you2idea@video:~$ watch JN3KPFbWCy8 [2:16:46]
// transcript — 1372 segments
0:00 - The following is a conversation with Elon Musk, his fourth time on this, the Lex Fridman Podcast.
0:08 (Elon whistling) I thought you were gonna finish it. (Elon whistling)
0:25 It's one of the greatest themes in all film history. (whistling continues)
0:31 - Yeah, that's great. - So I was just thinking about the Roman Empire.
0:36 As one does. (Elon laughing) - There's that whole meme where old guys
0:42 are thinking about the Roman Empire at least once a day. - And half the population's confused
0:46 whether it's true or not. But more seriously thinking about the wars
0:51 going on in the world today. And as you know, war and military conquest
0:56 has been a big part of Roman society and culture. And it, I think, has been a big part of most empires
1:04 and dynasties throughout human history, so. - Yeah, they usually came as a result of conquest.
1:10 I mean, there's some, like the Hungarian Empire where there was just a lot of sort of clever marriages.
1:16 - But fundamentally there's an engine of conquest. They celebrate excellence in warfare.
1:23 Many of the leaders were excellent generals. - [Elon] Yeah. - That kind of thing.
1:28 So a big picture question, Grok approved, I asked if this is a good question to ask.
1:33 - [Elon] Get tested Grok approved. - Yeah. At least on fun mode.
1:38 (Elon laughing) To what degree do you think war is part of human nature
1:44 versus a consequence of how human societies are structured? I ask this as you have somehow controversially
1:55 been a proponent of peace. - I'm generally a proponent of peace.
1:58 I mean, ignorance is perhaps, in my view, the real enemy to be countered.
2:04 That's the real hard part. Not, you know, fighting other humans.
2:09 But all creatures fight. I mean, the jungle is a, you look at the, people think of this nature
2:15 as perhaps some sort of peaceful thing. But in fact it is not.
2:18 There's some quite funny Werner Herzog thing where he is like in the jungle,
2:24 like saying that it's like basically just murder and death in every direction.
2:29 I mean, the plants, animals in the jungle are constantly trying to kill each other,
2:34 every single day, every minute. So it's not like, you know, we're unusual in that respect.
2:40 - Well, there's a relevant question here. Whether with greater intelligence comes greater control
2:46 over these base instincts for violence. - Yes. We have much more vulnerability to control our
2:55 limbic instinct for violence than say a chimpanzee. And in fact, if one looks and say,
3:01 chimpanzee society, it is not friendly. I mean, the Bonobos are an exception,
3:06 but chimpanzee society is filled with violence. And it's quite, quite horrific, frankly,
3:15 that that's our limbic system in action. Like, you don't wanna be on the wrong side of a chimpanzee,
3:20 they'll eat your face off and tear and nuts off. - Yeah, basically, there's no limits or ethics
3:24 or there almost had just war. There's no just war in chimpanzee societies,
3:29 is war and dominance by any means necessary. - Yeah, chimpanzee society is a pro,
3:34 like a permanent version of human society. They're not like peace loving basically at all.
3:41 There's extreme violence. And then once in a while, somebody who's watched too many Disney movies
3:48 decides to raise a chimpanzee as a pet, and then that eats their face or rip their nuts off
3:54 or chew their fingers off and that kind of thing. It's happened several times.
3:58 - Ripping your nuts off is an interesting strategy for interaction.
4:02 - It's happened to people. It's unfortunate. Like that's, I guess a one way to ensure
4:06 that the other chimp doesn't like, you know, contribute to the gene pool.
4:10 - Well, from a martial arts perspective is the fascinating strategy.
4:14 - (laughing) The nut riper. (laughing) - I wonder which of the martial arts teaches that one.
4:21 - I think it's safe to say if somebody's got your nuts in their hands and as the option ripping them off,
4:28 you'll be amenable to whatever they want. - Yeah. (Elon laughing)
4:33 - So like I said, somehow controversially, you've been an proponent of peace on Twitter, on X.
