex Fridman (1:23:58.720)
to have more like a hundred Hertz instead of 10 Hertz.
Lex Fridman (1:24:02.920)
And then you've got a 10 millisecond latency
Lex Fridman (1:24:04.400)
instead of a hundred millisecond worst case latency.
Lex Fridman (1:24:06.400)
And actually jitter is more of a challenge than latency.
Lex Fridman (1:24:09.560)
Cause latency is like, you can, you can,
Elon Musk (1:24:11.000)
you can anticipate and predict, but if you're,
Lex Fridman (1:24:13.120)
but if you've got a stack up of things going
Elon Musk (1:24:14.920)
from the camera to the, to the computer through
Lex Fridman (1:24:17.880)
then a series of other computers,
Lex Fridman (1:24:19.400)
and finally to an actuator on the car,
Lex Fridman (1:24:22.120)
if you have a stack up of tolerances of timing tolerances,
Elon Musk (1:24:26.960)
then you can have quite a variable latency,
Lex Fridman (1:24:29.000)
which is called jitter.
Elon Musk (1:24:30.360)
And, and that makes it hard to, to, to anticipate exactly
Lex Fridman (1:24:34.840)
what, how you should turn the car or accelerate,
Elon Musk (1:24:37.480)
because if you've got maybe a hundred,
Lex Fridman (1:24:40.640)
50, 200 milliseconds of jitter,
Elon Musk (1:24:42.440)
then you could be off by, you know, up to 0.2 seconds.
Lex Fridman (1:24:45.560)
And this can make, this could make a big difference.
Lex Fridman (1:24:47.560)
So you have to interpolate somehow to, to, to,
Lex Fridman (1:24:50.160)
to deal with the effects of jitter.
Lex Fridman (1:24:52.400)
So that you can make like robust control decisions.
Lex Fridman (1:24:57.920)
You have to, so the jitters and the sensor information,
Elon Musk (1:25:01.620)
or the jitter can occur at any stage in the pipeline.
Lex Fridman (1:25:05.000)
You can, if you have just, if you have fixed latency,
Elon Musk (1:25:07.760)
you can anticipate and, and like say, okay,
Lex Fridman (1:25:11.960)
we know that our information is for argument's sake,
Elon Musk (1:25:16.560)
150 milliseconds stale.
Lex Fridman (1:25:19.000)
Like, so for, for, for 150 milliseconds
Elon Musk (1:25:22.840)
from photon second camera to where you can measure a change
Lex Fridman (1:25:28.500)
in the acceleration of the vehicle.
Lex Fridman (1:25:33.720)
So then, then you can say, okay, well, we're going to enter,
Lex Fridman (1:25:38.080)
we know it's 150 milliseconds.
Lex Fridman (1:25:39.400)
So we're going to take that into account
Lex Fridman (1:25:40.960)
and, and compensate for that latency.
Elon Musk (1:25:44.240)
However, if you've got then 150 milliseconds of latency
Lex Fridman (1:25:47.320)
plus a hundred milliseconds of jitter,
Elon Musk (1:25:49.200)
that's which could be anywhere from zero,
Lex Fridman (1:25:50.760)
zero to a hundred milliseconds on top.
Elon Musk (1:25:52.240)
So, so then your latency could be from 150 to 250 milliseconds.
Lex Fridman (1:25:55.720)
Now you've got a hundred milliseconds
Elon Musk (1:25:56.680)
that you don't know what to do with.
Lex Fridman (1:25:58.040)
And that's basically random.
Lex Fridman (1:26:01.440)
So getting rid of jitter is extremely important.
Lex Fridman (1:26:04.260)
And that affects your control decisions
Lex Fridman (1:26:05.840)
and all those kinds of things.
Lex Fridman (1:26:07.400)
Okay.
Elon Musk (1:26:09.160)
Yeah, the car's just going to fundamentally maneuver better
Lex Fridman (1:26:11.140)
with lower jitter.
Elon Musk (1:26:12.920)
Got it.
Lex Fridman (1:26:13.760)
The cars will maneuver with superhuman ability
Lex Fridman (1:26:16.760)
and reaction time much faster than a human.
Lex Fridman (1:26:20.360)
I mean, I think over time the autopilot,
Elon Musk (1:26:24.320)
full self driving will be capable of maneuvers
Lex Fridman (1:26:26.400)
that are far more than what like James Bond could do
Elon Musk (1:26:34.840)
in like the best movie type of thing.
Lex Fridman (1:26:36.360)
That's exactly what I was imagining in my mind,
Elon Musk (1:26:38.620)
as you said it.
Lex Fridman (1:26:40.240)
It's like an impossible maneuvers
Elon Musk (1:26:41.820)
that a human couldn't do.
Lex Fridman (1:26:43.020)
Yeah, so.
Elon Musk (1:26:45.080)
Well, let me ask sort of looking back the six years,
Lex Fridman (1:26:48.720)
looking out into the future,
Elon Musk (1:26:50.280)
based on your current understanding,
Lex Fridman (1:26:51.840)
how hard do you think this,
Elon Musk (1:26:53.640)
this full self driving problem,
Lex Fridman (1:26:55.360)
when do you think Tesla will solve level four FSD?
Elon Musk (1:27:01.040)
I mean, it's looking quite likely
Lex Fridman (1:27:02.340)
that it will be next year.
Lex Fridman (1:27:05.520)
And what does the solution look like?
Lex Fridman (1:27:07.040)
Is it the current pool of FSD beta candidates,
Elon Musk (1:27:10.200)
they start getting greater and greater
Lex Fridman (1:27:13.120)
as they have been degrees of autonomy,
Lex Fridman (1:27:15.760)
and then there's a certain level
Lex Fridman (1:27:17.760)
beyond which they can do their own,
Elon Musk (1:27:20.800)
they can read a book.
Lex Fridman (1:27:22.880)
Yeah, so.
Elon Musk (1:27:25.720)
I mean, you can see that anybody
Lex Fridman (1:27:26.880)
who's been following the full self driving beta closely
Elon Musk (1:27:30.640)
will see that the rate of disengagements
Lex Fridman (1:27:35.260)
has been dropping rapidly.
Lex Fridman (1:27:37.440)
So like disengagement be where the driver intervenes
Lex Fridman (1:27:40.720)
to prevent the car from doing something dangerous,
Elon Musk (1:27:44.000)
potentially, so.
Lex Fridman (1:27:49.740)
So the interventions per million miles
Elon Musk (1:27:53.240)
has been dropping dramatically at some point.
Lex Fridman (1:27:57.520)
And that trend looks like it happens next year
Elon Musk (1:28:01.040)
is that the probability of an accident on FSD
Lex Fridman (1:28:06.040)
is less than that of the average human,
Lex Fridman (1:28:09.520)
and then significantly less than that of the average human.
Lex Fridman (1:28:13.480)
So it certainly appears like we will get there next year.
Elon Musk (1:28:21.080)
Then of course, then there's gonna be a case of,
Lex Fridman (1:28:24.000)
okay, well, we now have to prove this to regulators
Lex Fridman (1:28:26.040)
and prove it to, you know, and we want a standard
Lex Fridman (1:28:28.640)
that is not just equivalent to a human,
Lex Fridman (1:28:31.160)
but much better than the average human.
Lex Fridman (1:28:33.480)
I think it's gotta be at least two or three times
Elon Musk (1:28:35.560)
two or three times higher safety than a human.
Lex Fridman (1:28:39.000)
So two or three times lower probability of injury
Elon Musk (1:28:41.360)
than a human before we would actually say like,
Lex Fridman (1:28:44.840)
okay, it's okay to go, it's not gonna be equivalent,
Elon Musk (1:28:46.800)
it's gonna be much better.
Lex Fridman (1:28:48.340)
So if you look, FSD 10.6 just came out recently,
Elon Musk (1:28:53.400)
10.7 is on the way, maybe 11 is on the way,
Lex Fridman (1:28:57.120)
so we're in the future.
Elon Musk (1:28:58.440)
Yeah, we were hoping to get 11 out this year,
Lex Fridman (1:29:01.040)
but 11 actually has a whole bunch of fundamental rewrites
Elon Musk (1:29:07.180)
on the neural net architecture,
Lex Fridman (1:29:10.840)
and some fundamental improvements
Elon Musk (1:29:14.040)
in creating vector space, so.
Lex Fridman (1:29:19.680)
There is some fundamental like leap
Elon Musk (1:29:22.240)
that really deserves the 11,
Lex Fridman (1:29:24.000)
I mean, that's a pretty cool number.
Elon Musk (1:29:25.160)
Yeah, 11 would be a single stack
Lex Fridman (1:29:29.920)
for all, you know, one stack to rule them all.
Lex Fridman (1:29:36.160)
But there are just some really fundamental
Lex Fridman (1:29:40.520)
neural net architecture changes
Elon Musk (1:29:43.760)
that will allow for much more capability,
Lex Fridman (1:29:47.720)
but at first they're gonna have issues.
Lex Fridman (1:29:51.080)
So like we have this working on like sort of alpha software
Lex Fridman (1:29:54.680)
and it's good, but it's basically taking a whole bunch
Elon Musk (1:30:00.880)
of C, C++ code and leading a massive amount of C++ code
Lex Fridman (1:30:05.360)
and replacing it with a neural net.
Lex Fridman (1:30:06.440)
And Andre makes this point a lot,
Lex Fridman (1:30:09.160)
which is like neural nets are kind of eating software.
Elon Musk (1:30:12.480)
Over time there's like less and less conventional software,
Lex Fridman (1:30:15.880)
more and more neural net, which is still software,
Lex Fridman (1:30:18.240)
but it's, you know, still comes out the lines of software,
Lex Fridman (1:30:21.120)
but it's more neural net stuff
Lex Fridman (1:30:25.320)
and less, you know, heuristics basically.
Lex Fridman (1:30:33.080)
More matrix based stuff and less heuristics based stuff.
Elon Musk (1:30:38.080)
And, you know, like one of the big changes will be,
Lex Fridman (1:30:47.080)
like right now the neural nets will deliver
Elon Musk (1:30:54.080)
a giant bag of points to the C++ or C and C++ code.
Lex Fridman (1:31:00.080)
We call it the giant bag of points.
Lex Fridman (1:31:03.080)
And it's like, so you've got a pixel and something associated
Lex Fridman (1:31:08.080)
with that pixel.
Elon Musk (1:31:09.080)
Like this pixel is probably car.
Lex Fridman (1:31:11.080)
The pixel is probably lane line.
Elon Musk (1:31:13.080)
Then you've got to assemble this giant bag of points
Lex Fridman (1:31:16.080)
in the C code and turn it into vectors.
Lex Fridman (1:31:21.080)
And it does a pretty good job of it, but it's,
Lex Fridman (1:31:26.080)
we want to just, you know,
Elon Musk (1:31:30.080)
we need another layer of neural nets on top of that
Lex Fridman (1:31:35.080)
to take the giant bag of points and distill that down
Elon Musk (1:31:40.080)
to a vector space in the neural net part of the software,
Lex Fridman (1:31:45.080)
as opposed to the heuristics part of the software.
Elon Musk (1:31:48.080)
This is a big improvement.
Lex Fridman (1:31:51.080)
Neural nets all the way down.
Elon Musk (1:31:52.080)
That's what you want.
Lex Fridman (1:31:53.080)
It's not even all neural nets, but this will be just a,
Elon Musk (1:31:58.080)
this is a game changer to not have the bag of points,
Lex Fridman (1:32:01.080)
the giant bag of points that has to be assembled
Elon Musk (1:32:04.080)
with many lines of C++ and have the,
Lex Fridman (1:32:09.080)
and have a neural net just assemble those into a vector.
Lex Fridman (1:32:12.080)
So the neural net is outputting much, much less data.
Lex Fridman (1:32:19.080)
It's outputting, this is a lane line.
Elon Musk (1:32:22.080)
This is a curb.
Lex Fridman (1:32:23.080)
This is drivable space.
Elon Musk (1:32:24.080)
This is a car.
Lex Fridman (1:32:25.080)
This is a pedestrian or a cyclist or something like that.
Elon Musk (1:32:29.080)
It's outputting, it's really outputting proper vectors
Lex Fridman (1:32:35.080)
to the C, C++ control code,
Elon Musk (1:32:39.080)
as opposed to the sort of constructing the vectors in C,
Lex Fridman (1:32:50.080)
which we've done, I think, quite a good job of,
Lex Fridman (1:32:52.080)
but we're kind of hitting a local maximum
Lex Fridman (1:32:55.080)
on how well the C can do this.
Lex Fridman (1:32:59.080)
So this is really a big deal.
Lex Fridman (1:33:02.080)
And just all of the networks in the car
Elon Musk (1:33:04.080)
need to move to surround video.
Lex Fridman (1:33:06.080)
There's still some legacy networks that are not surround video.
Lex Fridman (1:33:11.080)
And all of the training needs to move to surround video
Lex Fridman (1:33:14.080)
and the efficiency of the training,
Elon Musk (1:33:16.080)
it needs to get better and it is.
Lex Fridman (1:33:18.080)
And then we need to move everything to raw photon counts
Elon Musk (1:33:25.080)
as opposed to processed images,
Lex Fridman (1:33:29.080)
which is quite a big reset on the training
Elon Musk (1:33:31.080)
because the system's trained on post processed images.
Lex Fridman (1:33:35.080)
So we need to redo all the training
Elon Musk (1:33:38.080)
to train against the raw photon counts
Lex Fridman (1:33:41.080)
instead of the post processed image.
Lex Fridman (1:33:43.080)
So ultimately, it's kind of reducing the complexity
Lex Fridman (1:33:46.080)
of the whole thing.
Lex Fridman (1:33:47.080)
So reducing the...
Lex Fridman (1:33:50.080)
Lines of code will actually go lower.
Elon Musk (1:33:52.080)
Yeah, that's fascinating.
Lex Fridman (1:33:54.080)
So you're doing fusion of all the sensors
Lex Fridman (1:33:56.080)
and reducing the complexity of having to deal with these...
Lex Fridman (1:33:58.080)
Fusion of the cameras.
Elon Musk (1:33:59.080)
Fusion of the cameras, really.
Lex Fridman (1:34:00.080)
Right, yes.
Elon Musk (1:34:03.080)
Same with humans.
Lex Fridman (1:34:05.080)
Well, I guess we've got ears too.
Elon Musk (1:34:07.080)
Yeah, we'll actually need to incorporate sound as well
Lex Fridman (1:34:11.080)
because you need to listen for ambulance sirens
Elon Musk (1:34:14.080)
or fire trucks or somebody yelling at you or something.
Lex Fridman (1:34:20.080)
I don't know.
Elon Musk (1:34:21.080)
There's a little bit of audio that needs to be incorporated as well.
Lex Fridman (1:34:24.080)
Do you need to go back for a break?
Elon Musk (1:34:26.080)
Yeah, sure, let's take a break.
Lex Fridman (1:34:27.080)
Okay.
Elon Musk (1:34:28.080)
Honestly, frankly, the ideas are the easy thing
Lex Fridman (1:34:33.080)
and the implementation is the hard thing.
Elon Musk (1:34:35.080)
The idea of going to the moon is the easy part.
Lex Fridman (1:34:37.080)
Not going to the moon is the hard part.
Elon Musk (1:34:39.080)
It's the hard part.
Lex Fridman (1:34:40.080)
And there's a lot of hardcore engineering
Elon Musk (1:34:42.080)
that's got to get done at the hardware and software level.
Lex Fridman (1:34:46.080)
Like I said, optimizing the C compiler
Lex Fridman (1:34:48.080)
and just cutting out latency everywhere.
Lex Fridman (1:34:55.080)
If we don't do this, the system will not work properly.
Lex Fridman (1:34:59.080)
So the work of the engineers doing this,
Lex Fridman (1:35:02.080)
they are like the unsung heroes,
Lex Fridman (1:35:05.080)
but they are critical to the success of the situation.
Lex Fridman (1:35:08.080)
I think you made it clear.
Elon Musk (1:35:09.080)
I mean, at least to me, it's super exciting.
Lex Fridman (1:35:11.080)
Everything that's going on outside of what Andre is doing.
Elon Musk (1:35:15.080)
Just the whole infrastructure, the software.
Lex Fridman (1:35:17.080)
I mean, everything is going on with Data Engine,
Elon Musk (1:35:19.080)
whatever it's called.
Lex Fridman (1:35:21.080)
The whole process is just work of art to me.
Elon Musk (1:35:24.080)
The sheer scale of it boggles my mind.
Lex Fridman (1:35:26.080)
Like the training, the amount of work done with,
Elon Musk (1:35:29.080)
like we've written all this custom software for training and labeling
Lex Fridman (1:35:33.080)
and to do auto labeling.
Elon Musk (1:35:34.080)
Auto labeling is essential.
Lex Fridman (1:35:38.080)
Because especially when you've got surround video, it's very difficult.
Elon Musk (1:35:42.080)
To label surround video from scratch is extremely difficult.
Lex Fridman (1:35:48.080)
Like take a human such a long time to even label one video clip,
Elon Musk (1:35:52.080)
like several hours.
Lex Fridman (1:35:54.080)
Or the auto label, basically we just apply heavy duty,
Elon Musk (1:36:00.080)
like a lot of compute to the video clips to preassign
Lex Fridman (1:36:06.080)
and guess what all the things are that are going on in the surround video.
Lex Fridman (1:36:09.080)
And then there's like correcting it.
Lex Fridman (1:36:10.080)
Yeah.
Lex Fridman (1:36:11.080)
And then all the human has to do is like tweet,
Lex Fridman (1:36:13.080)
like say, adjust what is incorrect.
Elon Musk (1:36:16.080)
This is like increases productivity by in fact a hundred or more.
Lex Fridman (1:36:21.080)
Yeah.
Lex Fridman (1:36:22.080)
So you've presented Tesla Bot as primarily useful in the factory.
Lex Fridman (1:36:25.080)
First of all, I think humanoid robots are incredible.
