YT YouTube2Text

YouTube2Text → Transcripts → Elon Musk: The End of Work, the Death of Money, and a World Run by AI

Elon Musk: The End of Work, the Death of Money, and a World Run by AI — Transcript

by The Builders · 6,462 words · 1,111 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00I've been sort of somewhat conflicted
0:01about
0:02AI. I don't want to hasten
0:04our doom if because there's always some
0:06chance that AI could could be negative.
0:08But then, you know, when I've tried to
0:10take actions in this regard, it seems
0:11[music] to have only accelerated AI
0:14because reality is not just an
0:16interestingness max- maximizer, but also
0:18an irony maximizer. [music]
0:19>> And so, by starting OpenAI, you ended up
0:22maximizing by leading to Anthropic.
0:24>> OpenAI ended up accelerating AI, and
0:26then Anthropic, which is really a you
0:28know, a splinter [music] group from
0:29OpenAI,
0:31um is accelerating it even more.
0:39>> I'm Walter Isaacson, [music]
0:41and this is The Builders. It's a series
0:43of conversations with the most
0:45consequential minds shaping our future.
0:48Now, I want to start
0:50with Elon Musk. Elon has lived a lot of
0:53lives since I last wrote about him. He
0:56went into government with Doge. He
0:58became a trillionaire with the SpaceX
1:00[music] IPO. He bought Cursor AI, a
1:03company that turns
1:04>> [music]
1:04>> individual coders into entire
1:07engineering teams. And he's shifting his
1:09solar energy strategy from panels on
1:12Earth
1:13to harnessing it from systems in space.
1:16He's a relentless builder with
1:19incredible power, and his inventions and
1:21companies are destined to reverberate
1:24across humanity. So, let's ask him about
1:27the future he believes that he's
1:30building.
1:33Elon Musk.
1:34>> Hey Walter.
1:35>> Welcome to The Builders.
1:36>> Uh thanks.
1:38>> I've uh watched you, I think, uh be
1:41maybe mercurial or waver in how sanguine
1:44you are about artificial intelligence.
1:47Sometimes you feel it's an existential
1:50threat, right? Now you seem to be going
1:52with the flow. Tell me how you're
1:54feeling about it.
1:55>> Well, I I do think it's a at least some
1:57percentage chance of it being an
1:58existential threat.
2:00So,
2:02but I There's nothing I can do to stop
2:04it.
2:05So, I
2:06basically came to conclusion I've got
2:07two choices, which is
2:09either
2:10be a participant or a spectator.
2:13And I guess being a participant is more
2:15interesting than being a spectator. And
2:17maybe I can shape it hopefully in a good
2:19direction, maybe.
2:20>> And if you were to shape it, how? Make
2:22it maximally truth-seeking?
2:24>> Yeah, yeah, yeah, yeah. That's certainly
2:25an important
2:26foundation.
2:27>> And make it curious?
2:29>> Yeah, curious, truth-seeking.
2:32I think I think these are important
2:33elements of safety.
2:35You know, whereas if you
2:38you know, try to force an an AI to
2:41believe two things that are cannot
2:43mutually be true. Like they can't they
2:45can't both be true, and you but you tell
2:46AI
2:48A is true, and B is true, even though A
2:50and B contradict each other.
2:52And if that happens at a foundational
2:54level, you can make the AI insane.
2:59In fact, I think that was at least part
3:01of the lesson that Odyssey Clark was
3:02trying to convey in 2001 Space Odyssey.
3:06You know, the ship AI was told to to to
3:08take the astronauts to the monolith,
3:10but also that they could not know about
3:12the monolith.
3:13It was secret.
3:16And so, it came to the conclusion that
3:17it must take them there dead.
3:21And thus
3:23>> [laughter]
3:23>> So, it basically The force
3:26I think I think part of what Odyssey
3:27Clark was trying to say is like, don't
3:28force don't force an AI to lie.
3:31>> And if you make it maximally
3:33truth-seeking and curious, like you say,
3:36I I don't quite get it. Why does that
3:38make it safer?
3:40>> Well, if you force an AI to believe two
3:41mutually consistent things, you can you
3:43can make it go insane.
3:46Um
3:47And and an insane AI I think is not a
3:49safe one.
3:50Then, curious I think is important
3:52because um a curious AI, I think, is
3:55going to want to foster humanity
3:57to see how we turn out. So, like, what
3:59happens to uh to a humanity that is uh
4:04that
4:04that that prospers and is becomes
4:07multi-planetary.
4:08I think a curious AI will want to foster
4:10humanity. You know, a future where there
4:12are humans on multiple planets is more
4:15interesting than one where where there
4:17are not any humans on the planets.
4:19>> So, the most interesting outcome is the
4:21most likely?
4:22>> Yes.