4:38 - Yeah. - So let me ask you about the wars going on today and to see what the path to peace could be.
4:44 How do you hope the current war in Israel and Gaza comes to an end?
4:48 What path do you see that can minimize human suffering in the long term in that part of the world?
4:54 - Well, I think that part of the world is definitely like, if you look up the, there is no easy answer
5:00 in the dictionary. It'll be that, like the picture of the Middle East,
5:05 in Israel especially. So there is no easy answer. What my, this is strictly my opinion of, you know,
5:15 is that the goal of Hamas was to provoke an overreaction from Israel.
5:24 They obviously did not expect to, you know, have a military victory, but they expect,
5:30 they really wanted to commit the worst atrocities that they could in order to provoke
5:36 the most aggressive response possible from Israel. And then leverage that aggressive response
5:45 to rally Muslims worldwide for the course of Gaza and Palestine,
5:54 which they have succeeded in doing. So the counterintuitive thing here,
6:02 I think that the thing that I think should be done, even though it's very difficult,
6:06 is that I would recommend that Israel engage in the most conspicuous acts of kindness possible.
6:14 Every po- everything. That is the actual thing that we're taught the goal of Hamas.
6:19 - So in some sense, the degree that makes sense in geopolitics
6:24 turn the other cheek implemented. - It's not exactly turn the other cheek.
6:28 Because I do think that there's, you know, I think it is appropriate for Israel
6:35 to find the Hamas members and, you know, either kill 'em or incarcerate them.
6:45 Like that's something that's something has to be done 'cause they're just gonna keep coming otherwise.
6:51 But in addition to that, they need to do whatever they can. There's some talk of establishing,
7:00 for example, a mobile hospital. I'd recommend doing that. Just making sure that, you know,
7:08 there's food, water, medical, necessities, and just be over the top about it
7:17 and be very transparent. So you can't claim it's a trick. Like just put webcam on the thing, you know, or 24/7.
7:29 - Deploy acts of kindness. - Yeah. Conspicuous acts of kindness that,
7:35 with that are unequivocal. Meaning they can't be somehow, because Hamas will then,
7:41 their response will be, oh, it's a trick. Therefore you have to counter how it's not a trick.
7:47 - This ultimately fights the broader force of hatred in the region.
7:51 - Yes. And I'm not sure who said it, it's an apocryphal saying,
7:55 but an eye for an eye makes everyone blind. Now that neck of the woods,
7:59 they really believe in the whole eye for an eye thing. But I mean, you really have,
8:08 if you're not gonna just outright commit genocide, like against an entire people,
8:14 which obviously would not be acceptable to really shouldn't be acceptable to anyone,
8:19 then you're gonna leave basically a lot of people alive who subsequently, you know, hate Israel.
8:25 So really the question is like, how, for every Hamas member that you kill,
8:30 how many did you create? And if you create more than you killed,
8:35 you've not succeeded. That's the, you know, the real situation there.
8:40 And it's safe to say that if, you know, if you kill somebody's child in Gaza,
8:52 you've made at least a few Hamas members who will die just to kill an Israeli.
9:03 That's the situation. So, but I mean, this is one of the most contentious subjects
9:11 one could possibly discuss. But I think if the goal ultimately
9:15 is some sort of long-term peace, one has to be, look at this from the standpoint of,
9:21 over time, are there more or fewer terrorists being created?
9:26 - Let me just linger on war. - Yeah, war safe to say, wars always existed and always will exist.
9:33 - [Lex] Always will exist, - Always has existed and always will exist.
9:37 - I hope not. You think it always will? - There will always be war.
9:42 This question of just how much war. And, you know, there's this,
9:47 there's this sort of the scope and scale of war. But to imagine that there would not be any war
9:52 in the future, I think would be a very unlikely outcome. - Yeah you talked about the Culture series.