Elon Musk (1:36:28.080)
From a fan of robotics, I think the elegance of movement
Lex Fridman (1:36:32.080)
that humanoid robots, that bipedal robots show are just so cool.
Lex Fridman (1:36:38.080)
So it's really interesting that you're working on this
Lex Fridman (1:36:40.080)
and also talking about applying the same kind of all the ideas,
Elon Musk (1:36:44.080)
some of which we've talked about with Data Engine,
Lex Fridman (1:36:46.080)
all the things that we're talking about with Tesla Autopilot,
Elon Musk (1:36:49.080)
just transferring that over to just yet another robotics problem.
Lex Fridman (1:36:54.080)
I have to ask, since I care about human robot interaction,
Lex Fridman (1:36:57.080)
so the human side of that.
Lex Fridman (1:36:59.080)
So you've talked about mostly in the factory.
Lex Fridman (1:37:01.080)
Do you see part of this problem that Tesla Bot has to solve
Lex Fridman (1:37:06.080)
is interacting with humans and potentially having a place like in the home.
Lex Fridman (1:37:10.080)
So interacting, not just not replacing labor, but also like, I don't know,
Lex Fridman (1:37:15.080)
being a friend or an assistant or something like that.
Elon Musk (1:37:18.080)
Yeah, I think the possibilities are endless.
Lex Fridman (1:37:27.080)
It's not quite in Tesla's primary mission direction
Elon Musk (1:37:32.080)
of accelerating sustainable energy,
Lex Fridman (1:37:34.080)
but it is an extremely useful thing that we can do for the world,
Elon Musk (1:37:38.080)
which is to make a useful humanoid robot that is capable of interacting with the world
Lex Fridman (1:37:44.080)
and helping in many different ways.
Elon Musk (1:37:49.080)
I think if you say extrapolate to many years in the future,
Lex Fridman (1:37:59.080)
I think work will become optional.
Elon Musk (1:38:04.080)
There's a lot of jobs that if people weren't paid to do it,
Lex Fridman (1:38:10.080)
they wouldn't do it.
Elon Musk (1:38:12.080)
It's not fun necessarily.
Lex Fridman (1:38:14.080)
If you're washing dishes all day,
Elon Musk (1:38:16.080)
it's like, you know, even if you really like washing dishes,
Lex Fridman (1:38:19.080)
you really want to do it for eight hours a day every day.
Elon Musk (1:38:22.080)
Probably not.
Lex Fridman (1:38:25.080)
And then there's like dangerous work.
Lex Fridman (1:38:27.080)
And basically, if it's dangerous, boring,
Lex Fridman (1:38:30.080)
it has like potential for repetitive stress injury, that kind of thing.
Elon Musk (1:38:34.080)
Then that's really where humanoid robots would add the most value initially.
Lex Fridman (1:38:40.080)
So that's what we're aiming for is for the humanoid robots to do jobs
Elon Musk (1:38:47.080)
that people don't voluntarily want to do.
Lex Fridman (1:38:51.080)
And then we'll have to pair that obviously
Elon Musk (1:38:53.080)
with some kind of universal basic income in the future.
Lex Fridman (1:38:56.080)
So I think.
Lex Fridman (1:39:00.080)
So do you see a world when there's like hundreds of millions of Tesla bots
Lex Fridman (1:39:05.080)
doing different, performing different tasks throughout the world?
Elon Musk (1:39:12.080)
Yeah, I haven't really thought about it that far into the future,
Lex Fridman (1:39:14.080)
but I guess there may be something like that.
Elon Musk (1:39:17.080)
So.
Lex Fridman (1:39:20.080)
Can I ask a wild question?
Lex Fridman (1:39:22.080)
So the number of Tesla cars has been accelerating.
Lex Fridman (1:39:25.080)
There's been close to two million produced.
Elon Musk (1:39:28.080)
Many of them have autopilot.
Lex Fridman (1:39:30.080)
I think we're over two million now.
Lex Fridman (1:39:32.080)
Do you think there will ever be a time when there will be more Tesla bots
Lex Fridman (1:39:36.080)
than Tesla cars?
Elon Musk (1:39:40.080)
Yeah.
Lex Fridman (1:39:42.080)
Actually, it's funny you asked this question,
Elon Musk (1:39:44.080)
because normally I do try to think pretty far into the future,
Lex Fridman (1:39:47.080)
but I haven't really thought that far into the future with the Tesla bot,
Elon Musk (1:39:52.080)
or it's codenamed Optimus.
Lex Fridman (1:39:55.080)
I call it Optimus subprime.
Elon Musk (1:39:59.080)
It's not like a giant transformer robot.
Lex Fridman (1:40:04.080)
So.
Lex Fridman (1:40:07.080)
But it's meant to be a general purpose, helpful bot.
Lex Fridman (1:40:14.080)
And basically, like the things that we're basically like, like,
Elon Musk (1:40:18.080)
Tesla, I think is the has the most advanced real world AI
Lex Fridman (1:40:25.080)
for interacting with the real world,
Elon Musk (1:40:26.080)
which are developed as a function of to make self driving work.
Lex Fridman (1:40:30.080)
And so along with custom hardware and like a lot of, you know,
Elon Musk (1:40:36.080)
hardcore low level software to have it run efficiently
Lex Fridman (1:40:39.080)
and be power efficient because it's one thing to do neural nets
Elon Musk (1:40:43.080)
if you've got a gigantic server room with 10,000 computers.
Lex Fridman (1:40:45.080)
But now let's say you just you have to now distill that down
Elon Musk (1:40:48.080)
into one computer that's running at low power in a humanoid robot or a car.
Lex Fridman (1:40:53.080)
That's actually very difficult.
Elon Musk (1:40:54.080)
A lot of hardcore software work is required for that.
Lex Fridman (1:40:59.080)
So since we're kind of like solving the navigate the real world
Elon Musk (1:41:05.080)
with neural nets problem for cars,
Lex Fridman (1:41:08.080)
which are kind of robots with four wheels,
Elon Musk (1:41:10.080)
then it's like kind of a natural extension of that is to put it
Lex Fridman (1:41:14.080)
in a robot with arms and legs and actuators.
Lex Fridman (1:41:20.080)
So like the two hard things are like you basically need to make the
Lex Fridman (1:41:31.080)
have the robot be intelligent enough to interact in a sensible way
Elon Musk (1:41:34.080)
with the environment.
Lex Fridman (1:41:36.080)
So you need real world AI and you need to be very good at manufacturing,
Elon Musk (1:41:43.080)
which is a very hard problem.
Lex Fridman (1:41:44.080)
Tesla is very good at manufacturing and also has the real world AI.
Lex Fridman (1:41:50.080)
So making the humanoid robot work is basically means developing custom motors
Lex Fridman (1:41:58.080)
and sensors that are different from what a car would use.
Lex Fridman (1:42:04.080)
But we've also we have I think we have the best expertise
Lex Fridman (1:42:10.080)
in developing advanced electric motors and power electronics.
Lex Fridman (1:42:15.080)
So it just has to be for humanoid robot application on a car.
Lex Fridman (1:42:22.080)
Still, you do talk about love sometimes.
Lex Fridman (1:42:25.080)
So let me ask.
Lex Fridman (1:42:26.080)
This isn't like for like sex robots or something like that.
Elon Musk (1:42:29.080)
Love is the answer.
Lex Fridman (1:42:30.080)
Yes.
Elon Musk (1:42:33.080)
There is something compelling to us, not compelling,
Lex Fridman (1:42:36.080)
but we connect with humanoid robots or even like robots like with the dog
Lex Fridman (1:42:41.080)
and shapes of dogs.
Elon Musk (1:42:43.080)
It just it seems like, you know, there's a huge amount of loneliness in this world.
Elon Musk (1:42:48.080)
All of us seek companionship with other humans, friendship
Lex Fridman (1:42:51.080)
and all those kinds of things.
Elon Musk (1:42:52.080)
We have a lot of here in Austin, a lot of people have dogs.
Lex Fridman (1:42:56.080)
There seems to be a huge opportunity to also have robots that decrease
Elon Musk (1:43:01.080)
the amount of loneliness in the world or help us humans connect with each other.
Lex Fridman (1:43:09.080)
So in a way that dogs can.
Lex Fridman (1:43:12.080)
Do you think about that with TeslaBot at all?
Lex Fridman (1:43:14.080)
Or is it really focused on the problem of performing specific tasks,
Lex Fridman (1:43:19.080)
not connecting with humans?
Elon Musk (1:43:23.080)
I mean, to be honest, I have not actually thought about it from the companionship
Elon Musk (1:43:27.080)
standpoint, but I think it actually would end up being it could be actually
Lex Fridman (1:43:31.080)
a very good companion.
Lex Fridman (1:43:34.080)
And it could develop like a personality over time that is that is like unique,
Lex Fridman (1:43:44.080)
like, you know, it's not like they're just all the robots are the same
Lex Fridman (1:43:47.080)
and that personality could evolve to be, you know,
Lex Fridman (1:43:53.080)
match the owner or the, you know, yes, the owner.
Elon Musk (1:43:59.080)
Well, whatever you want to call it.
Lex Fridman (1:44:02.080)
The other half, right?
Elon Musk (1:44:05.080)
In the same way that friends do.
Lex Fridman (1:44:06.080)
See, I think that's a huge opportunity.
Elon Musk (1:44:09.080)
Yeah, no, that's interesting.
Lex Fridman (1:44:14.080)
Because, you know, like there's a Japanese phrase I like, Wabi Sabi,
Elon Musk (1:44:19.080)
you know, the subtle imperfections are what makes something special.
Lex Fridman (1:44:23.080)
And the subtle imperfections of the personality of the robot mapped
Elon Musk (1:44:28.080)
to the subtle imperfections of the robot's human friend.
Lex Fridman (1:44:34.080)
I don't know, owner sounds like maybe the wrong word,
Lex Fridman (1:44:36.080)
but could actually make an incredible buddy, basically.
Lex Fridman (1:44:42.080)
In that way, the imperfections.
Elon Musk (1:44:43.080)
Like R2D2 or like a C3PO sort of thing, you know.
Lex Fridman (1:44:46.080)
So from a machine learning perspective,
Elon Musk (1:44:49.080)
I think the flaws being a feature is really nice.
Lex Fridman (1:44:53.080)
You could be quite terrible at being a robot for quite a while
Elon Musk (1:44:57.080)
in the general home environment or in general world.
Lex Fridman (1:45:00.080)
And that's kind of adorable.
Lex Fridman (1:45:02.080)
And that's like, those are your flaws and you fall in love with those flaws.
Lex Fridman (1:45:06.080)
So it's very different than autonomous driving
Elon Musk (1:45:09.080)
where it's a very high stakes environment you cannot mess up.
Lex Fridman (1:45:13.080)
And so it's more fun to be a robot in the home.
Elon Musk (1:45:17.080)
Yeah, in fact, if you think of like C3PO and R2D2,
Lex Fridman (1:45:21.080)
like they actually had a lot of like flaws and imperfections
Lex Fridman (1:45:24.080)
and silly things and they would argue with each other.
Lex Fridman (1:45:29.080)
Were they actually good at doing anything?
Elon Musk (1:45:32.080)
I'm not exactly sure.
Lex Fridman (1:45:34.080)
They definitely added a lot to the story.
Lex Fridman (1:45:38.080)
But there's sort of quirky elements and, you know,
Lex Fridman (1:45:43.080)
that they would like make mistakes and do things.
Elon Musk (1:45:45.080)
It was like it made them relatable, I don't know, enduring.
Lex Fridman (1:45:52.080)
So yeah, I think that that could be something that probably would happen.
Lex Fridman (1:45:59.080)
But our initial focus is just to make it useful.
Lex Fridman (1:46:03.080)
So I'm confident we'll get it done.
Elon Musk (1:46:06.080)
I'm not sure what the exact timeframe is,
Lex Fridman (1:46:08.080)
but like we'll probably have, I don't know,
Elon Musk (1:46:11.080)
a decent prototype towards the end of next year or something like that.
Lex Fridman (1:46:15.080)
And it's cool that it's connected to Tesla, the car.
Elon Musk (1:46:19.080)
Yeah, it's using a lot of, you know,
Lex Fridman (1:46:22.080)
it would use the autopilot inference computer
Lex Fridman (1:46:25.080)
and a lot of the training that we've done for cars
Lex Fridman (1:46:29.080)
in terms of recognizing real world things
Elon Musk (1:46:32.080)
could be applied directly to the robot.
Lex Fridman (1:46:38.080)
But there's a lot of custom actuators and sensors that need to be developed.
Lex Fridman (1:46:42.080)
And an extra module on top of the vector space for love.
Lex Fridman (1:46:47.080)
Yeah.
Elon Musk (1:46:48.080)
That's what I'm saying.
Lex Fridman (1:46:51.080)
We can add that to the car too.
Elon Musk (1:46:53.080)
That's true.
Lex Fridman (1:46:55.080)
Yeah, it could be useful in all environments.
Elon Musk (1:46:57.080)
Like you said, a lot of people argue in the car,
Lex Fridman (1:46:59.080)
so maybe we can help them out.
Elon Musk (1:47:02.080)
You're a student of history,
Lex Fridman (1:47:03.080)
fan of Dan Carlin's Hardcore History podcast.
Elon Musk (1:47:06.080)
Yeah, it's great.
Lex Fridman (1:47:07.080)
Greatest podcast ever?
Elon Musk (1:47:08.080)
Yeah, I think it is actually.
Lex Fridman (1:47:11.080)
It almost doesn't really count as a podcast.
Elon Musk (1:47:14.080)
It's more like an audio book.
Lex Fridman (1:47:16.080)
So you were on the podcast with Dan.
Elon Musk (1:47:18.080)
I just had a chat with him about it.
Lex Fridman (1:47:21.080)
He said you guys went military and all that kind of stuff.
Elon Musk (1:47:23.080)
Yeah, it was basically, it should be titled Engineer Wars, essentially.
Lex Fridman (1:47:32.080)
Like when there's a rapid change in the rate of technology,
Elon Musk (1:47:36.080)
then engineering plays a pivotal role in victory and battle.
Lex Fridman (1:47:43.080)
How far back in history did you go?
Lex Fridman (1:47:45.080)
Did you go World War II?
Elon Musk (1:47:47.080)
Well, it was supposed to be a deep dive on fighters and bomber technology in World War II,
Lex Fridman (1:47:55.080)
but that ended up being more wide ranging than that,
Elon Musk (1:47:58.080)
because I just went down the total rathole of studying all of the fighters and bombers of World War II
Lex Fridman (1:48:04.080)
and the constant rock, paper, scissors game that one country would make this plane,
Elon Musk (1:48:10.080)
then it would make a plane to beat that, and that country would make a plane to beat that.
Lex Fridman (1:48:15.080)
And really what matters is the pace of innovation
Lex Fridman (1:48:18.080)
and also access to high quality fuel and raw materials.
Elon Musk (1:48:25.080)
Germany had some amazing designs, but they couldn't make them
Lex Fridman (1:48:29.080)
because they couldn't get the raw materials,
Lex Fridman (1:48:31.080)
and they had a real problem with the oil and fuel, basically.
Lex Fridman (1:48:37.080)
The fuel quality was extremely variable.
Lex Fridman (1:48:40.080)
So the design wasn't the bottleneck?
Lex Fridman (1:48:42.080)
Yeah, the U.S. had kickass fuel that was very consistent.
Elon Musk (1:48:47.080)
The problem is if you make a very high performance aircraft engine,
Lex Fridman (1:48:50.080)
in order to make it high performance, the fuel, the aviation gas,
Elon Musk (1:48:59.080)
has to be a consistent mixture and it has to have a high octane.
Elon Musk (1:49:07.080)
High octane is the most important thing, but it also can't have impurities and stuff
Elon Musk (1:49:11.080)
because you'll foul up the engine.
Lex Fridman (1:49:14.080)
And Germany just never had good access to oil.
Elon Musk (1:49:16.080)
They tried to get it by invading the Caucasus, but that didn't work too well.
Lex Fridman (1:49:22.080)
It never worked so well.
Elon Musk (1:49:23.080)
It didn't work out for them.
Lex Fridman (1:49:24.080)
See you, Jerry.
Elon Musk (1:49:26.080)
Nice to meet you.
Lex Fridman (1:49:28.080)
Germany was always struggling with basically shitty oil,
Lex Fridman (1:49:31.080)
and they couldn't count on high quality fuel for their aircraft,
Lex Fridman (1:49:37.080)
so they had to have all these additives and stuff.
Elon Musk (1:49:43.080)
Whereas the U.S. had awesome fuel, and they provided that to Britain as well.
Lex Fridman (1:49:48.080)
So that allowed the British and the Americans to design aircraft engines
Elon Musk (1:49:53.080)
that were super high performance, better than anything else in the world.
Lex Fridman (1:49:58.080)
Germany could design the engines, they just didn't have the fuel.
Lex Fridman (1:50:01.080)
And then also the quality of the aluminum alloys that they were getting
Lex Fridman (1:50:06.080)
was also not that great.
Lex Fridman (1:50:09.080)
Is this like, you talked about all this with Dan?
Lex Fridman (1:50:11.080)
Yep.
Elon Musk (1:50:12.080)
Awesome.
Lex Fridman (1:50:13.080)
Broadly looking at history, when you look at Genghis Khan,
Elon Musk (1:50:16.080)
when you look at Stalin, Hitler, the darkest moments of human history,
Lex Fridman (1:50:22.080)
what do you take away from those moments?
Elon Musk (1:50:24.080)
Does it help you gain insight about human nature, about human behavior today,
Lex Fridman (1:50:28.080)
whether it's the wars or the individuals or just the behavior of people,
Lex Fridman (1:50:32.080)
any aspects of history?
Lex Fridman (1:50:41.080)
Yeah, I find history fascinating.
Elon Musk (1:50:49.080)
There's just a lot of incredible things that have been done, good and bad,
Lex Fridman (1:50:54.080)
that they help you understand the nature of civilization and individuals.