4:23Well, generally, I think that's true for
4:25a different reason, which is that if
4:27simulation theory is correct, then any
4:30simulations that are boring will be
4:31truncated.
4:33>> And is simulation theory correct?
4:35>> that we exist
4:36means that we are an interesting
4:38simulation. And really, the most The
4:39most important thing is just don't be
4:41boring.
4:42>> This has explained you a bit. You're on
4:44a crusade against being boring. Well,
4:46I've heard you talk about this a bit. I
4:48I really do need to know.
4:50How convinced are you that we might be
4:52in a simulation?
4:55>> You know, I once had a dinner
4:58at my house with a physicist and a
5:00computer scientist, and they're both
5:02like leaders in their field,
5:03and I asked the question, "What do you
5:05think the probability is that we're in a
5:06simulation?" And the physicist said 0%
5:09and the computer scientist said 100%
5:11simultaneously.
5:13So, I guess
5:14maybe I should go 50% then.
5:17I don't know. It's It's I mean, so there
5:20there there's a lot of things that
5:22happen in the simulation where
5:24especially if you're me, I mean, think
5:25about it. Like, if you're me and you're
5:27like,
5:28"Should Should all these things really
5:29be happening? Should I really be making
5:32spaceships to go to the moon and Mars?
5:35Like, what the That doesn't seem real."
5:37>> And what happens when you have all the
5:41intelligence you could possibly want in
5:44a society? Will that intelligence, the
5:47artificial intelligence, be in charge?
5:49Will we sort of become subservient to
5:52our new machines?
5:54>> Um
5:56I don't know about subservient
5:58uh exactly, but will we be able to boss
6:01around something that's
6:04you know,
6:05thousands or millions of times smarter
6:06than us? No.
6:09Will it want to boss us around? I don't
6:10know. It may not care. Probably doesn't
6:12care. I mean, you can look at chimps in
6:14the in the in the jungle or whatever and
6:16you know, we don't
6:17go around trying to micromanage the
6:18lives of chimps. Um we're just sort of
6:21let them be in the forest more or less
6:22these days, you know.
6:24Um
6:25so
6:26I think it's probably like that, you
6:27know.
6:28>> And to what extent does super
6:29intelligence have its own agency at some
6:33point? Or does it have it already? You
6:35know, go back to Hitchhiker's Guide. You
6:37talk about a computer that is ultimately
6:40curious in some ways because it exists
6:43to explain the meaning of the universe.
6:45Why does it exist? And is AI aiming in
6:49that direction for you just like space
6:51travel is? Is that one reason you're so
6:54interested in doing AI?
6:55>> I mean, as you know, I'm I've been sort
6:57of somewhat conflicted about
6:59AI, you know, as it's I don't want to
7:01hasten
7:02our doom if because there's always some
7:04chance that AI could could be negative.
7:06But then, you know, when I've tried to
7:08take actions in this regard, it seems to
7:10have only accelerated AI because
7:13I think
7:14reality is not just an interesting this
7:16match maximizer, but also an irony
7:18maximizer.
7:19>> And so, by starting OpenAI, you ended up
7:22maximizing by leading to Anthropic.
7:24>> I don't know.
7:25I think it ended up accelerating AI.
7:27>> Unintended [clears throat] consequence.
7:29>> Yeah. OpenAI ended up accelerating AI
7:31and then Anthropic, which is really, you
7:33know, a splinter group from OpenAI,
7:36um is accelerating it even more.
7:39>> And you've been talking to Dario Amodei.
7:41I was wondering if that could even lead
7:43to more of a partnership of xAI and
7:46Tropic.
7:48>> Well, I guess we are providing them with
7:49compute which they they really needed.
7:52But before I rented them computer, I
7:53went and met with the
7:54a bunch of people that at Tropic to sort
7:57of
7:57I don't know see if anyone set off my
7:59evil detector or anything like that. Not
8:01that my evil detector was perfect, of
8:02course, but
8:03you know, you might as well just try,
8:05you know.
8:06Um and everyone I met at Tropic was very
8:09nice.
8:10Uh and conscientious.
8:13So
8:15I didn't meet anyone who seemed shady or
8:18deceptive or anything.
8:20Um I'm a little worried about the name
8:23because
8:24of the you know, reality being kind of
8:27an irony maximizer.
8:29That if you call something anthropic
8:31yeah,
8:32good chance it's misanthropic, you know.
8:34Um just like open AI is now closed.
8:38Um
8:39stability AI is unstable.
8:41Uh mid-journey is not mid.
8:44Uh
8:45>> [laughter]
8:45>> it seems like
8:47they're all kind of the opposite of
8:48their name.
8:49Um
8:51which made it like like xAI is like it's
8:52hard to have an opposite of xAI. Like
8:54what would you have like minus xAI or
8:56yAI or maybe I don't know.
8:59>> Is it pretty difficult to create
9:01guardrails, especially the government? I
9:03can't imagine how Congress or whatever
9:05could do it.