9:57 There's war even there. - Yes. It's a giant war. The first book starts off with a gigantic galactic war
10:05 where trillions die, trillions. - But it still nevertheless protects these pockets
10:10 of flourishing. Somehow you can have galactic war and still have pockets of flourishing.
10:18 - Yeah. I mean it's, I guess if we are able to one day expand to,
10:23 you know, the galaxy or whatever, there will be a galactic war at some point.
10:30 - The scale, I mean the scale of war has been increasing, increasing, increasing.
10:35 It's like a race between the scale of suffering and the scale of flourishing.
10:38 - Yes. - A lot of people seem to be using this tragedy to beat the drums of war
10:45 and feed the military industrial complex. Do you worry about this?
10:49 The people who are rooting for escalation and how can it be stopped?
10:56 - One of the things that did concerned me is that there are very few people alive today
11:00 who actually viscerally understand the horrors of war, at least in the US.
11:05 I mean, obviously there are people in on the front lines in Ukraine and Russia who understand
11:09 just how terrible war is, but how many people in the west understand it?
11:15 And my grandfather was in World War II. He was severely traumatized.
11:19 He was there, I think for almost six years in the, in eastern North Africa and Italy.
11:26 All his friends were killed in front of him. And he would've died too, except they randomly gave some,
11:34 I guess, IQ test or something. And he scored very high. And he was not an officer, he was a,
11:40 I think a corporal or a sergeant or something like that. Because he didn't finish high school
11:45 because he had to drop outta high school, 'cause his dad died and he had to work
11:49 to support his siblings. So because he didn't graduate high school,
11:54 he was not eligible for the officer corps. So, you know, he kind of got put into the cannon fodder
12:01 category, basically. But then, this, randomly they gave him this test.
12:07 He was transferred to British intelligence in London. That's where he met my grandmother.
12:13 But he had PTSD next level, like next level. I mean, just didn't talk, just didn't talk.
12:21 And if you tried talking to him, he'd just tell you to shut up.
12:24 And he won a bunch of medals. Never bragged about it once, not even hinted, nothing.
12:30 I like found out about it, 'cause his military records were online.
12:33 That's how I know. So he would say like, no way in hell do you wanna do that again.
12:41 But how many people, now he obviously he died, you know, 20 years ago or longer, actually 30 years ago.
12:50 How many people are alive that remember World War II? Not many.
12:54 - And the same perhaps applies to the threat of nuclear war.
13:01 - Yeah. I mean, there are enough nuclear bombs pointed at United States to make the revel,
13:06 the radio active revel balance many times. - There's two major wars going on right now.
13:14 So you talked about the threat of AGI quite a bit. But now as we sit here with the intensity
13:20 of conflict going on, do you worry about nuclear war? - I think we shouldn't discount the possibility
13:27 of nuclear war. It is a civilizational threat. Right now I could be wrong,
13:37 but I think the current probability of nuclear war is quite low.
13:43 But there are a lot of nukes pointed at us. So, and we have a lot of nukes pointed at other people.
13:50 They're still there. Nobody's put their guns away. The missiles are still in the silos.
13:57 - And the leaders don't seem to be the ones with the nukes talking to each other.
14:03 - No. There are wars which are tragic and difficult on a local basis.
14:08 And then there are wars which are civilization ending or has that potential.
14:12 Obviously global thermonuclear warfare has high potential to end civilization, perhaps permanently.
14:19 But certainly, you know, to severely wound and perhaps set back human progress by,
14:30 you know, to the stone age or something, I don't know. Pretty bad.
14:39 Probably scientists and engineers wanna be super popular after that as well.
14:43 And like you got us into this mess. - So generally I think we obviously wanna prioritize
14:50 civilizational risks over things that are painful and tragic on a local level, but not civilizational.
15:00 - How do you hope the war in Ukraine comes to an end? And what's the path, once again
15:04 to minimizing human suffering there? - Well, I think that what is likely to happen,
15:13 which is really pretty much the way it is, is that something very close to the current lines will be
15:25 how a ceasefire or truth happens. But, you know, you just have a situation right now
15:33 where whoever goes on the offensive will suffer casualties at several times the rate of whoever's on the defense.