Lex Fridman (1:51:06.080)
Does it make you sad that humans do these kinds of things to each other?
Lex Fridman (1:51:09.080)
You look at the 20th century, World War II, the cruelty, the abuse of power,
Elon Musk (1:51:15.080)
talk about communism, Marxism, and Stalin.
Lex Fridman (1:51:20.080)
I mean, there's a lot of human history.
Elon Musk (1:51:24.080)
Most of it is actually people just getting on with their lives,
Lex Fridman (1:51:28.080)
and it's not like human history is just nonstop war and disaster.
Elon Musk (1:51:35.080)
Those are actually just, those are intermittent and rare.
Lex Fridman (1:51:38.080)
If they weren't, then humans would soon cease to exist.
Lex Fridman (1:51:46.080)
But it's just that wars tend to be written about a lot,
Lex Fridman (1:51:50.080)
whereas something being like, well,
Elon Musk (1:51:54.080)
a normal year where nothing major happened doesn't get written about much.
Lex Fridman (1:51:58.080)
But that's, most people just like farming and kind of living their life,
Elon Musk (1:52:04.080)
being a villager somewhere.
Lex Fridman (1:52:09.080)
And every now and again, there's a war.
Lex Fridman (1:52:16.080)
And I would have to say, there aren't very many books where I just had to stop reading
Lex Fridman (1:52:23.080)
because it was just too dark.
Lex Fridman (1:52:26.080)
But the book about Stalin, The Court of the Red Tsar, I had to stop reading.
Lex Fridman (1:52:32.080)
It was just too dark and rough.
Elon Musk (1:52:37.080)
Yeah.
Lex Fridman (1:52:39.080)
The 30s, there's a lot of lessons there to me,
Elon Musk (1:52:44.080)
in particular that it feels like humans,
Lex Fridman (1:52:48.080)
like all of us have that, it's the old Solzhenitsyn line,
Elon Musk (1:52:53.080)
that the line between good and evil runs through the heart of every man,
Lex Fridman (1:52:56.080)
that all of us are capable of evil, all of us are capable of good.
Elon Musk (1:52:59.080)
It's almost like this kind of responsibility that all of us have
Lex Fridman (1:53:04.080)
to tend towards the good.
Lex Fridman (1:53:07.080)
And so to me, looking at history is almost like an example of,
Lex Fridman (1:53:11.080)
look, you have some charismatic leader that convinces you of things.
Elon Musk (1:53:16.080)
It's too easy, based on that story, to do evil onto each other,
Lex Fridman (1:53:21.080)
onto your family, onto others.
Lex Fridman (1:53:23.080)
And so it's like our responsibility to do good.
Lex Fridman (1:53:26.080)
It's not like now is somehow different from history.
Elon Musk (1:53:29.080)
That can happen again. All of it can happen again.
Lex Fridman (1:53:32.080)
And yes, most of the time, you're right,
Elon Musk (1:53:35.080)
the optimistic view here is mostly people are just living life.
Lex Fridman (1:53:39.080)
And as you've often memed about, the quality of life was way worse
Elon Musk (1:53:44.080)
back in the day, and this keeps improving over time
Lex Fridman (1:53:47.080)
through innovation, through technology.
Lex Fridman (1:53:49.080)
But still, it's somehow notable that these blimps of atrocities happen.
Lex Fridman (1:53:54.080)
Sure.
Elon Musk (1:53:56.080)
Yeah, I mean, life was really tough for most of history.
Lex Fridman (1:54:02.080)
I mean, for most of human history, a good year would be one
Elon Musk (1:54:07.080)
where not that many people in your village died of the plague,
Lex Fridman (1:54:11.080)
starvation, freezing to death, or being killed by a neighboring village.
Elon Musk (1:54:16.080)
It's like, well, it wasn't that bad.
Lex Fridman (1:54:18.080)
It was only like we lost 5% this year.
Elon Musk (1:54:20.080)
That was a good year.
Lex Fridman (1:54:23.080)
That would be par for the course.
Elon Musk (1:54:25.080)
Just not starving to death would have been the primary goal
Lex Fridman (1:54:28.080)
of most people throughout history,
Elon Musk (1:54:31.080)
is making sure we'll have enough food to last through the winter
Lex Fridman (1:54:34.080)
and not freeze or whatever.
Elon Musk (1:54:36.080)
So, now food is plentiful.
Lex Fridman (1:54:42.080)
I have an obesity problem.
Elon Musk (1:54:46.080)
Well, yeah, the lesson there is to be grateful for the way things are now
Lex Fridman (1:54:50.080)
for some of us.
Elon Musk (1:54:53.080)
We've spoken about this offline.
Lex Fridman (1:54:56.080)
I'd love to get your thought about it here.
Elon Musk (1:55:00.080)
If I sat down for a long form in person conversation with the President of Russia,
Lex Fridman (1:55:05.080)
Vladimir Putin, would you potentially want to call in for a few minutes
Lex Fridman (1:55:10.080)
to join in on a conversation with him, moderated and translated by me?
Lex Fridman (1:55:15.080)
Sure, yeah.
Elon Musk (1:55:16.080)
Sure, I'd be happy to do that.
Lex Fridman (1:55:19.080)
You've shown interest in the Russian language.
Lex Fridman (1:55:22.080)
Is this grounded in your interest in history of linguistics, culture, general curiosity?
Lex Fridman (1:55:27.080)
I think it sounds cool.
Elon Musk (1:55:29.080)
Sounds cool and that looks cool.
Lex Fridman (1:55:32.080)
Well, it takes a moment to read Cyrillic.
Elon Musk (1:55:39.080)
Once you know what the Cyrillic characters stand for,
Lex Fridman (1:55:43.080)
actually then reading Russian becomes a lot easier
Elon Musk (1:55:47.080)
because there are a lot of words that are actually the same.
Lex Fridman (1:55:49.080)
Like bank is bank.
Lex Fridman (1:55:54.080)
So find the words that are exactly the same and now you start to understand Cyrillic.
Lex Fridman (1:55:59.080)
If you can sound it out, there's at least some commonality of words.
Lex Fridman (1:56:06.080)
What about the culture?
Lex Fridman (1:56:09.080)
You love great engineering, physics.
Elon Musk (1:56:12.080)
There's a tradition of the sciences there.
Lex Fridman (1:56:14.080)
You look at the 20th century from rocketry.
Elon Musk (1:56:17.080)
Some of the greatest rockets, some of the space exploration has been done in the former Soviet Union.
Lex Fridman (1:56:24.080)
So do you draw inspiration from that history?
Elon Musk (1:56:27.080)
Just how this culture that in many ways, one of the sad things is because of the language,
Lex Fridman (1:56:33.080)
a lot of it is lost to history because it's not translated.
Elon Musk (1:56:37.080)
Because it is in some ways an isolated culture.
Lex Fridman (1:56:40.080)
It flourishes within its borders.
Lex Fridman (1:56:45.080)
So do you draw inspiration from those folks, from the history of science engineering there?
Elon Musk (1:56:51.080)
The Soviet Union, Russia and Ukraine as well have a really strong history in space flight.
Elon Musk (1:57:02.080)
Some of the most advanced and impressive things in history were done by the Soviet Union.
Lex Fridman (1:57:11.080)
So one cannot help but admire the impressive rocket technology that was developed.
Elon Musk (1:57:20.080)
After the fall of the Soviet Union, there's much less that then happened.
Lex Fridman (1:57:29.080)
But still things are happening, but it's not quite at the frenetic pace that it was happening before the Soviet Union kind of dissolved into separate republics.
Elon Musk (1:57:46.080)
Yeah, there's Roscosmos, the Russian agency.
Lex Fridman (1:57:52.080)
I look forward to a time when those countries with China are working together.
Elon Musk (1:57:57.080)
The United States are all working together.
Lex Fridman (1:58:00.080)
Maybe a little bit of friendly competition.
Elon Musk (1:58:02.080)
I think friendly competition is good.
Elon Musk (1:58:04.080)
Governments are slow and the only thing slower than one government is a collection of governments.
Lex Fridman (1:58:09.080)
So the Olympics would be boring if everyone just crossed the finishing line at the same time.
Lex Fridman (1:58:16.080)
Nobody would watch.
Lex Fridman (1:58:18.080)
And people wouldn't try hard to run fast and stuff.
Lex Fridman (1:58:22.080)
So I think friendly competition is a good thing.
Elon Musk (1:58:25.080)
This is also a good place to give a shout out to a video titled,
Elon Musk (1:58:29.080)
The Entire Soviet Rocket Engine Family Tree by Tim Dodd, AKA Everyday Astronaut.
Elon Musk (1:58:34.080)
It's like an hour and a half.
Lex Fridman (1:58:35.080)
It gives the full history of Soviet rockets.
Lex Fridman (1:58:38.080)
And people should definitely go check out and support Tim in general.
Lex Fridman (1:58:41.080)
That guy is super excited about the future, super excited about space flight.
Elon Musk (1:58:45.080)
Every time I see anything by him, I just have a stupid smile on my face because he's so excited about stuff.
Lex Fridman (1:58:50.080)
Yeah, Tim Dodd is really great.
Elon Musk (1:58:53.080)
If you're interested in anything to do with space, he's, in terms of explaining rocket technology to your average person, he's awesome.
Lex Fridman (1:59:02.080)
The best, I'd say.
Lex Fridman (1:59:04.080)
And I should say like the part of the reason like I switched us from like Rafter at one point was going to be a hydrogen engine.
Lex Fridman (1:59:14.080)
But hydrogen has a lot of challenges.
Elon Musk (1:59:17.080)
It's very low density.
Lex Fridman (1:59:18.080)
It's a deep cryogen.
Lex Fridman (1:59:19.080)
So it's only liquid at a very, very close to absolute zero.
Lex Fridman (1:59:23.080)
Requires a lot of insulation.
Lex Fridman (1:59:26.080)
So it is a lot of challenges there.
Lex Fridman (1:59:30.080)
And I was actually reading a bit about Russian rocket engine developments.
Lex Fridman (1:59:35.080)
And at least the impression I had was that the Soviet Union, Russia and Ukraine primarily were actually in the process of switching to Methalox.
Lex Fridman (1:59:50.080)
And there were some interesting tests and data for ISP.
Elon Musk (1:59:55.080)
Like they were able to get like up to like a 380 second ISP with the Methalox engine.
Lex Fridman (20:00.400)
And like I said, it's much harder to
Elon Musk (20:06.720)
take an advanced technology product
Lex Fridman (20:09.520)
and bring it into volume manufacturing
Elon Musk (20:11.000)
than it is to design it in the first place.
Lex Fridman (20:12.880)
My orders of magnitude.
Lex Fridman (20:14.480)
So let's say you're trying to figure out
Lex Fridman (20:17.800)
is like why is this part or product expensive?
Elon Musk (20:23.960)
Is it because of something fundamentally foolish
Lex Fridman (20:27.400)
that we're doing or is it because our volume is too low?
Lex Fridman (20:31.320)
And so then you say, okay, well, what if our volume
Lex Fridman (20:32.840)
was a million units a year?
Lex Fridman (20:34.220)
Is it still expensive?
Lex Fridman (20:35.740)
That's what I mean by thinking about things in the limit.
Elon Musk (20:38.120)
If it's still expensive at a million units a year,
Lex Fridman (20:40.140)
then volume is not the reason why your thing is expensive.
Elon Musk (20:42.500)
There's something fundamental about the design.
Lex Fridman (20:44.700)
And then you then can focus on reducing the complexity
Elon Musk (20:47.440)
or something like that in the design.
Lex Fridman (20:48.280)
You gotta change the design to change the part
Elon Musk (20:50.600)
to be something that is not fundamentally expensive.
Lex Fridman (20:56.440)
But like that's a common thing in rocketry
Elon Musk (20:58.920)
because the unit volume is relatively low.
Lex Fridman (21:01.840)
And so a common excuse would be,
Elon Musk (21:04.120)
well, it's expensive because our unit volume is low.
Lex Fridman (21:06.480)
And if we were in like automotive or something like that,
Elon Musk (21:08.760)
or consumer electronics, then our costs would be lower.
Lex Fridman (21:10.920)
I'm like, okay, so let's say
Elon Musk (21:13.080)
now you're making a million units a year.
Lex Fridman (21:14.620)
Is it still expensive?
Elon Musk (21:16.080)
If the answer is yes, then economies of scale
Lex Fridman (21:20.560)
are not the issue.
Lex Fridman (21:22.120)
Do you throw into manufacturing,
Lex Fridman (21:24.120)
do you throw like supply chain?
Elon Musk (21:26.080)
Talked about resources and materials and stuff like that.
Lex Fridman (21:28.520)
Do you throw that into the calculation
Elon Musk (21:29.960)
of trying to reason from first principles,
Lex Fridman (21:31.800)
like how we're gonna make the supply chain work here?
Elon Musk (21:34.640)
Yeah, yeah.
Lex Fridman (21:35.760)
And then the cost of materials, things like that.
Lex Fridman (21:37.840)
Or is that too much?
Lex Fridman (21:38.960)
Exactly, so like a good example,
Elon Musk (21:41.880)
I think of thinking about things in the limit
Lex Fridman (21:44.440)
is if you take any machine or whatever,
Elon Musk (21:54.360)
like take a rocket or whatever,
Lex Fridman (21:56.040)
and say, if you look at the raw materials in the rocket,
Lex Fridman (22:03.640)
so you're gonna have like aluminum, steel, titanium,
Lex Fridman (22:07.560)
Inconel, specialty alloys, copper,
Lex Fridman (22:12.560)
and you say, what's the weight of the constituent elements,
Lex Fridman (22:19.200)
of each of these elements,
Lex Fridman (22:20.440)
and what is their raw material value?
Lex Fridman (22:22.600)
And that sets the asymptotic limit
Elon Musk (22:25.720)
for how low the cost of the vehicle can be
Lex Fridman (22:29.480)
unless you change the materials.
Elon Musk (22:31.800)
So, and then when you do that,
Lex Fridman (22:33.960)
I call it like maybe the magic one number
Elon Musk (22:35.760)
or something like that.
Lex Fridman (22:36.600)
So that would be like, if you had the,
Elon Musk (22:40.320)
like just a pile of these raw materials,
Lex Fridman (22:42.440)
again, you could wave the magic wand
Lex Fridman (22:43.680)
and rearrange the atoms into the final shape.
Lex Fridman (22:47.160)
That would be the lowest possible cost
Elon Musk (22:49.600)
that you could make this thing for
Lex Fridman (22:51.000)
unless you change the materials.
Lex Fridman (22:52.720)
So then, and that is almost always a very low number.
Lex Fridman (22:57.760)
So then what's actually causing things to be expensive
Elon Musk (23:01.200)
is how you put the atoms into the desired shape.
Lex Fridman (23:06.000)
Yeah, actually, if you don't mind me taking a tiny tangent,
Elon Musk (23:10.120)
I often talk to Jim Keller,
Lex Fridman (23:11.440)
who's somebody that worked with you as a friend.
Elon Musk (23:14.440)
Jim was, yeah, did great work at Tesla.
Lex Fridman (23:17.760)
So I suppose he carries the flame
Elon Musk (23:20.480)
with the same kind of thinking
Lex Fridman (23:22.680)
that you're talking about now.
Lex Fridman (23:26.240)
And I guess I see that same thing
Lex Fridman (23:27.880)
at Tesla and SpaceX folks who work there,
Elon Musk (23:31.800)
they kind of learn this way of thinking
Lex Fridman (23:33.760)
and it kind of becomes obvious almost.
Lex Fridman (23:36.720)
But anyway, I had argument, not argument,
Lex Fridman (23:39.680)
but he educated me about how cheap it might be
Elon Musk (23:44.800)
to manufacture a Tesla bot.
Lex Fridman (23:46.600)
We just, we had an argument.
Lex Fridman (23:48.360)
How can you reduce the cost of scale of producing a robot?
Lex Fridman (23:52.120)
Because I've gotten the chance to interact quite a bit,
Elon Musk (23:55.960)
obviously, in the academic circles with humanoid robots
Lex Fridman (23:59.480)
and then Boston Dynamics and stuff like that.
Lex Fridman (24:01.800)
And they're very expensive to build.
Lex Fridman (24:04.520)
And then Jim kind of schooled me on saying like,
Elon Musk (24:07.600)
okay, like this kind of first principles thinking
Lex Fridman (24:10.080)
of how can we get the cost of manufacturing down?
Elon Musk (24:13.760)
I suppose you do that,
Lex Fridman (24:14.720)
you have done that kind of thinking for Tesla bot
Lex Fridman (24:17.640)
and for all kinds of complex systems
Lex Fridman (24:21.880)
that are traditionally seen as complex.
Lex Fridman (24:23.680)
And you say, okay, how can we simplify everything down?
Lex Fridman (24:27.160)
Yeah, I mean, I think if you are really good
Elon Musk (24:30.320)
at manufacturing, you can basically make,
Lex Fridman (24:34.640)
at high volume, you can basically make anything
Elon Musk (24:36.560)
for a cost that asymptotically approaches
Lex Fridman (24:40.640)
the raw material value of the constituents
Elon Musk (24:42.880)
plus any intellectual property that you need to do license.
Lex Fridman (24:46.480)
Anything.
Lex Fridman (24:49.440)
But it's hard.
Lex Fridman (24:50.280)
It's not like that's a very hard thing to do,
Lex Fridman (24:51.840)
but it is possible for anything.
Lex Fridman (24:54.720)
Anything in volume can be made, like I said,
Elon Musk (24:57.480)
for a cost that asymptotically approaches
Lex Fridman (25:00.440)
this raw material constituents
Elon Musk (25:02.760)
plus intellectual property license rights.
Lex Fridman (25:05.360)
So what'll often happen in trying to design a product
Elon Musk (25:08.520)
is people will start with the tools and parts
Lex Fridman (25:11.960)
and methods that they are familiar with
Lex Fridman (25:14.520)
and then try to create the product
Lex Fridman (25:17.480)
using their existing tools and methods.