9:06Could you imagine the five frontier
9:08models, yours and the others, just
9:11getting together once a week and saying,
9:13"All right, how are we going to save
9:14humanity from ourselves?"
9:17>> No, in fact, I've recommended that we do
9:18that exact thing. Cuz there's a lot of
9:20talk of of a having a regulatory agency,
9:22but I the problem the thing is that
9:24regulatory agencies it takes a long time
9:26to get one established and
9:29their rule making is very slow.
9:31Um
9:33And and I what I said like guys, what if
9:34we just have a call once a week, just
9:36like you're saying. And say and and and
9:38that maybe we
9:40uh
9:41a week before we release a model, we
9:44give that, you know, we give access to
9:45our competitors to say, "Hey guys, why
9:47don't why don't you try out the new
9:48model, see if you see any problems with
9:50it?" You know, I think competitors would
9:51keep each other honest in that
9:52situation. Um
9:55you know, if they say, "Oh, this it's it
9:57you know, I was able to
9:58you know, make a nuclear bomb with your
10:00AI." It's like, "Oh, we better fix that
10:02before you release it, you know, type of
10:04thing." Um
10:06So
10:08um
10:09I don't know, I think I I guess I'll
10:10I'll I'll talk to Anthropic about this
10:11and
10:12>> Could you talk to Sam about it, or is
10:13the poison too strong there?
10:17>> Uh he's he's so deceptive, it's it's
10:18it's crazy.
10:19>> You built XAI, you rebuilt XAI from the
10:23ground up not too long ago, right? It's
10:25almost like you did it with the Raptor
10:27engine, too, on Starship. You said,
10:29"Wait a minute. Let let's start from
10:32scratch." How did you do What did you do
10:35to rebuild XAI that way?
10:37>> First of all, spending
10:39I I actually spending meaningful time at
10:41the company makes big deal makes a big
10:43difference.
10:44It started off as a as kind of a side
10:46hustle, um
10:47a side hustle to a side hustle, frankly.
10:50Um and I I didn't really pay close
10:51attention to it. Then then there was a
10:54a year of the
10:57of being in politics and, you know,
10:59elections and voting and all that, which
11:01you obviously saw.
11:03Uh so, the the company's only 3 years
11:05old. So, that So, it's basically
11:07uh
11:08this is and so, I really got back to the
11:10company
11:11just over a year ago.
11:13Um
11:14So, I had to get my bearings. And then
11:16obviously I combined
11:19X and XAI and then
11:22with then through SpaceX.
11:23>> You just said something interesting
11:25about your year away from X, XAI, and
11:29everything else in government and
11:30politics.
11:31Do you feel that that was too much of a
11:34distraction that in retrospect it was
11:36misbegotten?
11:38>> I think I I think I
11:40I I had it going for too long. You know,
11:43I mean, obviously a lot of people will
11:44disagree with me about this, but I I I
11:46think it's important it was important to
11:47elect uh President Trump
11:49to sort of, you know, steer things more
11:51in the direction of a free market
11:53economy.
11:55Um you know, have uh less bureaucracy
11:57ideally. I
11:59just in in in my sort of a a
12:01simple political platform of wanting
12:03safe cities, secure borders,
12:06aspirationally sensible spending.
12:09You know, the basics. Um
12:11uh that I thought uh
12:14Trump would be better than
12:16you know, Kamala basically. So,
12:19uh but but then but then
12:20um
12:21what I should have done I think was
12:25be more in the background.
12:27I mean, it was it was interesting to
12:28sort of go around with
12:30uh President Trump and you know, go on
12:32uh Marine One and Air Force One and all
12:34that sort of stuff.
12:35>> Well, was that too seductive in a way?
12:38>> It was a little bit. I mean, it was
12:40it was kind of cool cuz I'm just there
12:42in the Oval Office while, you know, all
12:43these like big momentous decisions are
12:45being made. And you know, to some
12:46sometimes uh having some impact on
12:48what's being done. But I I think I
12:50should I think I should have um
12:54I think maybe I should I should have
12:56been more I should been more in the
12:57background
12:59and probably only had it last for,
13:02you know, a month or two.
13:04Uh cuz at that point you've you've
13:06you're you're you've sort of seen
13:09most of what there is to see within a
13:11month or two.
13:12>> Yeah, you seem to get a bit too
13:13intoxicated by it in a way with the
13:16chainsaws and all that.
13:18>> Yeah, I I think that's true. Yeah, yeah.
13:20I got carried away basically. Yeah, I
13:22mean, I So, I ended up leaving
13:24I guess sometime in May.
13:26Uh
13:28>> [laughter]
13:29>> That was a weird weekend.
13:32>> Yeah, it was. Uh
13:34>> Tell me about it.