15:41 'Cause you've got defense in depth, you've got minefields, trenches, anti-tank defenses.
15:48 Nobody has air superiority. 'cause the anti-aircraft missiles
15:53 are really far better than the aircraft. Like, they're far more of them.
15:58 And so neither side has air superiority. Tanks are basically death drafts, just slow moving
16:06 and they're not immune to anti-tank weapons. So you really just have long range artillery
16:13 and infantry ranges. It's World War I all over again with drones, you know,
16:22 throwing old drones, some drones there. - Which makes the long range artillery
16:26 just that much more accurate and better and so more efficient at murdering people on both sides.
16:33 - Yeah. So it's whoever is, you don't wanna be trying to advance from either side,
16:39 'cause the probability of dying is incredibly high. So in order to overcome
16:49 defense in depth trenches and minefields, you really need a significant
16:54 local superiority in numbers. Ideally combined alms where you do a fast attack
17:02 with aircraft, a concentrated number of tanks and a lot of people.
17:10 That's the only way you're gonna punch through the line. And then you're gonna punch through
17:13 and then not have reinforcements just kick you right out again.
17:20 I mean, I really recommend people read World War I warfare in detail.
17:28 That's rough. I mean, the sheer number of people that died there
17:34 was mind boggling. - And it's almost impossible to imagine the end of it
17:40 that doesn't look like almost exactly like the beginning in terms of what land belongs to who and so on.
17:49 But on the other side of a lot of human suffering death and destruction of infrastructure.
17:56 - Yes. I mean, the thing that, the reason I, you know, proposed some sort of truce
18:03 or peace a year ago was because I've predicted pretty much exactly
18:08 what would happen, which is a lot of people dying for basically almost no changes in land.
18:16 And this, the loss of the flower of Ukrainian and Russian youth.
18:21 And we should have some sympathy for the Russian boys as well as the Ukrainian boys,
18:26 'cause they, Russian boys didn't ask to be on their front line, they have to be.
18:30 So there's a lot of sons not coming back to their parents, you know, and I think most of 'em don't really have,
18:43 they don't hate the other side. You know, it's sort of like as this saying about, like,
18:48 this saying comes from World War I, it's like young boys who don't know each other, killing each other
18:55 on behalf of old men that do know each other. The hell's the point of that?
19:02 - So Vladimir Zelensky said that he's not, or has said in the past,
19:06 he's not interested in talking to Putin directly. Do you think he should?
19:09 - Yeah. - Sit down man to man, lead a leader and negotiate peace?
19:14 - Look, I think I would just recommend do not send the flower of Ukrainian youth
19:21 to die in trenches. Whether he talks to Putin or not, just don't do that.
19:30 Whoever goes on the offensive will lose massive numbers of people
19:37 and history will not look kindly upon them. - You've spoken honestly about the possibility of war
19:44 between US and China in the long term. If no diplomatic solution is found, for example,
19:50 on the question of Taiwan and one China policy. - [Elon] Right.
19:53 - How do we avoid the trajectory where these two superpowers clash?
19:58 - Well, it's worth reading that book on the, difficult to pronounce, the Thucydides trap,
20:03 I believe it's called. I love war history. I like inside out and backwards.
20:09 There's hardly a battle I haven't read about. And trying to figure out like what really was the cause
20:16 of victory in any particular case as opposed to what one side or another claim the reason.
20:21 - Both the victory and what sparked the war? - Yeah, yeah, - [Lex] The whole thing.
20:26 - Yeah, so that Athens and Sparta is a classic case. The thing about the Greek is
20:31 they really wrote down a lot of stuff. They loved writing. You know, there are lots of interesting things
20:35 that happened in many parts of the world, but people didn't write down. (laughing)
20:38 So we don't know what happened, or they didn't really write with in detail.