Elon Musk (25:21.240)
The other way to think about it is actually imagine the,
Lex Fridman (25:25.080)
try to imagine the platonic ideal of the perfect product
Elon Musk (25:28.800)
or technology, whatever it might be.
Lex Fridman (25:31.360)
And say, what is the perfect arrangement of atoms
Lex Fridman (25:35.680)
that would be the best possible product?
Lex Fridman (25:38.560)
And now let us try to figure out
Lex Fridman (25:39.840)
how to get the atoms in that shape.
Lex Fridman (25:43.880)
I mean, it sounds,
Elon Musk (25:48.000)
it's almost like a Rick and Morty absurd
Lex Fridman (25:50.400)
until you start to really think about it
Lex Fridman (25:52.080)
and you really should think about it in this way
Lex Fridman (25:56.440)
because everything else is kind of,
Elon Musk (25:59.000)
if you think, you might fall victim to the momentum
Lex Fridman (26:03.000)
of the way things were done in the past
Elon Musk (26:04.440)
unless you think in this way.
Lex Fridman (26:06.000)
Well, just as a function of inertia,
Elon Musk (26:07.680)
people will want to use the same tools and methods
Lex Fridman (26:10.640)
that they are familiar with.
Elon Musk (26:13.720)
That's what they'll do by default.
Lex Fridman (26:15.720)
And then that will lead to an outcome of things
Elon Musk (26:18.720)
that can be made with those tools and methods
Lex Fridman (26:20.480)
but is unlikely to be the platonic ideal
Elon Musk (26:23.880)
of the perfect product.
Lex Fridman (26:25.040)
So then, so that's why it's good to think of things
Elon Musk (26:28.360)
in both directions.
Lex Fridman (26:29.200)
So like, what can we build with the tools that we have?
Lex Fridman (26:31.240)
But then, but also what is the perfect,
Lex Fridman (26:34.600)
the theoretical perfect product look like?
Lex Fridman (26:36.320)
And that theoretical perfect part
Lex Fridman (26:38.600)
is gonna be a moving target
Elon Musk (26:39.560)
because as you learn more,
Lex Fridman (26:41.720)
the definition of that perfect product will change
Elon Musk (26:45.960)
because you don't actually know what the perfect product is,
Lex Fridman (26:47.720)
but you can successfully approximate a more perfect product.
Lex Fridman (26:52.120)
So think about it like that and then saying,
Lex Fridman (26:54.520)
okay, now what tools, methods, materials,
Elon Musk (26:57.320)
whatever do we need to create
Lex Fridman (27:00.400)
in order to get the atoms in that shape?
Lex Fridman (27:03.880)
But people rarely think about it that way.
Lex Fridman (27:07.880)
But it's a powerful tool.
Elon Musk (27:10.480)
I should mention that the brilliant Siobhan Ziles
Lex Fridman (27:13.840)
is hanging out with us in case you hear a voice
Elon Musk (27:17.840)
of wisdom from outside, from up above.
Lex Fridman (27:23.720)
Okay, so let me ask you about Mars.
Elon Musk (27:26.320)
You mentioned it would be great for science
Lex Fridman (27:28.520)
to put a base on the moon to do some research.
Lex Fridman (27:32.960)
But the truly big leap, again,
Lex Fridman (27:36.760)
in this category of seemingly impossible,
Elon Musk (27:38.920)
is to put a human being on Mars.
Lex Fridman (27:41.880)
When do you think SpaceX will land a human being on Mars?
Elon Musk (27:44.920)
Hmm, best case is about five years, worst case, 10 years.
Lex Fridman (27:54.480)
Okay, so I'm gonna ask you about Mars.
Elon Musk (27:57.480)
You mentioned it would be great for science
Lex Fridman (28:00.480)
to put a base on the moon to do some research.
Lex Fridman (28:03.480)
But the truly big leap, again,
Lex Fridman (28:05.680)
in this category of seemingly impossible,
Elon Musk (28:08.760)
is to put a human being on Mars.
Lex Fridman (28:11.240)
But the truly big leap, again,
Elon Musk (28:15.040)
in this category of seemingly impossible,
Lex Fridman (28:18.040)
is to put a human being on Mars.
Lex Fridman (28:20.480)
When do you think SpaceX will land a human being on Mars?
Lex Fridman (28:23.480)
Hmm, best case is about five years, worst case, 10 years.
Lex Fridman (28:31.760)
What are the determining factors, would you say,
Lex Fridman (28:34.520)
from an engineering perspective?
Lex Fridman (28:36.240)
Or is that not the bottlenecks?
Lex Fridman (28:37.840)
I don't know, order of magnitude or something like that.
Elon Musk (28:40.760)
It's a lot, it's really next level.
Lex Fridman (28:43.000)
So, and the fundamental optimization of Starship
Elon Musk (28:49.200)
is minimizing cost per ton to orbit,
Lex Fridman (28:51.480)
and ultimately cost per ton to the surface of Mars.
Elon Musk (28:54.760)
This may seem like a mercantile objective,
Lex Fridman (28:56.320)
but it is actually the thing that needs to be optimized.
Elon Musk (29:00.320)
Like there is a certain cost per ton to the surface of Mars
Lex Fridman (29:04.000)
where we can afford to establish a self sustaining city,
Lex Fridman (29:08.800)
and then above that, we cannot afford to do it.
Lex Fridman (29:12.760)
So right now, you couldn't fly to Mars for a trillion dollars.
Elon Musk (29:16.760)
No amount of money could get you a ticket to Mars.
Lex Fridman (29:19.120)
So we need to get that above,
Elon Musk (29:22.440)
to get that something that is actually possible at all.
Lex Fridman (29:27.760)
But that's, we don't just want to have,
Elon Musk (29:32.240)
with Mars, flags and footprints,
Lex Fridman (29:33.800)
and then not come back for a half century,
Elon Musk (29:35.880)
like we did with the Moon.
Lex Fridman (29:37.920)
In order to pass a very important, great filter,
Elon Musk (29:43.360)
I think we need to be a multi planet species.
Lex Fridman (29:48.320)
This may sound somewhat esoteric to a lot of people,
Lex Fridman (29:51.400)
but eventually, given enough time,
Lex Fridman (29:57.160)
there's something, Earth is likely to experience
Lex Fridman (2:00:01.080)
And I was like, well, OK, that's actually really impressive.
Lex Fridman (2:00:05.080)
So I think you could actually get a much lower cost, like in optimizing cost per ton to orbit, cost per ton to Mars.
Elon Musk (2:00:19.080)
I think Methalox is the way to go.
Lex Fridman (2:00:25.080)
And I was partly inspired by the Russian work on the test stands with Methalox engines.
Lex Fridman (2:00:32.080)
And now for something completely different.
Lex Fridman (2:00:35.080)
Do you mind doing a bit of a meme review in the spirit of the great, the powerful PewDiePie?
Elon Musk (2:00:41.080)
Let's say 1 to 11.
Lex Fridman (2:00:42.080)
Just go over a few documents printed out.
Elon Musk (2:00:45.080)
We can try.
Lex Fridman (2:00:46.080)
Let's try this.
Elon Musk (2:00:49.080)
I present to you document numero uno.
Lex Fridman (2:00:56.080)
I don't know. OK.
Elon Musk (2:00:58.080)
Vladimir Impaler discovers marshmallows.
Lex Fridman (2:01:03.080)
That's not bad.
Lex Fridman (2:01:07.080)
So you get it? Because he likes impaling things.
Lex Fridman (2:01:11.080)
Yes, I get it.
Elon Musk (2:01:12.080)
I don't know, three, whatever.
Lex Fridman (2:01:14.080)
That's not very good.
Elon Musk (2:01:19.080)
This is grounded in some engineering, some history.
Lex Fridman (2:01:28.080)
Yeah, give us an eight out of ten.
Lex Fridman (2:01:31.080)
What do you think about nuclear power?
Lex Fridman (2:01:33.080)
I'm in favor of nuclear power.
Elon Musk (2:01:34.080)
I think in a place that is not subject to extreme natural disasters, I think nuclear power is a great way to generate electricity.
Lex Fridman (2:01:47.080)
I don't think we should be shutting down nuclear power stations.
Lex Fridman (2:01:51.080)
Yeah, but what about Chernobyl?
Lex Fridman (2:01:53.080)
Exactly.
Elon Musk (2:01:56.080)
I think there's a lot of fear of radiation and stuff.
Lex Fridman (2:02:04.080)
The problem is a lot of people just don't study engineering or physics.
Elon Musk (2:02:13.080)
Just the word radiation just sounds scary.
Lex Fridman (2:02:16.080)
They can't calibrate what radiation means.
Lex Fridman (2:02:21.080)
But radiation is much less dangerous than you think.
Lex Fridman (2:02:30.080)
For example, Fukushima, when the Fukushima problem happened due to the tsunami,
Elon Musk (2:02:41.080)
I got people in California asking me if they should worry about radiation from Fukushima.
Lex Fridman (2:02:46.080)
I'm like, definitely not, not even slightly, not at all.
Elon Musk (2:02:51.080)
That is crazy.
Elon Musk (2:02:54.080)
Just to show this is how the danger is so much overplayed compared to what it really is that I actually flew to Fukushima.
Elon Musk (2:03:09.080)
I donated a solar power system for a water treatment plant, and I made a point of eating locally grown vegetables on TV in Fukushima.
Lex Fridman (2:03:28.080)
I'm still alive.
Lex Fridman (2:03:31.080)
So it's not even that the risk of these events is low, but the impact of them is...
Lex Fridman (2:03:36.080)
The impact is greatly exaggerated.
Elon Musk (2:03:38.080)
It's human nature.
Lex Fridman (2:03:40.080)
People don't know what radiation is.
Lex Fridman (2:03:42.080)
I've had people ask me, what about radiation from cell phones causing brain cancer?
Lex Fridman (2:03:46.080)
I'm like, when you say radiation, do you mean photons or particles?
Lex Fridman (2:03:49.080)
They're like, I don't know, what do you mean photons or particles?
Lex Fridman (2:03:52.080)
Do you mean, let's say, photons?
Lex Fridman (2:03:56.080)
What frequency or wavelength?
Lex Fridman (2:03:59.080)
And they're like, I have no idea.
Lex Fridman (2:04:01.080)
Do you know that everything's radiating all the time?
Lex Fridman (2:04:04.080)
What do you mean?
Elon Musk (2:04:06.080)
Like, yeah, everything's radiating all the time.
Lex Fridman (2:04:08.080)
Photons are being emitted by all objects all the time, basically.
Lex Fridman (2:04:12.080)
And if you want to know what it means to stand in front of nuclear fire, go outside.
Lex Fridman (2:04:21.080)
The sun is a gigantic thermonuclear reactor that you're staring right at it.
Lex Fridman (2:04:28.080)
Are you still alive?
Lex Fridman (2:04:29.080)
Yes.
Elon Musk (2:04:30.080)
Okay, amazing.
Elon Musk (2:04:32.080)
Yeah, I guess radiation is one of the words that could be used as a tool to fear monger by certain people.
Elon Musk (2:04:39.080)
That's it.
Lex Fridman (2:04:40.080)
I think people just don't understand.
Elon Musk (2:04:42.080)
I mean, that's the way to fight that fear, I suppose, is to understand, is to learn.
Lex Fridman (2:04:46.080)
Yeah, just say, okay, how many people have actually died from nuclear accidents?
Elon Musk (2:04:50.080)
It's practically nothing.
Lex Fridman (2:04:51.080)
And say, how many people have died from coal plants?
Lex Fridman (2:04:57.080)
And it's a very big number.
Elon Musk (2:04:59.080)
So, like, obviously we should not be starting up coal plants and shutting down nuclear plants.
Elon Musk (2:05:05.080)
It just doesn't make any sense at all.
Elon Musk (2:05:08.080)
Coal plants, like, I don't know, 100 to 1,000 times worse for health than nuclear power plants.
Lex Fridman (2:05:15.080)
You want to go to the next one?
Lex Fridman (2:05:16.080)
This is really bad.
Elon Musk (2:05:20.080)
It's 90, 180, and 360 degrees.
Lex Fridman (2:05:24.080)
Everybody loves the math.
Elon Musk (2:05:25.080)
Nobody gives a shit about 270.
Lex Fridman (2:05:28.080)
It's not super funny.
Elon Musk (2:05:30.080)
I don't know, like, 203.
Lex Fridman (2:05:32.080)
This is not a, you know, LOL situation.
Elon Musk (2:05:37.080)
Yeah.
Lex Fridman (2:05:43.080)
That was pretty good.
Elon Musk (2:05:44.080)
The United States oscillating between establishing and destroying dictatorships.
Lex Fridman (2:05:48.080)
It's like a metronome.
Lex Fridman (2:05:49.080)
Is that a metronome?
Lex Fridman (2:05:50.080)
Yeah, it's out of 7 out of 10.
Elon Musk (2:05:53.080)
It's kind of true.
Lex Fridman (2:05:54.080)
Oh, yeah, this is kind of personal for me.
Elon Musk (2:05:57.080)
Next one.
Lex Fridman (2:05:58.080)
Oh, man.
Lex Fridman (2:05:59.080)
Is this Leica?
Lex Fridman (2:06:00.080)
Yeah.
Elon Musk (2:06:01.080)
Well, no.
Lex Fridman (2:06:02.080)
Or it's like referring to Leica or something?
Elon Musk (2:06:03.080)
As Leica's, like, husband.
Lex Fridman (2:06:06.080)
Husband.
Elon Musk (2:06:07.080)
Yeah, yeah.
Lex Fridman (2:06:08.080)
Hello.
Elon Musk (2:06:09.080)
Yes, this is dog.
Lex Fridman (2:06:10.080)
Your wife was launched into space.
Lex Fridman (2:06:11.080)
And then the last one is him with his eyes closed and a bottle of vodka.
Lex Fridman (2:06:16.080)
Yeah.
Elon Musk (2:06:17.080)
Leica didn't come back.
Lex Fridman (2:06:18.080)
No.
Elon Musk (2:06:19.080)
They don't tell you the full story of, you know, what the impact they had on the loved
Lex Fridman (2:06:24.080)
ones.
Elon Musk (2:06:25.080)
True.
Lex Fridman (2:06:26.080)
Yeah.
Elon Musk (2:06:27.080)
It's like 711 for me.
Lex Fridman (2:06:28.080)
Sure.
Elon Musk (2:06:29.080)
The Soviet shadow.
Lex Fridman (2:06:30.080)
Oh, yeah.
Elon Musk (2:06:31.080)
This keeps going on the Russian theme.
Lex Fridman (2:06:33.400)
First man in space.
Elon Musk (2:06:35.160)
Nobody cares.
Lex Fridman (2:06:36.160)
First man on the moon.
Elon Musk (2:06:37.160)
Well, I think people do care.
Lex Fridman (2:06:38.160)
No, I know.
Elon Musk (2:06:39.160)
But...
Lex Fridman (2:06:40.160)
There is...
Elon Musk (2:06:41.160)
Yuri Gagarin's name will be forever in history, I think.
Elon Musk (2:06:45.760)
There is something special about placing, like, stepping foot onto another totally foreign
Elon Musk (2:06:52.080)
land.
Lex Fridman (2:06:53.080)
It's not the journey, like, people that explore the oceans.
Elon Musk (2:06:56.960)
It's not as important to explore the oceans as to land on a whole new continent.
Lex Fridman (2:07:01.400)
Yeah.
Elon Musk (2:07:02.400)
Well, this is about you.
Lex Fridman (2:07:05.080)
Oh, yeah, I'd love to get your comment on this.
Elon Musk (2:07:08.200)
Elon Musk, after sending 6.6 billion dollars to the UN to end world hunger, you have three
Lex Fridman (2:07:14.680)
hours.
Elon Musk (2:07:15.680)
Yeah, well, I mean, obviously, 6 billion dollars is not going to end world hunger.
Lex Fridman (2:07:24.240)
So I mean, the reality is at this point, the world is producing far more food than it can
Elon Musk (2:07:30.680)
really consume.
Lex Fridman (2:07:31.680)
Like, we don't have a caloric constraint at this point.
Lex Fridman (2:07:35.920)
So where there is hunger, it is almost always due to, like, civil war or strife or some
Elon Musk (2:07:43.280)
like, it's not a thing that is extremely rare for it to be just a matter of, like, lack
Elon Musk (2:07:52.860)
of money.
Elon Musk (2:07:53.860)
It's like, you know, it's like some civil war in some country and like one part of the
Elon Musk (2:07:59.000)
country is literally trying to starve the other part of the country.
Lex Fridman (2:08:03.040)
So it's much more complex than something that money could solve.
Elon Musk (2:08:05.840)
It's geopolitics.
Lex Fridman (2:08:07.440)
It's a lot of things.
Elon Musk (2:08:09.840)
It's human nature.
Lex Fridman (2:08:10.840)
It's governments.
Elon Musk (2:08:11.840)
It's money, monetary systems, all that kind of stuff.
Lex Fridman (2:08:14.520)
Yeah, food is extremely cheap these days.
Elon Musk (2:08:17.640)
It's like, I mean, the US at this point, you know, among low income families, obesity is
Lex Fridman (2:08:26.720)
actually another problem.
Elon Musk (2:08:27.960)
It's not, like, obesity, it's not hunger.
Lex Fridman (2:08:31.120)
It's like too much, you know, too many calories.
Lex Fridman (2:08:34.660)
So it's not that nobody's hungry anywhere.
Lex Fridman (2:08:37.880)
It's just, this is not a simple matter of adding money and solving it.
Elon Musk (2:08:43.960)
Hmm.
Lex Fridman (2:08:44.960)
What do you think that one gets?
Elon Musk (2:08:48.800)
It's getting...
Lex Fridman (2:08:49.800)
Two.
Lex Fridman (2:08:50.800)
We're just going after empires, world, where did you get those artifacts?