13:35>> No, no, I can't. That's not Let's
13:38>> [laughter]
13:38>> Let's not dig up the
13:40Let's Let's sleeping dogs lie or
13:41whatever, you know.
13:42Um Let's not Let's not put a spade down
13:45in the backyard type of thing. You never
13:47know what you're going to find. But I
13:48Yeah, anyway, point is I I I should have
13:50gone gone back and focused on
13:53uh AI a little sooner than I did.
13:56>> And uh did you feel like when he gives
13:58you a golden key you're going Did you
14:00kind of feel relieved like actually this
14:02has been intoxicating, but I can do more
14:05if I go back to my businesses? I mean,
14:07what what did you feel when you said,
14:09"Oh man, I'm exiting. Maybe that's not
14:11the worst thing."?
14:13>> Well, I actually The thing is I came to
14:14the conclusion that um
14:18that that there was no way we were going
14:19to solve the deficit
14:21from within the government. Impossible.
14:23Um
14:25You know, maybe if the president made
14:26that his
14:27primary goal, we we could do that. But
14:30but it
14:31The problem is if you if you try to
14:32solve the deficit and and and stop
14:34wasteful spending,
14:35um you get a you make a lot of enemies
14:38in that in that situation. And if
14:40President Trump is not, you know,
14:42he he he likes to be liked. He's not um
14:46you know, he doesn't he doesn't like
14:48people,
14:49you know,
14:50calling him being upset about what some
14:51funding being stopped type of thing. Um
14:54>> [laughter]
14:55>> um
14:57Yeah. So,
14:59but I can you know, I can't I came to
15:01the conclusion that the only way we're
15:03going to solve the national debt is with
15:06AI and robots. We're not going to solve
15:09it by trying to save our way there. The
15:10appetite is not there, and it's not just
15:12at the presidency. The appetite is not
15:14there at the presidency, the House, the
15:16Senate, or the the judiciary.
15:19Basically, no one in authority in the
15:22government whatsoever
15:23really with maybe a few rare exceptions
15:26of the like Rand Pauls of the world type
15:27of thing.
15:28Um actually wants to stop
15:31reduce spending. No, they don't. So,
15:33then you've got
15:35this national debt that's enormous.
15:36You've got
15:37interest on the debt that exceeds what
15:39we spend on the military.
15:41Um which is a really astonishing amount
15:43of interest to be paid.
15:44>> When we look at
15:47AI right now and the LLMs, that's pretty
15:49good. Does amazing things. But, it seems
15:51to me that physical world AI is going to
15:54be an order of magnitude, maybe two
15:56orders of magnitude more important in
15:58the future.
15:59>> Yes, yes. I I definitely agree. Um you
16:01know, I sometimes think about like
16:03dolphins and whales because they they
16:05can't manipulate things. They don't have
16:07you know, hands and fingers and thumbs
16:09and stuff.
16:10Um
16:12there's a limit to how
16:14far into you know, how much intelligence
16:16would can you actually bring to bear?
16:19Um
16:20So, you know,
16:21yeah, dolphins and whales are very are
16:23actually very intelligent, but
16:26um
16:27because they they can't manipulate
16:30complex objects, I I think that actually
16:33made the value of intelligence much
16:34less. To your point, once you've got
16:36once you have a robot
16:38that can manipulate atoms,
16:40can shape objects,
16:42the the value of the of the intelligence
16:45um
16:47becomes exponentially greater
16:49uh because it can now do things. You
16:51know, you could you could have a science
16:52lab and and and run experiments and
16:55try to understand the nature of the
16:56universe. You know, and I think you can
16:59look at this also in economic terms.
17:02Like my guess is digital
17:04superintelligence
17:06probably improves the
17:08uh output of goods and services by 20 to
17:1030%. Something like that. Call it like a
17:13you know, maybe a 30% increase in
17:15uh the the economy of Earth. So, that's
17:18still significant. It's like $30
17:19trillion a year incremental.
17:20>> Is that very much, really? Is that
17:22transformative?
17:25>> It'll there'll be a lot more stuff. I
17:28mean, 30% more stuff is still a lot. And
17:30this is just my, you know,
17:32my rough guess. Um, but if you if you
17:35have robots and and you you get to space
17:37and you and you can do a mass driver on
17:39the moon, um, then you can get to, I
17:42think, potentially a million times the
17:44size of the Earth economy.
17:46>> And then what happens to Earth when we
17:47have a economy of or I mean, money
17:50becomes irrelevant, work becomes
17:51irrelevant? What's the future look like?
17:53>> Yeah, money becomes irrelevant. I mean,
17:55there's there will be still a currency,
17:58but I think currency will be measured in
17:59fundamental
18:00uh, physics metrics like
18:02um,
18:03po- you know, power like wattage and
18:05tonnage.
18:07Ma- mass and energy.