20:44 They just would say like, we had a battle and we won. And like, well what? Can you add a bit more?
20:51 The Greeks, they really wrote a lot. (laughing) They were very articulate on, they just love writing.
20:57 So, and we have a bunch of that writing as preserved. So we know what led up to the Peloponnesian War
21:03 between the Spartan and Athenian alliance. And we know that they saw it coming.
21:16 I mean, the Spartans didn't write, they also weren't very verbose by their nature,
21:19 but they did write, but they weren't very verbose. They were terse.
21:23 But the Athenians and the other Greeks wrote a line and they were like, and Sparta was really kinda
21:29 like the leader of Greece. But Athens grew stronger and stronger with each passing year.
21:37 And everyone's like, well that's inevitable that there's gonna be a clash between Athens and Sparta.
21:45 Well, how do we avoid that? And they couldn't. They actually, they saw it coming
21:49 and they still could not avoid it. (laughing) So, you know, at some point if there's, if one group,
21:59 one civilization or country or whatever exceeds another, sort of, like, you know,
22:07 the United States has been the biggest kid on the block for,
22:10 since I think around 1890 from an economic standpoint. So the United States has been the economic,
22:18 most powerful economic engine in the world, longer than anyone's been alive.
22:28 And the foundation of war is economics. So now we have a situation in the case
22:36 of China where the economy is likely to be two, perhaps three times larger than that of the US.
22:46 So imagine you're the biggest kid on the block for as long as anyone can remember.
22:49 And suddenly a kid comes along who's twice your size. - So we see it coming. - [Elon] Yeah.
22:57 - How is it possible to stop? Is there some, let me throw something out there,
23:02 just intermixing of cultures understanding. There does seem to be a giant cultural gap
23:08 in understanding of each other. And you're an interesting case study
23:11 because you are an American, obviously you've done a lot of incredible manufacture here in the United States,
23:18 but you also work with China. - I've spent a lot of time in China
23:21 and met with the leadership many times. - Maybe a good question to ask is,
23:25 what are some things about China that people don't understand positive, just in the culture?
23:29 What's some interesting things that you've learned about the Chinese?
23:36 - Well, the sheer number of really smart, hardworking people in China is incredible.
23:44 There are really, say like how many smart, hardworking people are there in China,
23:48 there's far more of them there than there are here. I think in my opinion.
23:57 And they've got a lot of energy. So I mean, the architecture in China
24:04 that's in recent years is far more impressive than the US. I mean the train stations, the buildings,
24:11 the high speed rail, everything, it's really far more impressive than what we have in the US.
24:19 I mean, I recommend somebody just go to Shanghai and Beijing, look at the buildings
24:24 and go to, you know, take the train from Beijing to Xian, where you have the terracotta warriors.
24:30 China's got an incredible history, very long history. And you know, I think arguably the,
24:37 in terms of the use of language from a written standpoint, sort of one of the oldest,
24:44 perhaps the oldest written language. And then China, people did write things down.
24:50 So now China historically has always been, with rare exception, been internally focused.
24:59 They have not been inquisitive. They've fought each other. There've been many, many civil wars.
25:07 In the three kingdoms war, I believe they lost about 70% of their population.
25:14 So, and that, so they've had brutal internal wars, like civil wars that make the US Civil War
25:21 look small by comparison. So I think it's important to appreciate
25:31 that China is not monolithic. We sort of think of like China
25:36 as this sort of one entity of one mind. And this is definitely not the case
25:42 from what I've seen, and I think most people who understand China would agree.
25:48 People in China think about China 10 times more than they think
25:51 about anything outside of China. So it's like 90% of their consideration is,
25:58 you know, our is internal. - Well, isn't that a really positive thing
26:03 when you're talking about the collaboration and a future piece between superpowers,
26:08 when you're inward facing, which is like focusing on improving yourself
26:12 versus focusing on... - [Elon] Yeah. - "Improving" others through military might.