Lex Fridman (2:08:57.560)
The British Museum.
Elon Musk (2:08:58.560)
Shout out to Monty Python.
Lex Fridman (2:09:01.600)
We found them.
Elon Musk (2:09:02.600)
Yeah.
Lex Fridman (2:09:03.600)
The British Museum is pretty great.
Elon Musk (2:09:05.800)
I mean, admittedly Britain did take these historical artifacts from all around the world
Lex Fridman (2:09:10.200)
and put them in London, but, you know, it's not like people can't go see them.
Lex Fridman (2:09:16.120)
So it is a convenient place to see these ancient artifacts is London for, you know, for a large
Lex Fridman (2:09:23.480)
segment of the world.
Lex Fridman (2:09:25.080)
So I think, you know, on balance, the British Museum is a net good, although I'm sure a
Lex Fridman (2:09:29.640)
lot of countries will argue about that.
Elon Musk (2:09:31.440)
Yeah.
Elon Musk (2:09:32.440)
It's like you want to make these historical artifacts accessible to as many people as
Elon Musk (2:09:35.960)
possible and the British Museum, I think, does a good job of that.
Elon Musk (2:09:41.520)
Even if there's a darker aspect to like the history of empire in general, whatever the
Elon Musk (2:09:45.200)
empire is, however things were done, it is the history that happened.
Lex Fridman (2:09:52.120)
You can't sort of erase that history, unfortunately.
Elon Musk (2:09:54.280)
You could just become better in the future.
Lex Fridman (2:09:55.920)
That's the point.
Elon Musk (2:09:57.520)
Yeah.
Lex Fridman (2:09:58.520)
I mean, it's like, well, how are we going to pass moral judgment on these things?
Elon Musk (2:10:04.040)
Like it's like if, you know, if one is going to judge, say, the Russian Empire, you've
Elon Musk (2:10:10.400)
got to judge, you know, what everyone was doing at the time and how were the British
Elon Musk (2:10:14.720)
relative to everyone.
Lex Fridman (2:10:18.200)
And I think the British would actually get like a relatively good grade, relatively good
Elon Musk (2:10:22.640)
grade, not in absolute terms, but compared to what everyone else was doing, they were
Lex Fridman (2:10:30.040)
not the worst.
Elon Musk (2:10:31.600)
Like I said, you got to look at these things in the context of the history at the time
Lex Fridman (2:10:35.400)
and say, what were the alternatives and what are you comparing it against?
Lex Fridman (2:10:39.200)
And I do not think it would be the case that Britain would get a bad grade when looking
Lex Fridman (2:10:47.220)
at history at the time.
Elon Musk (2:10:48.560)
You know, if you judge history from, you know, from what is morally acceptable today, you
Lex Fridman (2:10:56.280)
basically are going to give everyone a failing grade.
Elon Musk (2:10:58.400)
I'm not clear.
Elon Musk (2:10:59.400)
It's not, I don't think anyone would get a passing grade in their morality of like you
Lex Fridman (2:11:04.720)
could go back 300 years ago, like who's getting a passing grade?
Lex Fridman (2:11:08.960)
Basically no one.
Lex Fridman (2:11:10.760)
And we might not get a passing grade from generations that come after us.
Lex Fridman (2:11:16.280)
What does that one get?
Elon Musk (2:11:18.880)
Sure.
Lex Fridman (2:11:19.880)
Six, seven.
Elon Musk (2:11:20.880)
For the Monty Python, maybe.
Lex Fridman (2:11:21.880)
I always love Monty Python.
Elon Musk (2:11:22.880)
They're great.
Lex Fridman (2:11:23.880)
The Life of Brian and the Quest of the Holy Grail are incredible.
Elon Musk (2:11:28.880)
Yeah.
Lex Fridman (2:11:29.880)
Yeah.
Elon Musk (2:11:30.880)
Yeah.
Lex Fridman (2:11:31.880)
Those serious eyebrows.
Lex Fridman (2:11:32.880)
How important do you think is facial hair to great leadership?
Lex Fridman (2:11:37.480)
Well, you got a new haircut.
Lex Fridman (2:11:41.480)
How does that affect your leadership?
Lex Fridman (2:11:42.480)
I don't know.
Elon Musk (2:11:43.480)
Hopefully not.
Lex Fridman (2:11:44.480)
It doesn't.
Lex Fridman (2:11:45.480)
Is that the second no one?
Lex Fridman (2:11:46.480)
Yeah.
Elon Musk (2:11:47.480)
The second is no one.
Lex Fridman (2:11:48.480)
There is no one competing with Brezhnev.
Elon Musk (2:11:49.480)
No one, too.
Lex Fridman (2:11:50.480)
Those are like epic eyebrows.
Elon Musk (2:11:51.480)
Yeah.
Lex Fridman (2:11:52.480)
Sure.
Elon Musk (2:11:53.480)
That's ridiculous.
Lex Fridman (2:11:54.480)
Give it a six or seven, I don't know.
Elon Musk (2:11:55.480)
I like this Shakespearean analysis of memes.
Lex Fridman (2:11:56.480)
Brezhnev, he had a flair for drama as well.
Elon Musk (2:11:57.480)
Like, you know, showmanship.
Lex Fridman (2:11:58.480)
Yeah.
Elon Musk (2:11:59.480)
Yeah.
Lex Fridman (2:12:00.480)
It must come from the eyebrows.
Elon Musk (2:12:01.480)
All right.
Elon Musk (2:12:02.480)
Invention, great engineering, look what I invented, that's the best thing since ripped
Elon Musk (2:12:20.960)
up bread.
Lex Fridman (2:12:21.960)
Yeah.
Lex Fridman (2:12:22.960)
Because they invented sliced bread, am I just explaining memes at this point?
Elon Musk (2:12:29.000)
This is where my life has become a meme, what it like, you know, like a scribe that like
Elon Musk (2:12:40.000)
runs around with the kings and just like writes down memes.
Lex Fridman (2:12:44.120)
I mean, when was the cheeseburger invented?
Elon Musk (2:12:46.280)
That's like an epic invention.
Lex Fridman (2:12:47.280)
Yeah.
Elon Musk (2:12:48.280)
Like, like, wow.
Lex Fridman (2:12:49.280)
Yeah.
Elon Musk (2:12:50.280)
Versus just like a burger or a burger, I guess a burger in general is like, you know, then
Elon Musk (2:12:57.680)
there's like, what is a burger, what's a sandwich, and then you start getting a pizza sandwich
Lex Fridman (2:13:01.800)
and what is the original, it gets into an ontology argument.
Lex Fridman (2:13:05.720)
Yeah.
Lex Fridman (2:13:06.720)
But everybody knows like if you order like a burger or cheeseburger or whatever and you
Elon Musk (2:13:08.920)
like, you got like, you know, tomato and some lettuce and onions and whatever and, you know,
Elon Musk (2:13:14.000)
mayor and ketchup and mustard, it's like epic.
Lex Fridman (2:13:16.040)
Yeah.
Lex Fridman (2:13:17.040)
But I'm sure they've had bread and meat separately for a long time and it was kind of a burger
Elon Musk (2:13:21.080)
on the same plate, but somebody who actually combined them into the same thing and then
Elon Musk (2:13:25.960)
you bite it and hold it makes it convenient.
Lex Fridman (2:13:29.040)
It's a materials problem.
Elon Musk (2:13:30.040)
Yeah.
Lex Fridman (2:13:31.040)
Like your hands don't get dirty and whatever.
Elon Musk (2:13:33.520)
Yeah.
Lex Fridman (2:13:34.520)
It's brilliant.
Elon Musk (2:13:35.520)
Well, that is not what I would have guessed, but everyone knows like you, if you order
Elon Musk (2:13:43.280)
a cheeseburger, you know what you're getting, you know, it's not like some obtuse, like,
Elon Musk (2:13:46.800)
well, I wonder what I'll get, you know, um, you know, uh, fries are, I mean, great.
Lex Fridman (2:13:52.800)
I mean, they were the devil, but fries are awesome.
Lex Fridman (2:13:56.320)
And uh, yeah, pizza is incredible.
Lex Fridman (2:14:00.320)
Food innovation doesn't get enough love, I guess is what we're getting at.
Elon Musk (2:14:05.480)
Great.
Elon Musk (2:14:06.480)
Um, uh, what about the, uh, Matthew McConaughey, Austinite here, uh, president Kennedy, do
Lex Fridman (2:14:13.520)
you know how to put men on the moon yet?
Lex Fridman (2:14:15.200)
NASA?
Elon Musk (2:14:16.200)
No.
Lex Fridman (2:14:17.200)
President Kennedy, it'd be a lot cooler if you did.
Elon Musk (2:14:19.440)
Pretty much sure, six, six or seven, I suppose.
Lex Fridman (2:14:25.360)
All right.
Lex Fridman (2:14:26.360)
And this is the last one that's funny.
Lex Fridman (2:14:30.880)
Someone drew a bunch of dicks all over the walls, 16 chapel boys bath.
Elon Musk (2:14:35.680)
Sure.
Lex Fridman (2:14:36.680)
I'll give it a nine.
Elon Musk (2:14:37.680)
It's super.
Lex Fridman (2:14:38.680)
It's really true.
Elon Musk (2:14:39.680)
All right.
Lex Fridman (2:14:40.680)
This is our highest ranking meme for today.
Elon Musk (2:14:41.680)
I mean, it's true.
Lex Fridman (2:14:42.680)
Like, how do they get away with it?
Elon Musk (2:14:44.880)
Lots of nakedness.
Lex Fridman (2:14:45.880)
I mean, dick pics are, I mean, just something throughout history.
Elon Musk (2:14:49.640)
Uh, as long as people can draw things, there's been a dick pic.
Lex Fridman (2:14:53.080)
It's a staple of human history.
Elon Musk (2:14:55.080)
It's a staple.
Lex Fridman (2:14:56.080)
It's just throughout human history.
Elon Musk (2:14:58.840)
You tweeted that you aspire to comedy.
Lex Fridman (2:15:00.680)
You're friends with Joe Rogan.
Elon Musk (2:15:02.640)
Might you, uh, do a short standup comedy set at some point in the future, maybe, um, open
Lex Fridman (2:15:08.080)
for Joe, something like that.
Elon Musk (2:15:09.640)
Is that, is that...
Lex Fridman (2:15:10.640)
Really?
Lex Fridman (2:15:11.640)
Standup?
Lex Fridman (2:15:12.640)
Actual, just full on standup?
Elon Musk (2:15:13.640)
Full on standup.
Lex Fridman (2:15:14.640)
Is that in there or is that...
Elon Musk (2:15:15.640)
It's extremely difficult if, uh, at least that's what, uh, like Joe says and the comedians
Lex Fridman (2:15:22.840)
say.
Elon Musk (2:15:23.840)
Huh.
Elon Musk (2:15:24.840)
I wonder if I could, um, I mean, I, I, you know, I, I have done standup for friends,
Elon Musk (2:15:32.880)
just, uh, impromptu, you know, I'll get, get on like a roof, uh, and they, they do laugh,
Lex Fridman (2:15:40.320)
but they're our friends too.
Lex Fridman (2:15:41.440)
So I don't know if, if you've got to call, you know, like a room of strangers, are they
Lex Fridman (2:15:45.320)
going to actually also find it funny, but I could try, see what happens.
Elon Musk (2:15:51.400)
I think you'd learn something either way.
Lex Fridman (2:15:53.560)
Um, yeah.
Elon Musk (2:15:54.560)
I kind of love, um, both the, when you bomb and when, when you do great, just watching
Lex Fridman (2:16:00.080)
people, how they deal with it, it's so difficult.
Elon Musk (2:16:03.320)
It's so, you're so fragile up there.
Lex Fridman (2:16:07.000)
It's just you and you, you think you're going to be funny.
Lex Fridman (2:16:09.760)
And when it completely falls flat, it's just, it's beautiful to see people deal with like
Lex Fridman (2:16:14.400)
that.
Elon Musk (2:16:15.400)
Yeah.
Lex Fridman (2:16:16.400)
I might have enough material to do standup.
Elon Musk (2:16:17.400)
I've never thought about it, but I might have enough material.
Lex Fridman (2:16:21.240)
Um, I don't know, like 15 minutes or something.
Elon Musk (2:16:25.480)
Oh yeah.
Lex Fridman (2:16:26.480)
Yeah.
Elon Musk (2:16:27.480)
Do a Netflix special.
Lex Fridman (2:16:28.480)
Netflix special.
Elon Musk (2:16:29.480)
Sure.
Elon Musk (2:16:30.480)
Um, what's your favorite Rick and Morty concept, uh, just to spring that on you.
Elon Musk (2:16:36.240)
Is there, there's a lot of sort of scientific engineering ideas explored there.
Lex Fridman (2:16:39.960)
There's the butter robot.
Elon Musk (2:16:42.880)
That's a great, uh, that's a great show.
Lex Fridman (2:16:44.800)
Um, yeah.
Elon Musk (2:16:45.800)
Rick and Morty is awesome.
Lex Fridman (2:16:47.440)
Somebody that's exactly like you from an alternate dimension showed up there.
Elon Musk (2:16:50.560)
Elon Tusk.
Lex Fridman (2:16:51.560)
Yeah, that's right.
Elon Musk (2:16:52.560)
That you voiced.
Lex Fridman (2:16:53.560)
Yeah.
Elon Musk (2:16:54.560)
Rick and Morty certainly explores a lot of interesting concepts.
Lex Fridman (2:16:57.200)
Uh, so like what's the favorite one?
Elon Musk (2:17:00.200)
I don't know.
Elon Musk (2:17:01.200)
The butter robot certainly is, uh, you know, it's like, it's certainly possible to have
Elon Musk (2:17:04.960)
too much sentience in a device.
Lex Fridman (2:17:06.840)
Um, like you don't want to have your toast to be like a super genius toaster.
Elon Musk (2:17:12.080)
It's going to hate, hate life cause all it could do is make his toast.
Lex Fridman (2:17:15.280)
But if it's like, you don't want to have like super intelligent stuck in a very limited
Elon Musk (2:17:19.720)
device.
Elon Musk (2:17:20.720)
Um, do you think it's too easy from a, if we're talking about from the engineering perspective
Elon Musk (2:17:24.960)
of super intelligence, like with Marvin the robot, like, is it, it seems like it might
Lex Fridman (2:17:30.760)
be very easy to engineer just the depressed robot.
Elon Musk (2:17:33.800)
Like it's not obvious to engineer and robot that's going to find a fulfilling existence.
Elon Musk (2:17:40.760)
Sometimes humans I suppose, but, um, I wonder if that's like the default, if you don't do
Elon Musk (2:17:47.160)
a good job on building a robot, it's going to be sad a lot.
Lex Fridman (2:17:52.800)
Well we can reprogram robots easier than we can reprogram humans.
Lex Fridman (2:17:58.160)
So I guess if you let it evolve without tinkering, then it might get a sad, uh, but you can change
Lex Fridman (2:18:06.520)
the optimization function and have it be a cheery robot.
Elon Musk (2:18:13.000)
You uh, like I mentioned with, with SpaceX, you give a lot of people hope and a lot of
Lex Fridman (2:18:17.560)
people look up to you.
Elon Musk (2:18:18.560)
Millions of people look up to you.
Elon Musk (2:18:19.920)
Uh, if we think about young people in high school, maybe in college, um, what advice
Elon Musk (2:18:26.400)
would you give to them about if they want to try to do something big in this world,
Lex Fridman (2:18:31.680)
they want to really have a big positive impact.
Lex Fridman (2:18:33.680)
What advice would you give them about their career, maybe about life in general?
Lex Fridman (2:18:39.440)
Try to be useful.
Elon Musk (2:18:40.440)
Um, you know, do things that are useful to your fellow human beings, to the world.
Lex Fridman (2:18:46.480)
It's very hard to be useful.
Elon Musk (2:18:48.480)
Um, very hard.
Lex Fridman (2:18:51.680)
Um, you know, are you contributing more than you consume?
Lex Fridman (2:18:57.200)
You know, like, uh, like can you try to have a positive net contribution to society?
Elon Musk (2:19:05.760)
Um, I think that's the thing to aim for, you know, not, not to try to be sort of a leader
Elon Musk (2:19:11.600)
for just for the sake of being a leader or whatever.
Elon Musk (2:19:15.200)
Um, a lot of time people, a lot of times the people you want as leaders are the people
Elon Musk (2:19:22.080)
who don't want to be leaders.
Elon Musk (2:19:24.440)
So, um, if you live a useful life, that is a good life, a life worth having lived.
Elon Musk (2:19:36.640)
Um, you know, and I, like I said, I would, I would encourage people to use the mental
Lex Fridman (2:19:45.040)
tools of physics and apply them broadly in life.
Elon Musk (2:19:48.360)
There are the best tools.
Lex Fridman (2:19:49.360)
When you think about education and self education, what do you recommend?
Lex Fridman (2:19:54.040)
So there's the university, there's a self study, there is a hands on sort of finding
Elon Musk (2:20:01.920)
a company or a place or a set of people that do the thing you're passionate about and joining
Elon Musk (2:20:05.840)
them as early as possible.
Elon Musk (2:20:08.080)
Um, there's, uh, taking a road trip across Europe for a few years and writing some poetry,
Lex Fridman (2:20:13.520)
which, uh, which, which trajectory do you suggest?
Elon Musk (2:20:18.440)
In terms of learning about how you can become useful, as you mentioned, how you can have
Elon Musk (2:20:24.000)
the most positive impact.
Elon Musk (2:20:27.320)
Well, I encourage people to read a lot of books, just read, basically try to ingest
Elon Musk (2:20:39.160)
as much information as you can, uh, and try to also just develop a good general knowledge.
Elon Musk (2:20:46.000)
Um, so, so you at least have like a rough lay of the land of the knowledge landscape.
Elon Musk (2:20:54.320)
Like try to learn a little bit about a lot of things, um, cause you might not know what
Lex Fridman (2:20:59.040)
you're really interested in.