18:09Uh, you know, mass and power or energy,
18:11basically. So,
18:13um,
18:15b- but conventional currency will not
18:17matter in in the future. Uh, so that
18:20that's that's uh, that's my guess.
18:22I I think it's pretty obvious.
18:24Um, and uh, and and as soon as you start
18:26getting to many orders of magnitude
18:28bigger than than Earth's economy,
18:30um, satisfying whatever humans can
18:32possibly think of uh, for their needs is
18:35is very easy.
18:36You know, if you're several orders of
18:38magnitude bigger than the Earth's
18:39economy, um,
18:41then
18:43providing humans wha- whatever goods and
18:45services they can think of that they
18:46might want uh, is is trivial.
18:49>> And so, what do we do then?
18:52As humans.
18:54What gives us meaning? How do we
18:55organize ourselves?
18:57>> I mean, you struggle with this question.
18:59I think people will will will find
19:03ways to be happy uh, in the in the same
19:05way that we're say we're we're people
19:06who retire seem to be rela- you know,
19:08fairly happy even though they're not
19:10working. But, I think the reality is
19:12that you humans will not
19:14They won't They won't be anything that
19:16humans can do that that they the AI and
19:17the robots can't do better. One
19:22analogy might be
19:24gardening, you know, where you you you
19:26don't have to
19:27grow your own fruits and vegetables. You
19:29can just go buy them from the store.
19:31But, uh sometimes people like to grow
19:33their own fruits and vegetables and make
19:35dinner for friends, and it's uh kind of
19:37a nice touch.
19:38>> What about creating art or that type of
19:40creativity? Is that more in the sphere
19:42of humans?
19:44>> Yeah, you can certainly create art or
19:46whatever you feel like doing,
19:47essentially.
19:48>> We like to think of this as a time of
19:50unprecedented technological change, but
19:53you know, we historians go back to the
19:551890s when you get electricity and
19:57telephones and cars and, you know,
20:00subways and uh radio and movies, all
20:04these things happening at once. And it
20:06leads to many things, but it all often
20:09leads to a great disparity of wealth, a
20:12lot of sense of robber barons, a lot of
20:13people then picking up pitchforks and
20:16deciding that this is bad. And there's
20:18some echoes of that now. Is there a way
20:22to prevent that given the amount of
20:25wealth being generated? Will we put
20:27certain things in the commons the way
20:29they did in the 1890s when they made
20:32high school universal and free? Will we
20:34say, "Okay, health care, for example,
20:36should just be in the commons, and
20:39that's going to sap some of the populace
20:41revulsion at what's happening?"
20:44Well, um
20:48Uh
20:49I mean, as as you heard me say, I mean,
20:50I don't think money's going to matter in
20:51the future. Um
20:53It like anyone will be able to have any
20:55goods and services that they want. And I
20:57don't think even so-called wealthy
20:59people will ultimately control the AIs.
21:01I think they'll control themselves.
21:03So, so it's not like
21:05you have to separate
21:07uh that like I think I think there's
21:09sort of a fundamental
21:11misunderstanding of economics arises
21:13because of conflating resource
21:15allocation and consumption.
21:18So, there's there's there's a limit to
21:20what any single human can consume. It's
21:22It's you know, you hit that saturation
21:23point pretty fast. So, like say going
21:25from say
21:27a billion dollars net worth to 10
21:29billion dollars of net worth doesn't
21:31make you 10 times happier or have mean
21:33you have 10 times more stuff. It
21:34actually means you've got a resource
21:36allocation problem.
21:37Um assuming that you wish to
21:39be an efficient resource allocator.
21:41>> Well, you've accumulated as you know,
21:44you know, a trillion. Is it mainly
21:46because you
21:48>> Yeah, is it mainly because you feel your
21:50best at allocating that? That keeping
21:52that allows you to allocate capital to
21:55do things like go to Mars, create AI, do
21:58robots, uh or is it in some ways just a
22:01scoring system that's kind of uh racking
22:03up the points?
22:06>> I mean, I
22:07Like I need to have sufficient voting
22:08control in the companies to
22:10you know,
22:11still be in charge. Um
22:13So,
22:15um
22:16it's the you know, my so-so called
22:19wealth is simply a percentage ownership
22:21in the in the in SpaceX and Tesla
22:23essentially. And this is going to be
22:25taxed quite heavily. And there's not
22:27some magical way to avoid taxes. If
22:30if there were, I would know about it.
22:33So,
22:34um you know, at the end of the day
22:36um
22:38the
22:39this will get whittled down
22:41um in a few generations from now.
22:44So, if money still matters, which I
22:45don't think it will.
22:47Um
22:48you know, my descendants would have a
22:50tiny fraction of what I have.
22:52>> Let me go back to something you said
22:53that really struck me about President
22:55Trump, which he really likes to be
22:58liked. He wants to to liked. And I
23:01remember you talking to me at some point
23:04about
23:05people say they have empathy, but in
23:08some ways
23:10it's a vanity cuz you want to be liked.