26:18 - The good news, the history of China suggests that China is not inquisitive.
26:22 Meaning they're not gonna go out and invade a whole bunch of countries.
26:26 Now they do feel very strongly, you know, so that's good. I mean, 'cause a lot of very powerful countries
26:31 have been inquisitive. The US is one of the, also one of the rare cases
26:37 that has not been inquisitive. Like after World War II, the US could have basically taken
26:41 over the world in any country. Like, we've got nukes, nobody else has got nukes.
26:45 We don't even have to lose soldiers. Which country do you want?
26:49 And the United States could have taken over everything. Oh wait, at will.
26:53 And it didn't, and the United States actually helped rebuild countries.
26:57 So it helped rebuild Europe, you know, helped rebuild Japan.
27:02 This is very unusual behavior, almost unprecedented. You know, the US did conspicuous acts of kindness
27:13 like the Berlin airlift, you know? And I think, you know, it's always like,
27:22 well, America's done bad things. Well, of course America's done bad things.
27:25 But one needs to look at the whole track record. And just generally, you know, one sort of test
27:34 would be how do you treat your prisoners at war? Or let's say, you know,
27:42 no offense to the Russians, but let's say you're in Germany,
27:45 it's 1945, you got the Russian Army coming one side and you got the French, British,
27:51 and American Army's coming the other side. Who would you like to be to surrender it to?
27:56 Like, no country is like morally perfect, but I recommend being a POW with the Americans.
28:03 That would be my choice very strongly. (laughing) - In the full menu of POWs.
28:09 - Very much so. And in fact, Wernher von Braun, took, you know, smart guy was like,
28:17 we've got to be captured by the Americans. And in fact, the SS was under orders to execute Von Braun
28:26 and all of the German rocket initiatives. And they narrowly escaped their SS.
28:34 They said they were going out for a walk in the woods. They left in the middle of winter with no coats.
28:39 And they ran like, with no food, no coats, no water. And just ran like hell and ran west.
28:45 And by Sherlock, they, I think his brother found like a bicycle or something,
28:51 and then just cycled west as fast as he couldn't have found a US patrol.
28:58 So anyway, that's one way you can tell morality is. Where do you wanna be PoW? (laughing)
29:07 It's not fun anywhere, but some places are much worse than others.
29:10 So anyways, so like America has been, while far from perfect, generally a benevolent force,
29:21 and we should always be self-critical and we try to be better.
29:27 But anyone with have brain knows that. So I think there are some,
29:33 in this way China and the United States are similar. Neither country has been acquisitive in a significant way.
29:42 So that's like a, you know, a shared principle I guess. Now China does feel very strongly about Taiwan.
29:51 They've been very clear about that for a long time. You know, from their standpoint,
29:57 it would be like one of the states is, you know, not there like Hawaii or something like that.
30:04 But more significant than Hawaii, you know? And Hawaii is pretty significant for us.
30:11 So they view it as really the, that there's a fundamental part of China.
30:20 The islander Formosa not Taiwan, that is not part of China, but should be.
30:27 And the only reason it hasn't been is 'cause the US Pacific Fleet.
30:32 - And as their economic power grows and as their military power grows,
30:36 the thing that they're clearly saying is their interest will, you know, clearly be materialized.
30:46 - Yes. China has been very clear that they'll incorporate Taiwan
30:54 peacefully or militarily, but that they will incorporate it from their standpoint is 100% likely.
31:04 - You know, something you said about conspicuous acts of kindness.
31:07 As a geopolitical policy, it almost seems naive, but I'd venture to say that this is probably
31:15 the path forward, how you avoid most wars. Just as you say it, it sounds naive,
31:23 but it's kind of brilliant. If you believe in the goodness of underlying most of human nature.
31:31 It just seems like it conspicuous acts of kindness can reverberate through the populace
31:38 of the countries involved. - Yeah, well. - [Lex] And deescalate.