Lex Fridman (2:21:00.040)
How would you know what you're really interested in if you at least aren't like doing a peripheral
Lex Fridman (2:21:04.280)
explore exploration of broadly of, of the knowledge landscape?
Elon Musk (2:21:10.240)
Um, and you talk to people from different walks of life and different, uh, industries
Lex Fridman (2:21:17.320)
and professions and skills and occupations, like just try to learn as much as possible.
Elon Musk (2:21:27.520)
Man's search for meaning.
Lex Fridman (2:21:31.120)
Isn't the whole thing a search for meaning?
Elon Musk (2:21:32.920)
Yeah.
Elon Musk (2:21:33.920)
What's the meaning of life and all, you know, but just generally, like I said, I would encourage
Elon Musk (2:21:38.920)
people to read broadly, um, in many different subject areas, um, and, and, and then try
Elon Musk (2:21:45.760)
to find something where there's an overlap of your talents and, and what you're interested
Elon Musk (2:21:51.580)
in.
Lex Fridman (2:21:52.580)
So people may, may, may be good at something, but, or they may have skill at a particular
Elon Musk (2:21:56.120)
thing, but they don't like doing it.
Elon Musk (2:21:57.760)
Um, so you want to try to find a thing where you have your, that's a good, a good, a combination
Elon Musk (2:22:04.680)
of, of your, of the things that you're inherently good at, but you also like doing, um, and,
Lex Fridman (2:22:12.600)
um,
Lex Fridman (2:22:13.600)
And reading is a super fast shortcut to, to figure out which, where are you, you both
Lex Fridman (2:22:18.680)
good at it.
Elon Musk (2:22:19.680)
You like doing it and it will actually have positive impact.
Lex Fridman (2:22:22.680)
Well, you got to learn about things somehow.
Lex Fridman (2:22:25.040)
So read, reading a broad range, just really read, read it.
Elon Musk (2:22:31.320)
You know, one point was that kid I read through the encyclopedia, uh, so that was pretty helpful.
Elon Musk (2:22:38.760)
Um, and, uh, there are also things that I didn't even know existed a lot, so obviously
Lex Fridman (2:22:44.280)
and
Elon Musk (2:22:45.280)
It's like as broad as it gets.
Lex Fridman (2:22:46.280)
Encyclopedias were digestible, I think, uh, you know, whatever, 40 years ago.
Elon Musk (2:22:51.240)
Um, so, um, you know, maybe read through the, the condensed version of the encyclopedia
Lex Fridman (2:22:58.360)
of Britannica.
Lex Fridman (2:22:59.360)
And that, um, you can always like skip subjects or you read a few paragraphs and you know
Lex Fridman (2:23:05.120)
you're not interested, just jump to the next one.
Elon Musk (2:23:07.560)
That sort of read the encyclopedia or scan, skim, skim through it.
Elon Musk (2:23:12.360)
Um, and, um, but I, you know, I put a lot of stock and certainly have a lot of respect
Elon Musk (2:23:19.920)
for someone who puts in an honest day's work, uh, to do useful things and, and just generally
Elon Musk (2:23:27.480)
to have like a, not a zero sum mindset, um, or, uh, like have, have more of a grow the
Elon Musk (2:23:34.560)
pie mindset.
Elon Musk (2:23:35.840)
Like the, if you, if you sort of say like when, when I see people like perhaps, um,
Elon Musk (2:23:42.280)
including some very smart people kind of taking an attitude of, uh, like, like, like doing
Elon Musk (2:23:48.320)
things that seem like morally questionable, it's often because they have at a base sort
Elon Musk (2:23:53.640)
of axiomatic level, a zero sum mindset.
Elon Musk (2:23:57.480)
Um, and, and they, without realizing it, they don't realize they have a zero sum mindset
Elon Musk (2:24:02.720)
or at least that they don't realize it consciously.
Elon Musk (2:24:04.920)
Um, and so if you have a zero sum mindset, then the only way to get ahead is by taking
Elon Musk (2:24:09.260)
things from others.
Elon Musk (2:24:11.520)
If it's like, if the, if the, if the pie is fixed, then the only way to have more pie
Elon Musk (2:24:17.160)
is to take someone else's pie.
Lex Fridman (2:24:19.360)
But this is false.
Elon Musk (2:24:20.680)
Like obviously the pie has grown dramatically over time, the economic pie.
Elon Musk (2:24:24.120)
Um, so the reality, in reality you can have the, so overuse this analogy, if you have
Elon Musk (2:24:31.640)
a lot of, there's a lot of pie, pie is not fixed.
Elon Musk (2:24:37.960)
Um, uh, so you really want to make sure you don't, you're not operating, um, without realizing
Elon Musk (2:24:43.640)
it from a zero sum mindset where, where the only way to get ahead is to take things from
Lex Fridman (2:24:47.960)
others.
Lex Fridman (2:24:48.960)
And you take, try to take things from others, which is not, not good.
Elon Musk (2:24:52.040)
It's much better to work on, uh, adding to the economic pie, maybe, you know, so creating,
Elon Musk (2:25:02.280)
like I said, create, creating more than you consume, uh, doing more than you.
Lex Fridman (2:25:06.200)
Yeah.
Elon Musk (2:25:07.200)
Um, so that's, that's a big deal.
Elon Musk (2:25:08.960)
Um, I think there's like, you know, a fair number of people in, in finance that, uh,
Elon Musk (2:25:15.040)
do have a bit of a zero sum mindset.
Lex Fridman (2:25:16.680)
I mean, it's all walks of life.
Elon Musk (2:25:19.160)
I've seen that one of the, one of the reasons, uh, Rogan inspires me is he celebrates others
Lex Fridman (2:25:25.680)
a lot.
Elon Musk (2:25:26.680)
There's not, not creating a constant competition.
Lex Fridman (2:25:29.400)
Like there's a scarcity of resources.
Lex Fridman (2:25:31.400)
What happens when you celebrate others and you promote others, the ideas of others, it,
Lex Fridman (2:25:36.800)
it, uh, it actually grows that pie.
Elon Musk (2:25:38.960)
I mean, it, every, like the, uh, the resource, the resources become less scarce and that,
Lex Fridman (2:25:45.280)
that applies in a lot of kinds of domains.
Elon Musk (2:25:46.760)
It applies in academia where a lot of people are very, uh, see some funding for academic
Lex Fridman (2:25:51.600)
research is zero sum and it is not.
Elon Musk (2:25:54.120)
If you celebrate each other, if you make, if you get everybody to be excited about AI,
Elon Musk (2:25:58.080)
about physics, above mathematics, I think it, there'll be more and more funding and
Elon Musk (2:26:02.440)
I think everybody wins.
Lex Fridman (2:26:03.640)
Yeah.
Elon Musk (2:26:04.640)
That applies, I think broadly.
Lex Fridman (2:26:05.640)
Uh,
Elon Musk (2:26:06.640)
yeah, yeah, exactly.
Lex Fridman (2:26:08.680)
So last, last question about love and meaning, uh, what is the role of love?
Elon Musk (2:26:15.240)
In the human condition, broadly and more specific to you, how has love, romantic love or otherwise
Lex Fridman (2:26:21.720)
made you a better person, a better human being?
Lex Fridman (2:26:27.360)
Better engineer?
Lex Fridman (2:26:28.360)
Now you're asking really perplexing questions.
Elon Musk (2:26:31.520)
Um, it's hard to give a, I mean, there are many books, poems and songs written about
Lex Fridman (2:26:41.240)
what is love and what is, what exactly, you know, um, you know, what is love, baby don't
Elon Musk (2:26:50.960)
hurt me.
Lex Fridman (2:26:51.960)
Um, that's one of the great ones.
Elon Musk (2:26:53.920)
Yes.
Lex Fridman (2:26:54.920)
Yeah.
Elon Musk (2:26:55.920)
You've, you've earlier quoted Shakespeare, but that that's really up there.
Lex Fridman (2:26:58.120)
Yeah.
Elon Musk (2:26:59.120)
Love is a many splinter thing.
Elon Musk (2:27:02.120)
Uh, I mean there's, um, it's cause we've talked about so many inspiring things like be useful
Elon Musk (2:27:07.960)
in the world, sort of like solve problems, alleviate suffering, but it seems like connection
Elon Musk (2:27:13.720)
between humans is a source, you know, it's a, it's a source of joys, a source of meaning
Lex Fridman (2:27:20.320)
and that, that's what love is, friendship, love.
Elon Musk (2:27:24.160)
I just wonder if you think about that kind of thing where you talk about preserving the
Elon Musk (2:27:29.040)
light of human consciousness and us becoming a multi planetary, multi planetary species.
Elon Musk (2:27:35.240)
I mean, to me at least, um, that, that means like if we're just alone and conscious and
Lex Fridman (2:27:43.240)
intelligent and it doesn't mean nearly as much as if we're with others, right?
Lex Fridman (2:27:48.880)
And there's some magic created when we're together, the, uh, the, the friendship of
Elon Musk (2:27:54.560)
it.
Lex Fridman (2:27:55.560)
And I think the highest form of it is love, which I think broadly is, is much bigger than
Elon Musk (2:27:59.440)
just sort of romantic, but also yes, romantic love and, um, family and those kinds of things.
Elon Musk (2:28:05.800)
Well, I mean, the reason I guess I care about us becoming multi planet species in a space
Elon Musk (2:28:10.420)
frank civilization is foundationally, I love humanity, um, and, and so I wish to see it
Elon Musk (2:28:19.600)
prosper and do great things and be happy and, um, and if I did not love humanity, I would
Elon Musk (2:28:27.520)
not care about these things.
Lex Fridman (2:28:29.640)
So when you look at the whole of it, the human history, all the people who's ever lived,
Elon Musk (2:28:35.080)
all the people live now, it's pretty, we're, we're okay.
Lex Fridman (2:28:41.840)
On the whole, we're pretty interesting bunch.
Elon Musk (2:28:44.960)
Yeah.
Elon Musk (2:28:45.960)
All things considered, and I've read a lot of history, including the darkest, worst parts
Elon Musk (2:28:51.680)
of it, and, uh, despite all that, I think on balance, I still love humanity.
Elon Musk (2:28:59.560)
You joked about it with the 42, uh, what do you, what do you think is the meaning of this
Lex Fridman (2:29:03.280)
whole thing?
Lex Fridman (2:29:05.600)
Is like, is there a non numerical representation?
Elon Musk (2:29:08.520)
Yeah.
Elon Musk (2:29:09.520)
Well, really, I think what Douglas Adams was saying in Hitchhiker's Guide to the Galaxy
Elon Musk (2:29:13.120)
is that, um, the universe is the answer and, uh, what we really need to figure out are
Lex Fridman (2:29:22.000)
what questions to ask about the answer that is the universe and that the question is the
Elon Musk (2:29:27.680)
really the hard part.
Lex Fridman (2:29:28.680)
And if you can properly frame the question, then the answer, relatively speaking, is easy.
Elon Musk (2:29:33.640)
Uh, so, so, so therefore, if you want to understand what questions to ask about the universe,
Elon Musk (2:29:40.840)
you want to understand the meaning of life, we need to expand the scope and scale of consciousness
Lex Fridman (2:29:45.860)
so that we're better able to understand the nature of the universe and, and understand
Lex Fridman (2:29:51.120)
the meaning of life.
Lex Fridman (2:29:52.580)
And ultimately, the most important part would be to ask the right question, thereby elevating
Lex Fridman (2:30:00.240)
the role of the interviewer as the most important human in the room.
Elon Musk (2:30:07.880)
Good questions are, you know, it's a hard, it's hard to come up with good questions.
Lex Fridman (2:30:12.880)
Absolutely.
Elon Musk (2:30:13.880)
Um, but yeah, like, it's like that, that is the foundation of my philosophy is that, um,
Elon Musk (2:30:20.000)
I am curious about the nature of the universe and, uh, you know, and obviously I will die.
Elon Musk (2:30:27.240)
I don't know when I'll die, but I won't live forever.
Elon Musk (2:30:31.840)
Um, but I would like to know that we are on a path to understanding the nature of the
Elon Musk (2:30:36.560)
universe and the meaning of life and what questions to ask about the answer that is
Lex Fridman (2:30:40.040)
the universe.
Elon Musk (2:30:41.040)
And, um, and so if we expand the scope and scale of humanity and consciousness in general,
Elon Musk (2:30:46.160)
um, which includes silicon consciousness, then that, you know, that, that, that seems
Elon Musk (2:30:51.960)
like a fundamentally good thing.
Elon Musk (2:30:53.920)
Elon, like I said, um, I'm deeply grateful that you would spend your extremely valuable
Elon Musk (2:31:00.520)
time with me today and also that you have given millions of people hope in this difficult
Lex Fridman (2:31:07.040)
time, this divisive time, in this, uh, cynical time.
Lex Fridman (2:31:11.380)
So I hope you do continue doing what you're doing.
Lex Fridman (2:31:14.160)
Thank you so much for talking today.
Elon Musk (2:31:15.160)
Oh, you're welcome.
Lex Fridman (2:31:16.160)
Uh, thanks for your excellent questions.
Elon Musk (2:31:18.880)
Thanks for listening to this conversation with Elon Musk.
Lex Fridman (2:31:21.280)
To support this podcast, please check out our sponsors in the description.
Lex Fridman (2:31:25.480)
And now let me leave you with some words from Elon Musk himself.
Elon Musk (2:31:29.640)
When something is important enough, you do it, even if the odds are not in your favor.
Elon Musk (2:31:35.840)
Thank you for listening and hope to see you next time.
Lex Fridman (30:00.480)
some calamity that could be something
Elon Musk (30:05.200)
that humans do to themselves,
Lex Fridman (30:06.720)
or an external event like happened to the dinosaurs.
Lex Fridman (30:12.680)
But eventually, if none of that happens,
Lex Fridman (30:17.920)
and somehow magically we keep going,
Elon Musk (30:21.240)
then the Sun is gradually expanding,
Lex Fridman (30:24.520)
and will engulf the Earth,
Lex Fridman (30:26.600)
and probably Earth gets too hot for life
Lex Fridman (30:31.080)
in about 500 million years.
Elon Musk (30:34.800)
It's a long time, but that's only 10% longer
Lex Fridman (30:37.280)
than Earth has been around.
Lex Fridman (30:39.360)
And so if you think about the current situation,
Lex Fridman (30:43.240)
it's really remarkable, and kind of hard to believe,
Lex Fridman (30:45.640)
but Earth's been around four and a half billion years,
Lex Fridman (30:50.120)
and this is the first time in four and a half billion years
Elon Musk (30:52.360)
that it's been possible to extend life beyond Earth.
Lex Fridman (30:55.720)
And that window of opportunity may be open
Elon Musk (30:58.120)
for a long time, and I hope it is,
Lex Fridman (30:59.520)
but it also may be open for a short time.
Lex Fridman (31:01.740)
And I think it is wise for us to act quickly
Lex Fridman (31:09.440)
while the window is open, just in case it closes.
Elon Musk (31:13.800)
Yeah, the existence of nuclear weapons, pandemics,
Lex Fridman (31:17.800)
all kinds of threats should kind of give us some motivation.
Elon Musk (31:22.800)
I mean, civilization could die with a bang or a whimper.
Lex Fridman (31:29.560)
If it dies as a demographic collapse,
Elon Musk (31:32.760)
then it's more of a whimper, obviously.
Lex Fridman (31:35.160)
And if it's World War III, it's more of a bang.
Lex Fridman (31:38.480)
But these are all risks.
Lex Fridman (31:41.160)
I mean, it's important to think of these things
Lex Fridman (31:42.000)
and just think of things like probabilities, not certainties.
Lex Fridman (31:46.240)
There's a certain probability
Elon Musk (31:47.680)
that something bad will happen on Earth.
Lex Fridman (31:50.200)
I think most likely the future will be good.
Lex Fridman (31:53.200)
But there's like, let's say for argument's sake,
Lex Fridman (31:56.400)
a 1% chance per century of a civilization ending event.
Elon Musk (32:00.920)
Like that was Stephen Hawking's estimate.
Lex Fridman (32:05.080)
I think he might be right about that.
Lex Fridman (32:07.720)
So then we should basically think of this
Lex Fridman (32:15.440)
being a multi planet species,
Elon Musk (32:16.760)
just like taking out a planet from the sky,
Lex Fridman (32:18.600)
multi planet species, just like taking out insurance
Elon Musk (32:20.720)
for life itself.
Lex Fridman (32:21.560)
Like life insurance for life.
Elon Musk (32:27.280)
It's turned into an infomercial real quick.
Lex Fridman (32:29.360)
Life insurance for life, yes.
Lex Fridman (32:31.600)
And we can bring the creatures from,
Lex Fridman (32:36.240)
plants and animals from Earth to Mars
Lex Fridman (32:38.040)
and breathe life into the planet
Lex Fridman (32:41.280)
and have a second planet with life.
Elon Musk (32:44.640)
That would be great.
Lex Fridman (32:46.000)
They can't bring themselves there.
Lex Fridman (32:47.560)
So if we don't bring them to Mars,
Lex Fridman (32:50.000)
then they will just for sure all die
Elon Musk (32:52.280)
when the sun expands anyway.
Lex Fridman (32:54.320)
And then that'll be it.
Lex Fridman (32:56.200)
What do you think is the most difficult aspect
Lex Fridman (32:58.560)
of building a civilization on Mars?
Elon Musk (33:00.960)
Terraforming Mars, like from an engineering perspective,
Lex Fridman (33:03.760)
from a financial perspective, human perspective,
Elon Musk (33:07.240)
to get a large number of folks there
Lex Fridman (33:13.000)
who will never return back to Earth?
Elon Musk (33:15.880)
No, they could certainly return.
Lex Fridman (33:17.160)
Some will return back to Earth.
Elon Musk (33:18.440)
They will choose to stay there
Lex Fridman (33:19.960)
for the rest of their lives.
Elon Musk (33:21.280)
Yeah, many will.