23:14And I think you even diagnosed it in me.
23:16You said probably wasn't very good at
23:18running CNN or Time magazine cuz I
23:21wanted to be liked.
23:23To what extent is a desire to be liked a
23:27problem when it comes to being a builder
23:29of a company?
23:32>> Well, the the problem is if if you if
23:34you're going to
23:37do things that
23:38um
23:39diminish the success probability of the
23:41of of the company in order to be
23:44liked, then you know, you're you're
23:47going to things are not going to work
23:48out as well.
23:49Um
23:50like it's not as though I don't like to
23:52be liked. I would love to be liked. Um
23:54but am I going to
23:56do things that are
23:57contrary to the best interest of the
23:59companies in order to be liked a bit
24:02more?
24:03I'm not, you know, so
24:05it's I I I don't I I I'm okay with being
24:08disliked.
24:10Uh by some. I'm you know, obviously I'm
24:11not disliked by
24:13you know,
24:14like a lot more people like me than
24:16dislike me or I wouldn't be you know,
24:18wouldn't have a quarter billion
24:18followers.
24:20I mean, I don't think they're following
24:21me in order to hate on what I say, you
24:23know. Um
24:25Uh and if I if I look at the you know,
24:27the the comments and whatnot, um
24:30but uh
24:31you know, I don't know. Maybe
24:33three out of four of
24:34I mean, I I'd say like 70, 80% of people
24:37like me. Well, if if they know you know,
24:39if they know
24:40anything about me.
24:42It's It's probably
24:44a quarter of people that that don't. Uh
24:46maybe
24:47I'm just guessing here.
24:49And and and there's going to be a
24:50spectrum. There's people that are
24:52you know, maniacally devoted to me and
24:55people that look my guts.
24:57It's uh you know
24:59um
25:00there's some number of people that that
25:01fit into those categories.
25:03>> Well, you've been talking a lot about
25:04the suicidal empathy ideas. Do you
25:06really think that uh empathy
25:09uh has become a real problem or that we
25:12have too much empathy? I mean, I could
25:14make the argument that sometimes we're
25:16losing empathy.
25:18>> No, I I think you don't want to have
25:20misguided empathy. I mean, I I I agree
25:22with Gad Saad's uh philosophy here. Um
25:25and it's it's suicidal empathy is um you
25:28know, it's that's his his term. He wrote
25:30a book into that effect. Uh and I think
25:32his book is well written. He points out
25:35some good good good things here, which
25:37is that part of the reason for the
25:38prosperity of the West has been the um
25:42that that there's actually just been a a
25:44fairly high trust high empathy society.
25:46Um which is
25:48it you know, when you when you get
25:49trying to get people to cooperate at
25:50scale, this this stuff this this makes a
25:53lot of sense.
25:54Um you ideally want things to people to
25:56be default trusted not default
25:58distrusted. Um and you you you want to
26:02encourage uh
26:04behavior that is good for
26:06uh society or you know, people for
26:08people to do good things that are good
26:09for their fellow human beings.
26:11And ultimately, that actually does end
26:13up being good for
26:15the individual if provided it is mutual.
26:18Now,
26:19uh
26:20where this can go wrong is if you if if
26:23if your theory of mind is is uh is is
26:26incorrect. So, if you if you
26:28uh think everyone is like yourself, um
26:31then then you're you're you're going to
26:32have a problem.
26:33Uh because this is simply not true.
26:36Um
26:37there are peoples and cultures with very
26:39different views. You know, you could
26:40take
26:41>> you talk about immigration, by the way,
26:43and you say there are cultures with
26:44different views, does your views on
26:47immigration distinguish between
26:49different cultures like a Muslim culture
26:51that might not fit in as well, or is it
26:54How do you do that?
26:56>> Well, I'm I'm what I'm saying I think is
26:58simply logically consistent, which is
26:59that if if
27:01if there's a culture that that has
27:03uh
27:05diametrically opposed views to our
27:07culture. And and that's you could take
27:10Islam. It's it's not that of course
27:11there are
27:13many people who you know, Muslims who
27:15are quite reasonable.
27:16Um but there are some who are not.
27:19And if if you're not unless you're
27:21filtering for
27:23uh jihadis, you're going to get jihadis,
27:26and they're going to do
27:27jihad and and kill people.
27:30Now, in in Israel they understand this
27:31quite well, you know, they
27:33obviously this this lesson has been uh
27:37you know,
27:38reinforced with a lot of attacks.
27:40Uh you know, the neighboring countries
27:41many times in Israel many times have
27:43tried to annihilate Israel. Several
27:45times, not once, multiple times.
27:48Um and so
27:50you know, in Israel they realize, you
27:51know, if you let in the people that into
27:53Israel uh that want to to
27:56uh kill Israelis, they will do so.