Lex Fridman (33:23.760)
But we need the spaceships back,
Elon Musk (33:28.440)
like the ones that go to Mars.
Lex Fridman (33:29.480)
We need them back.
Lex Fridman (33:30.320)
So you can hop on if you want.
Lex Fridman (33:32.680)
But we can't just not have the spaceships come back.
Elon Musk (33:34.960)
Those things are expensive.
Lex Fridman (33:35.800)
We need them back.
Elon Musk (33:36.640)
I'd like to come back and do another trip.
Lex Fridman (33:38.760)
I mean, do you think about the terraforming aspect,
Lex Fridman (33:40.680)
like actually building?
Lex Fridman (33:41.640)
Are you so focused right now on the spaceships part
Lex Fridman (33:44.560)
that's so critical to get to Mars?
Lex Fridman (33:46.960)
We absolutely, if you can't get there,
Elon Musk (33:48.200)
nothing else matters.
Lex Fridman (33:49.600)
So, and like I said, we can't get there
Elon Musk (33:53.040)
at some extraordinarily high cost.
Lex Fridman (33:54.680)
I mean, the current cost of, let's say,
Elon Musk (33:57.800)
one ton to the surface of Mars
Lex Fridman (34:00.080)
is on the order of a billion dollars.
Elon Musk (34:02.680)
So, because you don't just need the rocket
Lex Fridman (34:04.840)
and the launch and everything,
Elon Musk (34:05.680)
you need like heat shield, you need guidance system,
Lex Fridman (34:09.840)
you need deep space communications,
Elon Musk (34:12.280)
you need some kind of landing system.
Lex Fridman (34:15.000)
So, like rough approximation would be a billion dollars
Elon Musk (34:19.520)
per ton to the surface of Mars right now.
Lex Fridman (34:22.200)
This is obviously way too expensive
Elon Musk (34:26.880)
to create a self sustaining civilization.
Lex Fridman (34:30.680)
So we need to improve that by at least a factor of 1,000.
Lex Fridman (34:36.840)
A million per ton?
Lex Fridman (34:38.440)
Yes, ideally much less than a million ton.
Lex Fridman (34:40.880)
But if it's not, like it's gotta be,
Lex Fridman (34:44.280)
so you have to say like, well,
Lex Fridman (34:45.200)
how much can society afford to spend
Lex Fridman (34:47.960)
or just want to spend on a self sustaining city on Mars?
Elon Musk (34:52.400)
The self sustaining part is important.
Lex Fridman (34:53.780)
Like it's just the key threshold,
Elon Musk (34:57.720)
the great filter will have been passed
Lex Fridman (35:01.360)
when the city on Mars can survive
Elon Musk (35:05.280)
even if the spaceships from Earth stop coming
Lex Fridman (35:07.720)
for any reason, doesn't matter what the reason is,
Lex Fridman (35:09.960)
but if they stop coming for any reason,
Lex Fridman (35:12.120)
will it die out or will it not?
Lex Fridman (35:13.720)
And if there's even one critical ingredient missing,
Lex Fridman (35:16.400)
then it still doesn't count.
Elon Musk (35:18.320)
It's like, you know, if you're on a long sea voyage
Lex Fridman (35:20.440)
and you've got everything except vitamin C,
Elon Musk (35:23.920)
it's only a matter of time, you know, you're gonna die.
Lex Fridman (35:26.280)
So we're gonna get a Mars city
Elon Musk (35:28.880)
to the point where it's self sustaining.
Lex Fridman (35:32.260)
I'm not sure this will really happen in my lifetime,
Lex Fridman (35:33.880)
but I hope to see it at least have a lot of momentum.
Lex Fridman (35:37.320)
And then you could say, okay,
Lex Fridman (35:38.520)
what is the minimum tonnage necessary
Lex Fridman (35:40.380)
to have a self sustaining city?
Lex Fridman (35:44.060)
And there's a lot of uncertainty about this.
Lex Fridman (35:46.480)
You could say like, I don't know,
Elon Musk (35:48.560)
it's probably at least a million tons
Lex Fridman (35:52.040)
cause you have to set up a lot of infrastructure on Mars.
Elon Musk (35:55.440)
Like I said, you can't be missing anything
Lex Fridman (35:58.680)
that in order to be self sustaining,
Elon Musk (36:00.960)
you can't be missing, like you need,
Lex Fridman (36:02.420)
you know, semiconductor, fabs, you need iron ore refineries,
Elon Musk (36:07.120)
like you need lots of things, you know.
Lex Fridman (36:09.360)
So, and Mars is not super hospitable.
Elon Musk (36:13.640)
It's the least inhospitable planet,
Lex Fridman (36:15.680)
but it's definitely a fixer of a planet.
Elon Musk (36:18.120)
Outside of Earth.
Lex Fridman (36:19.360)
Yes.
Elon Musk (36:20.200)
Earth is pretty good.
Lex Fridman (36:21.020)
Earth is like easy, yeah.
Lex Fridman (36:22.280)
And also I should, we should clarify in the solar system.
Lex Fridman (36:25.480)
Yes, in the solar system.
Elon Musk (36:26.600)
There might be nice like vacation spots.
Lex Fridman (36:29.760)
There might be some great planets out there,
Lex Fridman (36:31.160)
but it's hopeless.
Lex Fridman (36:32.760)
Too hard to get there?
Elon Musk (36:33.800)
Yeah, way, way, way, way, way too hard to say the least.
Lex Fridman (36:37.440)
Let me push back on that, not really a push back,
Lex Fridman (36:39.720)
but a quick curve ball of a question.
Lex Fridman (36:42.000)
So you did mention physics as the first starting point.
Elon Musk (36:44.840)
So, general relativity allows for wormholes.
Lex Fridman (36:51.400)
They technically can exist.
Lex Fridman (36:53.080)
Do you think those can ever be leveraged
Lex Fridman (36:55.720)
by humans to travel fast in the speed of light?
Elon Musk (36:59.400)
Well, the wormhole thing is debatable.
Lex Fridman (37:03.600)
The, we currently do not know of any means
Elon Musk (37:08.200)
of going faster than the speed of light.
Lex Fridman (37:11.760)
There is like, there are some ideas about having space.
Elon Musk (37:21.400)
Like so, you can only move at the speed of light
Lex Fridman (37:26.520)
through space, but if you can make space itself move,
Elon Musk (37:30.740)
that's like, that's warping space.
Lex Fridman (37:36.140)
Space is capable of moving faster than the speed of light.
Elon Musk (37:39.580)
Right.
Lex Fridman (37:40.740)
Like the universe in the Big Bang,
Elon Musk (37:42.100)
the universe expanded at much,
Lex Fridman (37:44.180)
much more than the speed of light by a lot.
Elon Musk (37:46.820)
Yeah.
Lex Fridman (37:48.820)
So, but the, if this is possible,
Elon Musk (37:56.380)
the amount of energy required to warp space
Lex Fridman (37:58.660)
is so gigantic, it boggles the mind.
Lex Fridman (38:03.100)
So all the work you've done with propulsion,
Lex Fridman (38:05.060)
how much innovation is possible with rocket propulsion?
Elon Musk (38:08.100)
Is this, I mean, you've seen it all,
Lex Fridman (38:11.140)
and you're constantly innovating in every aspect.
Lex Fridman (38:14.420)
How much is possible?
Lex Fridman (38:15.340)
Like how much, can you get 10x somehow?
Elon Musk (38:17.380)
Is there something in there in physics
Lex Fridman (38:19.660)
that you can get significant improvement
Elon Musk (38:21.260)
in terms of efficiency of engines
Lex Fridman (38:22.620)
and all those kinds of things?
Elon Musk (38:24.640)
Well, as I was saying, really the Holy Grail
Lex Fridman (38:27.940)
is a fully and rapidly reusable orbital system.
Lex Fridman (38:33.020)
So right now, the Falcon 9
Lex Fridman (38:38.100)
is the only reusable rocket out there,
Lex Fridman (38:41.460)
but the booster comes back and lands,
Lex Fridman (38:44.300)
and you've seen the videos,
Lex Fridman (38:46.180)
and we get the nose cone fairing back,
Lex Fridman (38:47.620)
but we do not get the upper stage back.
Lex Fridman (38:49.740)
So that means that we have a minimum cost
Lex Fridman (38:54.260)
of building an upper stage.
Lex Fridman (38:56.660)
And you can think of like a two stage rocket
Lex Fridman (38:59.300)
of sort of like two airplanes,
Elon Musk (39:00.700)
like a big airplane and a small airplane.
Lex Fridman (39:03.180)
And we get the big airplane back,
Lex Fridman (39:04.600)
but not the small airplane.
Lex Fridman (39:05.860)
And so it still costs a lot.
Lex Fridman (39:07.860)
So that upper stage is at least $10 million.
Lex Fridman (39:13.300)
And then the degree of,
Elon Musk (39:15.460)
the booster is not as rapidly and completely reusable
Lex Fridman (39:19.060)
as we'd like in order of the fairings.
Lex Fridman (39:20.900)
So our kind of minimum marginal cost
Lex Fridman (39:25.180)
and our counting overhead for per flight
Elon Musk (39:27.180)
is on the order of 15 to $20 million, maybe.
Lex Fridman (39:33.400)
So that's extremely good for,
Elon Musk (39:38.060)
it's by far better than any rocket ever in history.
Lex Fridman (39:41.660)
But with full and rapid reusability,
Elon Musk (39:45.380)
we can reduce the cost per ton to orbit
Lex Fridman (39:48.840)
by a factor of 100.
Lex Fridman (39:51.880)
But just think of it like,
Lex Fridman (39:54.200)
like imagine if you had an aircraft or something or a car.
Lex Fridman (39:58.800)
And if you had to buy a new car
Lex Fridman (40:02.920)
every time you went for a drive,
Elon Musk (40:05.080)
that would be very expensive.
Lex Fridman (40:06.760)
It'd be silly, frankly.
Lex Fridman (40:08.320)
But in fact, you just refuel the car or recharge the car.
Lex Fridman (40:13.480)
And that makes your trip like,
Elon Musk (40:18.480)
I don't know, a thousand times cheaper.
Lex Fridman (40:20.320)
So it's the same for rockets.
Elon Musk (40:23.800)
If you, it's very difficult to make this complex machine
Lex Fridman (40:27.240)
that can go to orbit.
Lex Fridman (40:28.680)
And so if you cannot reuse it
Lex Fridman (40:30.560)
and have to throw even any part of,
Elon Musk (40:32.800)
any significant part of it away,
Lex Fridman (40:34.000)
that massively increases the cost.
Elon Musk (40:36.640)
So, you know, Starship in theory
Lex Fridman (40:40.320)
could do a cost per launch of like a million,
Elon Musk (40:44.360)
maybe $2 million or something like that.
Lex Fridman (40:46.400)
And put over a hundred tons in orbit, which is crazy.
Elon Musk (40:53.560)
Yeah, that's incredible.
Lex Fridman (40:55.960)
So you're saying like it's by far the biggest bang
Elon Musk (40:58.200)
for the buck is to make it fully reusable
Lex Fridman (41:00.720)
versus like some kind of brilliant breakthrough
Elon Musk (41:04.160)
in theoretical physics.
Lex Fridman (41:05.760)
No, no, there's no, there's no brilliant break.
Elon Musk (41:07.680)
No, there's no, just make the rocket reusable.
Lex Fridman (41:11.240)
This is an extremely difficult engineering problem.
Elon Musk (41:13.520)
Got it.
Lex Fridman (41:14.360)
No new physics is required.
Elon Musk (41:17.960)
Just brilliant engineering.
Lex Fridman (41:19.320)
Let me ask a slightly philosophical fun question.
Elon Musk (41:22.200)
Gotta ask, I know you're focused on getting to Mars,
Lex Fridman (41:24.840)
but once we're there on Mars, what do you,
Lex Fridman (41:27.240)
what form of government, economic system, political system
Lex Fridman (41:32.360)
do you think would work best
Lex Fridman (41:33.760)
for an early civilization of humans?
Lex Fridman (41:37.200)
Is, I mean, the interesting reason to talk about this stuff,
Elon Musk (41:41.560)
it also helps people dream about the future.
Lex Fridman (41:44.320)
I know you're really focused
Elon Musk (41:45.560)
about the short term engineering dream,
Lex Fridman (41:48.040)
but it's like, I don't know,
Elon Musk (41:49.320)
there's something about imagining an actual civilization
Lex Fridman (41:51.840)
on Mars that gives people, really gives people hope.
Elon Musk (41:55.440)
Well, it would be a new frontier and an opportunity
Lex Fridman (41:57.680)
to rethink the whole nature of government,
Elon Musk (41:59.640)
just as was done in the creation of the United States.
Lex Fridman (42:02.680)
So, I mean, I would suggest having a direct democracy,
Elon Musk (42:14.680)
people vote directly on things
Lex Fridman (42:16.200)
as opposed to representative democracy.
Lex Fridman (42:18.440)
So representative democracy, I think,
Lex Fridman (42:21.520)
is too subject to a special interest
Lex Fridman (42:25.120)
and a coercion of the politicians and that kind of thing.
Lex Fridman (42:29.580)
So I'd recommend that there was just direct democracy,
Elon Musk (42:39.360)
people vote on laws, the population votes on laws themselves,
Lex Fridman (42:42.680)
and then the laws must be short enough
Elon Musk (42:44.520)
that people can understand them.
Lex Fridman (42:46.880)
Yeah, and then like keeping a well informed populace,
Elon Musk (42:49.920)
like really being transparent about all the information,
Lex Fridman (42:52.240)
about what they're voting for.
Elon Musk (42:53.080)
Yeah, absolute transparency.
Lex Fridman (42:54.800)
Yeah, and not make it as annoying as those cookies
Elon Musk (42:57.480)
we have to accept. Accept cookies.
Lex Fridman (42:59.920)
I've always, like, you know,
Elon Musk (43:01.760)
there's like always like a slight amount of trepidation
Lex Fridman (43:03.640)
when you click accept cookies,
Elon Musk (43:05.360)
like I feel as though there's like perhaps
Lex Fridman (43:07.280)
like a very tiny chance that it'll open a portal to hell
Elon Musk (43:10.680)
or something like that.
Lex Fridman (43:12.320)
That's exactly how I feel.
Lex Fridman (43:13.800)
Why do they, why do they keep wanting me to accept it?
Lex Fridman (43:16.640)
What do they want with this cookie?
Elon Musk (43:19.000)
Like somebody got upset with accepting cookies
Lex Fridman (43:21.020)
or something somewhere, who cares?
Elon Musk (43:23.040)
Like so annoying to keep accepting all these cookies.
Lex Fridman (43:26.920)
To me, this is just a great example.
Elon Musk (43:29.600)
Yes, you can have my damn cookie.
Lex Fridman (43:30.720)
I don't care, whatever.
Elon Musk (43:32.360)
Heard it from Ilhan first.
Lex Fridman (43:33.720)
He accepts all your damn cookies.
Elon Musk (43:35.960)
Yeah, and stop asking me.
Lex Fridman (43:40.240)
It's annoying.
Elon Musk (43:41.400)
Yeah, it's one example of implementation
Lex Fridman (43:46.000)
of a good idea done really horribly.
Elon Musk (43:50.200)
Yeah, it's somebody who was like,
Lex Fridman (43:51.480)
there's some good intentions of like privacy or whatever,
Lex Fridman (43:54.920)
but now everyone's just has to accept cookies
Lex Fridman (43:57.320)
and it's not, you know, you have billions of people
Elon Musk (43:59.360)
who have to keep clicking accept cookie.
Lex Fridman (44:00.720)
It's super annoying.
Elon Musk (44:02.440)
Then we just accept the damn cookie, it's fine.
Lex Fridman (44:05.000)
There is like, I think a fundamental problem that we're,
Elon Musk (44:08.660)
because we've not really had a major,
Lex Fridman (44:12.320)
like a world war or something like that in a while.
Lex Fridman (44:14.360)
And obviously we would like to not have world wars.
Lex Fridman (44:18.320)
There's not been a cleansing function
Elon Musk (44:19.840)
for rules and regulations.
Lex Fridman (44:21.420)
So wars did have, you know, some sort of aligning
Elon Musk (44:24.340)
in that there would be a reset on rules
Lex Fridman (44:27.580)
and regulations after a war.
Lex Fridman (44:29.660)
So World Wars I and II,
Lex Fridman (44:30.500)
there were huge resets on rules and regulations.
Elon Musk (44:34.260)
Now, if society does not have a war
Lex Fridman (44:37.900)
and there's no cleansing function
Elon Musk (44:39.460)
or garbage collection for rules and regulations,
Lex Fridman (44:41.420)
then rules and regulations will accumulate every year
Elon Musk (44:43.500)
because they're immortal.
Lex Fridman (44:45.020)
There's no actual, humans die, but the laws don't.
Lex Fridman (44:48.580)
So we need a garbage collection function
Lex Fridman (44:51.660)
for rules and regulations.
Elon Musk (44:52.980)
They should not just be immortal
Lex Fridman (44:55.660)
because some of the rules and regulations
Elon Musk (44:57.340)
that are put in place will be counterproductive,
Lex Fridman (45:00.260)
done with good intentions, but counterproductive.
Elon Musk (45:02.140)
Sometimes not done with good intentions.
Lex Fridman (45:03.940)
So if rules and regulations just accumulate every year
Lex Fridman (45:09.100)
and you get more and more of them,
Lex Fridman (45:10.740)
then eventually you won't be able to do anything.
Elon Musk (45:13.220)
You're just like Gulliver with, you know,
Lex Fridman (45:15.180)
tied down by thousands of little structures
Elon Musk (45:17.420)
by thousands of little strings.
Lex Fridman (45:19.620)
And we see that in, you know, US and like basically
Elon Musk (45:26.460)
all economies that have been around for a while
Lex Fridman (45:31.460)
and regulators and legislators create new rules
Lex Fridman (45:35.340)
and regulations every year,
Lex Fridman (45:36.580)
but they don't put effort into removing them.
Lex Fridman (45:38.740)
And I think that's very important that we put effort
Lex Fridman (45:40.300)
into removing rules and regulations.
Lex Fridman (45:44.020)
But it gets tough because you get special interests
Lex Fridman (45:45.620)
that then are dependent on, like they have, you know,
Elon Musk (45:48.900)
a vested interest in that whatever rule and regulation
Lex Fridman (45:51.940)
and then they fight to not get it removed.
Elon Musk (45:57.620)
Yeah, so I mean, I guess the problem with the constitution
Lex Fridman (46:00.940)
is it's kind of like C versus Java
Elon Musk (46:04.140)
because it doesn't have any garbage collection built in.
Lex Fridman (46:06.740)
I think there should be, when you first said
Elon Musk (46:09.020)
the metaphor of garbage collection, I loved it.
Lex Fridman (46:10.860)
Yeah, from a coding standpoint.
Elon Musk (46:12.380)
From a coding standpoint, yeah, yeah.
Lex Fridman (46:14.340)
It would be interesting if the laws themselves
Elon Musk (46:16.900)
kind of had a built in thing
Lex Fridman (46:19.020)
where they kind of die after a while
Elon Musk (46:20.700)
unless somebody explicitly publicly defends them.
Lex Fridman (46:23.660)
So that's sort of, it's not like somebody has to kill them.
Elon Musk (46:26.220)
They kind of die themselves.
Lex Fridman (46:28.100)
They disappear.
Elon Musk (46:29.180)
Yeah.
Lex Fridman (46:32.540)
Not to defend Java or anything, but you know, C++,
Elon Musk (46:36.220)
you know, you could also have a great garbage collection
Lex Fridman (46:38.540)
in Python and so on.
Elon Musk (46:39.940)
Yeah, so yeah, something needs to happen
Lex Fridman (46:43.740)
or just the civilization's arteries just harden over time
Lex Fridman (46:48.580)
and you can just get less and less done
Lex Fridman (46:50.820)
because there's just a rule against everything.
Lex Fridman (46:54.820)
So I think like, I don't know, for Mars or whatever,
Lex Fridman (46:57.900)
I'd say, or even for, you know, obviously for Earth as well,
Elon Musk (47:00.260)
like I think there should be an active process
Lex Fridman (47:02.620)
for removing rules and regulations
Lex Fridman (47:04.940)
and questioning their existence.
Lex Fridman (47:07.100)
Just like if we've got a function
Elon Musk (47:10.260)
for creating rules and regulations,
Lex Fridman (47:11.580)
because rules and regulations can also think of us
Elon Musk (47:13.140)
like they're like software or lines of code
Lex Fridman (47:15.780)
for operating civilization.
Elon Musk (47:18.860)
That's the rules and regulations.
Lex Fridman (47:21.300)
So it's not like we shouldn't have rules and regulations,
Lex Fridman (47:22.980)
but you have code accumulation, but no code removal.
Lex Fridman (47:27.100)
And so it just gets to become basically archaic bloatware
Elon Musk (47:31.460)
after a while.
Lex Fridman (47:33.620)
And it's just, it makes it hard for things to progress.
Lex Fridman (47:37.900)
So I don't know, maybe Mars, you'd have like,
Lex Fridman (47:40.460)
you know, any given law must have a sunset, you know,
Lex Fridman (47:44.300)
and require active voting to keep it up there, you know.
Lex Fridman (47:52.140)
And I actually also say like, and these are just,
Elon Musk (47:54.660)
I don't know, recommendations or thoughts,
Lex Fridman (47:58.260)
ultimately will be up to the people on Mars to decide.
Lex Fridman (48:00.700)
But I think it should be easier to remove a law
Lex Fridman (48:06.660)
than to add one because of the,
Elon Musk (48:08.620)
just to overcome the inertia of laws.
Lex Fridman (48:10.700)
So maybe it's like, for argument's sake,
Elon Musk (48:15.220)
you need like say 60% vote to have a law take effect,
Lex Fridman (48:19.780)
but only a 40% vote to remove it.
Lex Fridman (48:23.420)
So let me be the guy, you posted a meme on Twitter recently
Lex Fridman (48:26.620)
where there's like a row of urinals,
Lex Fridman (48:30.140)
and a guy just walks all the way across,
Lex Fridman (48:33.100)
and he tells you about crypto.
Elon Musk (48:36.300)
I mean, that's happened to me so many times.
Lex Fridman (48:38.340)
I think maybe even literally.
Elon Musk (48:40.420)
Yeah.
Lex Fridman (48:41.820)
Do you think, technologically speaking,
Lex Fridman (48:43.500)
there's any room for ideas of smart contracts or so on?
Lex Fridman (48:47.340)
Because you mentioned laws.
Elon Musk (48:49.300)
That's an interesting use of things like smart contracts
Lex Fridman (48:52.980)
to implement the laws by which governments function.
Elon Musk (48:57.300)
Like something built on Ethereum,
Lex Fridman (48:58.980)
or maybe a dog coin that enables smart contracts somehow.
Elon Musk (49:04.860)
I don't quite understand this whole smart contract thing.
Lex Fridman (49:08.260)
You know.
Elon Musk (49:09.960)
I mean, I'm too dumb to understand smart contracts.
Lex Fridman (49:15.020)
That's a good line.
Elon Musk (49:17.900)
I mean, my general approach to any kind of deal or whatever
Lex Fridman (49:21.460)
is just make sure there's clarity of understanding.
Elon Musk (49:23.780)
That's the most important thing.
Lex Fridman (49:25.820)
And just keep any kind of deal very short and simple,
Elon Musk (49:29.660)
plain language, and just make sure everyone understands
Lex Fridman (49:33.460)
this is the deal, is it clear?
Lex Fridman (49:36.580)
And what are the consequences if various things
Lex Fridman (49:40.660)
don't happen?
Lex Fridman (49:42.620)
But usually deals are, business deals or whatever,
Lex Fridman (49:47.220)
are way too long and complex and overly lawyered
Lex Fridman (49:50.940)
and pointlessly.
Lex Fridman (49:52.740)
You mentioned that Doge is the people's coin.
Elon Musk (49:57.020)
Yeah.
Lex Fridman (49:57.860)
And you said that you were literally going,
Elon Musk (49:59.660)
SpaceX may consider literally putting a Doge coin
Lex Fridman (50:04.660)
on the moon, is this something you're still considering?
Elon Musk (50:09.940)
Mars, perhaps, do you think there's some chance,
Lex Fridman (50:13.620)
we've talked about political systems on Mars,
Elon Musk (50:16.060)
that Doge coin is the official currency of Mars
Lex Fridman (50:20.220)
at some point in the future?
Elon Musk (50:22.580)
Well, I think Mars itself will need to have
Lex Fridman (50:25.700)
a different currency because you can't synchronize
Elon Musk (50:29.060)
due to speed of light, or not easily.
Lex Fridman (50:32.660)
So it must be completely stand alone from Earth.
Elon Musk (50:36.480)
Well, yeah, because Mars is, at closest approach,
Lex Fridman (50:41.340)
it's four light minutes away, roughly,
Lex Fridman (50:43.020)
and then at furthest approach, it's roughly
Lex Fridman (50:45.580)
20 light minutes away, maybe a little more.
Lex Fridman (50:50.100)
So you can't really have something synchronizing
Lex Fridman (50:52.980)
if you've got a 20 minute speed of light issue,
Elon Musk (50:55.500)
if it's got a one minute blockchain.
Lex Fridman (50:58.180)
It's not gonna synchronize properly.
Lex Fridman (50:59.980)
So Mars, I don't know if Mars would have
Lex Fridman (51:03.960)
a cryptocurrency as a thing, but probably, seems likely.
Lex Fridman (51:07.640)
But it would be some kind of localized thing on Mars.
Lex Fridman (51:12.280)
And you let the people decide.
Elon Musk (51:14.800)
Yeah, absolutely.
Lex Fridman (51:17.680)
The future of Mars should be up to the Martians.
Elon Musk (51:20.720)
Yeah, so, I think the cryptocurrency thing
Lex Fridman (51:25.720)
is an interesting approach to reducing
Elon Musk (51:30.800)
the error in the database that is called money.
Lex Fridman (51:41.400)
I think I have a pretty deep understanding
Elon Musk (51:42.920)
of what money actually is on a practical day to day basis
Lex Fridman (51:46.720)
because of PayPal.
Elon Musk (51:50.400)
We really got in deep there.
Lex Fridman (51:52.920)
And right now, the money system, actually,
Elon Musk (51:57.560)
for practical purposes, is really a bunch
Lex Fridman (52:01.120)
of heterogeneous mainframes running old COBOL.
Elon Musk (52:07.480)
Okay, you mean literally.
Lex Fridman (52:08.720)
That's literally what's happening.
Elon Musk (52:10.840)
In batch mode.
Lex Fridman (52:12.760)
Okay.
Elon Musk (52:13.600)
In batch mode.
Lex Fridman (52:14.420)
Yeah, pretty the poor bastards who have
Elon Musk (52:16.720)
to maintain that code.
Lex Fridman (52:19.000)
Okay, that's a pain.
Elon Musk (52:22.240)
Not even Fortran, it's COBOL.
Lex Fridman (52:24.240)
It's COBOL.
Lex Fridman (52:26.040)
And the banks are still buying mainframes in 2021
Lex Fridman (52:30.120)
and running ancient COBOL code.
Lex Fridman (52:33.000)
And the Federal Reserve is probably even older
Lex Fridman (52:37.800)
than what the banks have, and they have
Elon Musk (52:39.280)
an old COBOL mainframe.
Lex Fridman (52:41.880)
And so, the government effectively has editing privileges
Elon Musk (52:47.080)
on the money database.
Lex Fridman (52:48.660)
And they use those editing privileges to make more money,
Elon Musk (52:53.740)
whatever they want.
Lex Fridman (52:55.420)
And this increases the error in the database that is money.
Elon Musk (52:59.060)
So, I think money should really be viewed
Lex Fridman (53:00.820)
through the lens of information theory.
Lex Fridman (53:03.580)
And so, it's kind of like an internet connection.
Lex Fridman (53:08.300)
Like what's the bandwidth, total bit rate,
Lex Fridman (53:12.580)
what is the latency, jitter, packet drop,
Lex Fridman (53:16.420)
you know, errors in network communication.
Elon Musk (53:21.820)
Just think of money like that, basically.
Lex Fridman (53:24.460)
I think that's probably the right way to think of it.
Lex Fridman (53:26.460)
And then say what system from an information theory
Lex Fridman (53:31.460)
standpoint allows an economy to function the best.
Elon Musk (53:35.620)
And, you know, crypto is an attempt to reduce
Lex Fridman (53:40.620)
the error in money that is contributed
Elon Musk (53:48.780)
by governments diluting the money supply
Lex Fridman (53:53.300)
as basically a pernicious form of taxation.
Elon Musk (53:58.900)
So, both policy in terms of with inflation
Lex Fridman (54:01.900)
and actual like technological COBOL,
Elon Musk (54:05.780)
like cryptocurrency takes us into the 21st century
Lex Fridman (54:08.880)
in terms of the actual systems
Elon Musk (54:10.740)
that allow you to do the transaction,
Lex Fridman (54:12.140)
to store wealth, all those kinds of things.
Elon Musk (54:16.920)
Like I said, just think of money as information.
Lex Fridman (54:18.580)
People often will think of money
Elon Musk (54:20.900)
as having power in and of itself.
Lex Fridman (54:24.100)
It does not.
Elon Musk (54:24.980)
Money is information and it does not have power
Lex Fridman (54:28.980)
in and of itself.
Elon Musk (54:31.460)
Like, you know, applying the physics tools
Lex Fridman (54:35.060)
of thinking about things in the limit is helpful.
Elon Musk (54:37.540)
If you are stranded on a tropical island
Lex Fridman (54:41.020)
and you have a trillion dollars, it's useless
Elon Musk (54:47.660)
because there's no resource allocation.
Lex Fridman (54:50.540)
Money is a database for resource allocation,
Lex Fridman (54:52.660)
but there's no resource to allocate except yourself.
Lex Fridman (54:55.020)
So, money is useless.
Elon Musk (55:01.020)
If you're stranded on a desert island with no food,
Lex Fridman (55:04.100)
all the Bitcoin in the world will not stop you
Elon Musk (55:10.340)
from starving.
Lex Fridman (55:12.420)
So, just think of money as a database
Elon Musk (55:20.820)
for resource allocation across time and space.
Lex Fridman (55:24.980)
And then what system, in what form
Elon Musk (55:29.980)
should that database or data system,
Lex Fridman (55:37.020)
what would be most effective?
Elon Musk (55:39.060)
Now, there is a fundamental issue
Lex Fridman (55:41.380)
with say Bitcoin in its current form
Elon Musk (55:46.020)
in that the transaction volume is very limited
Lex Fridman (55:50.700)
and the latency for a properly confirmed transaction
Elon Musk (55:55.700)
is too long, much longer than you'd like.
Lex Fridman (55:58.100)
So, it's actually not great from a transaction volume
Elon Musk (56:02.460)
standpoint or a latency standpoint.
Lex Fridman (56:07.260)
So, it is perhaps useful to solve an aspect
Elon Musk (56:12.420)
of the money database problem, which is the sort of store
Lex Fridman (56:17.420)
of wealth or an accounting of relative obligations,
Elon Musk (56:22.020)
I suppose, but it is not useful as a currency,
Lex Fridman (56:27.500)
as a day to day currency.
Lex Fridman (56:28.780)
But people have proposed different technological solutions.
Lex Fridman (56:31.260)
Like Lightning.
Elon Musk (56:32.100)
Yeah, Lightning Network and the layer two technologies
Lex Fridman (56:34.740)
on top of that.
Elon Musk (56:35.580)
I mean, it seems to be all kind of a trade off,
Lex Fridman (56:38.820)
but the point is, it's kind of brilliant to say
Elon Musk (56:41.060)
that just think about information,
Lex Fridman (56:42.460)
think about what kind of database,
Lex Fridman (56:44.100)
what kind of infrastructure enables
Lex Fridman (56:45.860)
that exchange of information.
Elon Musk (56:46.700)
Yeah, just say like you're operating an economy
Lex Fridman (56:49.180)
and you need to have some thing that allows
Elon Musk (56:55.260)
for the efficient, to have efficient value ratios
Lex Fridman (56:59.740)
between products and services.
Elon Musk (57:01.380)
So, you've got this massive number of products
Lex Fridman (57:03.220)
and services and you need to, you can't just barter.
Elon Musk (57:06.380)
It's just like, that would be extremely unwieldy.
Lex Fridman (57:09.620)
So, you need something that gives you the ratio
Elon Musk (57:13.420)
of exchange between goods and services.
Lex Fridman (57:20.540)
And then something that allows you
Elon Musk (57:22.580)
to shift obligations across time, like debt.
Lex Fridman (57:26.580)
Debt and equity shift obligations across time.
Lex Fridman (57:29.180)
Then what does the best job of that?
Lex Fridman (57:33.300)
Part of the reason why I think there's some
Elon Musk (57:36.020)
Merit Doge coin, even though it was obviously created
Lex Fridman (57:38.660)
as a joke, is that it actually does have
Elon Musk (57:44.100)
a much higher transaction volume capability than Bitcoin.
Lex Fridman (57:49.820)
And the costs of doing a transaction,
Elon Musk (57:53.180)
the Doge coin fee is very low.
Lex Fridman (57:55.980)
Like right now, if you want to do a Bitcoin transaction,
Elon Musk (57:58.220)
the price of doing that transaction is very high.
Lex Fridman (58:00.460)
So, you could not use it effectively for most things.
Lex Fridman (58:04.340)
And nor could it even scale to a high volume.
Lex Fridman (58:11.860)
And when Bitcoin started, I guess around 2008
Elon Musk (58:15.260)
or something like that, the internet connections
Lex Fridman (58:18.740)
were much worse than they are today.
Elon Musk (58:20.820)
Like Order of magnitude, I mean,
Lex Fridman (58:23.620)
just way, way worse in 2008.
Elon Musk (58:26.860)
So, like having a small block size or whatever
Lex Fridman (58:31.860)
is, and a long synchronization time made sense in 2008.
Lex Fridman (58:37.820)
But 2021, or fast forward 10 years,
Lex Fridman (58:41.380)
it's like comically low.
Lex Fridman (58:50.540)
And I think there's some value to having a linear increase
Lex Fridman (58:55.580)
in the amount of currency that is generated.
Elon Musk (58:58.620)
So, because some amount of the currency,
Lex Fridman (59:01.060)
like if a currency is too deflationary,
Elon Musk (59:05.060)
or I should say, if a currency is expected
Lex Fridman (59:10.140)
to increase in value over time,
Elon Musk (59:11.700)
there's reluctance to spend it.
Lex Fridman (59:14.220)
Because you're like, oh, I'll just hold it and not spend it
Elon Musk (59:17.380)
because it's scarcity is increasing with time.
Lex Fridman (59:19.380)
So, if I spend it now, then I will regret spending it.
Elon Musk (59:22.180)
So, I will just, you know, hodl it.
Lex Fridman (59:24.980)
But if there's some dilution of the currency occurring
Elon Musk (59:27.980)
over time, that's more of an incentive
Lex Fridman (59:29.860)
to use it as a currency.
Elon Musk (59:31.260)
So, those coins, somewhat randomly has just a fixed,
Lex Fridman (59:39.220)
a number of sort of coins or hash strings
Elon Musk (59:43.220)
that are generated every year.
Lex Fridman (59:46.540)
So, there's some inflation, but it's not a percentage base.
Elon Musk (59:49.860)
It's a percentage of the total amount of money
Lex Fridman (59:54.340)
it's a fixed number.
Elon Musk (59:55.700)
So, the percentage of inflation
Lex Fridman (59:58.380)
will necessarily decline over time.
🔗 相关节目