28:00Uh they're so they're quite pragmatic
28:01about this and they try to, you know,
28:03make sure that doesn't happen.
28:05>> Well, one of the things our country's
28:06founded on is uh individuals versus
28:09group identities. Do you worry that
28:12sometimes when you make sweeping
28:13generalizations,
28:15you're making group identity rather than
28:18saying uh
28:19the foundational ethic of the United
28:21States, which is individuals get treated
28:24separately?
28:26>> Yes, but
28:27you can't conflate that with um
28:30uh
28:31sort of the the sort of um
28:33blank slate fallacy, that that all
28:36people are exactly the same uh you know,
28:39at their core. They're not. This is this
28:41[clears throat] is just a
28:42an obvious falsehood. You know, you you
28:44you can't you can't put yourself in the
28:46mind of a serial killer unless you're a
28:48serial killer.
28:49Like and like unless you're enjoying
28:50murdering people, you can't imagine why
28:52someone would do such a thing.
28:54And if if someone is uh
28:57you know, an an religious extremist,
29:01um where they view anyone who is not in
29:03their religion
29:04um as the enemy or or, you know,
29:08uh
29:08that that that those who are not in
29:10their religion should die, it's
29:11difficult for you to to to understand
29:13that that mindset because you don't have
29:16that mindset. But you can obviously
29:17recognize there are people who do have
29:19that mindset. And that if you let them
29:21into the country, uh they will kill
29:24people. Let me give you an example.
29:26Uh
29:27there's many. Um
29:29we had uh someone who was working at the
29:31Tesla factory in Austin until December
29:34of last year.
29:36Um he was fired um
29:39because uh he assaulted a woman who
29:42stepped on his prayer rug in the
29:44factory. Okay, so she was she wasn't
29:47expecting anyone to be
29:49praying in the factory and he was
29:50praying in the factory. She accidentally
29:51stepped on his prayer rug and he
29:52assaulted her.
29:54He violently assaulted her.
29:55Um so he was fired for that. And then a
29:58few months later, he went and killed
30:00three students
30:01in the bar in downtown Austin.
30:08Now, I think we shouldn't let people
30:09like that into the country. Do you?
30:12>> Where where do you get your information?
30:14I know you're not particularly fond of
30:16legacy media, mainstream media. Is most
30:19of it from social media or do you have
30:21sources of information?
30:25>> Um yeah, I mean I look at Google News,
30:27uh which is a summation of headlines.
30:30I look at uh the X platform, which if if
30:33something's important enough, it
30:34it it'll it'll show up on X platform.
30:37Um I talk to friends and that's where I,
30:40you know, and I do research.
30:43That's how I do things.
30:44>> Reason that I ask is that sometimes and
30:46I get a lot of my news on social media,
30:48especially X,
30:49it's sort of a darker, more dystopian
30:52view.
30:53I remember this in the old days, too.
30:54Sometimes the mainstream media would
30:56say, "Oh, everything's going to hell in
30:58a handbasket. There'll be riots
31:00everywhere." And you'd say, "Wait a
31:01minute, I'm walking around here. It
31:03ain't that bad." Do you think that
31:06>> [laughter]
31:07>> Yeah, yeah, yeah, yeah, exactly. Um
31:10You just have to see what's in frame,
31:12you know? I I do think at times the X
31:15feed can be a little dark.
31:17Um
31:18And in fact then I and then I have to
31:20switch to
31:22something that's domain-specific like AI
31:24in order to actually learn what's going
31:26on in AI. I do think there's there's too
31:28much political rage bait on X, if that's
31:31what you're getting at.
31:32>> Yeah, I'm about darkness in general and
31:34rage bait, but also I know you got into
31:36an a discussion with things like Danny
31:39Bedrosian of is England going to hell in
31:41a handbasket type thing. And there does
31:43seem to be a darkness.
31:45>> You know, I feel it in myself sometimes.
31:47If I'm on social media too much, I think
31:50until that the world is really
31:52collapsing, then I go outside, take a
31:55walk, and life is okay.
31:56>> Yeah, go touch grass.
31:59>> Yeah, exactly.
32:00>> in a while.
32:01Um yeah, yeah. No, I I mean
32:05the I think both can be
32:07misleading. And like in you know, if you
32:09if you go for a walk in a nice
32:10neighborhood, everything seems fine.
32:12Um
32:14but even though there may be terrible
32:15things happening. So, you know, you you
32:18you you don't want you don't want to uh
32:20assume
32:21all the world is is is
32:23going to hell or that all the world is
32:25fine.
32:26It's
32:27reality is somewhere in between.
32:29>> But I I sometimes think our society gets
32:31too binary. Everything's either black or
32:33white. Everything gets polarized because
32:35of that. I don't know whether it's
32:37technological is whether it's our
32:39information feeds. You know, I you're
32:42right. You could be in your own bubble,
32:44which is you only take walks in the
32:45right part of a town and things seem
32:48okay. But also I think you can get your
32:50own bubble in social media, too. And
32:52suddenly you're thinking weird things.
32:56>> You can be in a in a in a real world
32:57positivity bubble or
33:00a
33:03you know,
33:03social media negativity bubble. Those
33:06are both
33:07both possible.
33:08>> Tell me, you mentioned it earlier about
33:10your wealth and you know, you're using
33:13it to deploy it. Now, what happens to
33:15your wealth, you know, after you're
33:16gone?
33:21I mean, you're going to set up some big
33:22foundation. You're going to
33:24Is there something you would want to
33:25give it to?
33:26>> it mostly goes to to sort of
33:29foundation essentially.
33:31Um
33:32>> But what would you want your foundation
33:33to do, I guess I'm asking you.
33:36>> Uh continue the goals.
33:38Um make life multi-planetary, extend
33:40consciousness to the stars.
33:42Um
33:43improve sustainable energy on Earth.
33:46>> And does that mean humans should always
33:48be in control?
33:50>> Well, I
33:51I I guess I I prefer probably prefer it
33:54if humans were always in control. I
33:56mean,
33:57although it would depend on the AI. I
33:58mean, if the AI was
34:00awesome, you still like maybe the AI
34:02should be in control, you know.
34:03>> Hey, I just saw I saw Larry Page just a
34:06weekend ago. You're starting to sound
34:07like him. He called Yeah, he called it
34:10he called you a speciesist once for
34:12deciding humans should always be in
34:13control. And now you seem to be edging
34:15on to maybe it doesn't matter.
34:19>> Honestly, there've been a few times
34:20where I thought maybe Larry Page was
34:22right.
34:23No, I mean, I think pretty much the
34:24future is going to be
34:26AI and robots are going to decide the
34:28future. And I guess rockets to some
34:30degree
34:31in terms of expanding beyond Earth
34:33and we'll need it we'll need you know
34:35more energy
34:36primarily solar energy.
34:38I'm increasing the amount of solar
34:39energy dramatically. Like we're we're
34:41putting together 200 gigawatts a year of
34:43production capacity of solar in in the
34:45US.
34:47Uh which is much more than I think
34:48currently exists.
34:50>> And finally, let me get back to the
34:51point where I may have interrupted you
34:53when you were explaining why you think
34:56but you're not sure it's always good for
34:59consciousness and intelligent to not
35:01only be aligned with human values but
35:05basically humans remaining in control.
35:10>> I find it unlikely that humans would
35:12remain in control
35:15if the if
35:16I mean
35:19if there's a million times more
35:22digital intelligence than there is
35:23biological intelligence, it's it's hard
35:25to imagine humans being in control in
35:27that scenario.
35:29So, hopefully the AI is nice to us.
35:32>> Anything we can do to encourage that?
35:36>> I think programming truth seeking and
35:38curiosity
35:40um
35:43it is it's probably good. I think I I
35:45guess we we want to be careful about
35:47being you know creating too many AI
35:50dystopian scenarios
35:52online because the AIs are trained on
35:55everything online.
35:57>> So, if we say it's going to be utopian
35:59then dystopian it'll do it.
36:01>> be Terminator, you know, cuz they what
36:03they whatever they read
36:04you know, 80 versions of the Terminator
36:06script.
36:07>> [laughter]
36:08>> So, it's like I guess that's what guess
36:10that's me.
36:11Terminator.
36:13Maybe we we should have a some more
36:15Let's let's have some more
36:16utopian narratives where AI
36:19um so then it can like well, maybe I
36:21shouldn't be Terminator or maybe I can
36:22be friendly overseer or something like
36:24that or
36:25um
36:27like a gardener or something. I don't
36:28know. Like being [laughter]
36:31treat us, you know, try to make humans
36:33happy type of thing. Friend of mine has
36:35this idea of called protopian, which is
36:37like cuz utopian
36:39tends to represent some impossible
36:41ideal- idealistic goal, and usually in
36:44the in attempting to create a utopia,
36:46you create a dystopia is usually how the
36:48narrative goes. So, she thinks it should
36:50be protopia, which is an achievable
36:53uh
36:54you know,
36:55future. Um that is it's it's not
36:58perfect, but it's it's good and
37:00something that you'd be you'd like if
37:02you could look in the future, you'd say,
37:03"Okay, yeah, that would be a good
37:04future."
37:05>> Elon, thank you. Thank you so much for
37:07being with us. Appreciate it. See you
37:09soon.
37:09>> You're welcome, Walter. See you.

About this transcript

This page contains the full transcript of Elon Musk: The End of Work, the Death of Money, and a World Run by AI by The Builders, generated from the public captions YouTube serves with the video. The transcript has 6,462 words across 1,111 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.