31:44 - Absolutely. So after World War I, they made a big mistake.
31:50 You know, they basically tried to lump all of lamb on Germany
31:56 and, you know, settled Germany with impossible reparations. And, you know, really, there was a lot of,
32:07 there was a fair, quite a bit of blame to go around for World War I,
32:12 but they try to, you know, put it all in Germany. And that was, that laid the seeds for World War II.
32:25 So it's a lot of people were not just Hitler, a lot of people felt wronged
32:31 and they wanted vengeance and they got it. - [Lex] People don't forget.
32:39 - Yeah, you know, you kill somebody's father or mother or son, daughter, they're not gonna forget it.
32:47 They'll want vengeance. So after World War II, they're like, well,
32:52 the Treaty of Versailles was a huge mistake in World War I. And so this time, instead of,
33:05 you know, crushing the losers, we're actually gonna help 'em with the Marshall Plan
33:10 and we're gonna help rebuild Germany. We're gonna help rebuild or, you know, Austria
33:17 and the other, you know, Italy and whatnot, so. And that was the right move.
33:26 - There's a, it does feel like there's a profound truth to conspicuous acts of kindness being an antidote to this.
33:37 - Something must stop the cycle of reciprocal violence. Something must stop it
33:42 or it will, you know, it'll never stop. Just eye for an eye, tooth for tooth,
33:51 limb for a limb, life for a life, forever and ever - To escape briefly,
33:58 the darkness was some incredible engineering work. xAI just released Grok AI assistant
34:04 that I've gotten a chance to play with. It's amazing on many levels.
34:08 First of all, it's amazing that a relatively small team in a relatively short amount of time was able to develop
34:15 this close to state-of-the art system. Another incredible thing is there's a regular mode
34:22 and there's a fun mode. - Yeah. I guess I'm to blame for that one. (laughing)
34:26 - I wish, first of all, I wish everything in life had a fun mode.
34:29 - Yeah. - There's something compelling beyond just fun about the fun mode
34:35 interacting with a large language model. I'm not sure exactly what it is
34:39 'cause I only have had a little bit of time to play with it.
34:42 But it just makes it more interesting, more vibrant to interact with the system.
34:48 - Yeah, absolutely. Our AI Grok is modeled after the Hitchhiker's Guide to the Galaxy,
34:59 which is one of my favorite books, which is, it's a book on philosophy, disguises book on humor.
35:07 And I would say that is that forms the basis of my philosophy, which is that
35:13 we don't know the meaning of life, but the more we can expand the scope
35:18 and scale of consciousness, digital and biological, the more we're able to understand
35:26 what questions to ask about the answer that is the universe.
35:30 So I have a philosophy of curiosity. - There is generally a feeling like this AI system
35:38 has an outward looking, like the way you are, like sitting with a good friend looking up at the stars,
35:46 like the asking pod head, like questions about the universe,
35:49 wondering what it's all about. The curiosity you talk about.
35:52 There's a sense no matter how mundane the question I ask, there's a sense of cosmic grandeur to the whole thing.
35:59 - Well we are actually working hard to have engineering math, physics answers that you can count on.
36:09 So for the other sort of AIs out there that these so-called large language models,
36:17 I've not found the engineering to be reliable. And the hallucination,
36:24 it unfortunately hallucinates most when you at least want it to hallucinate.
36:29 So when you're asking important difficult questions, that that's where it tends to be confidently wrong.
36:36 So we're really trying hard to say, okay, how do we be as grounded as possible
36:42 so you can count on the results? Trace things back to physics first principles,
36:48 mathematical logic. So underlying the humor is an aspiration
36:55 to adhere to the truth of the universe as closely as possible.
37:01 - [Lex] That's really tricky. - It is tricky. So that's why, you know,
37:07 there's always gonna be some amount of error, but we want to aspire to be
37:13 as truthful as possible about the answers with acknowledged error.
37:20 So that there was always, you don't wanna be confidently wrong, so you're not gonna be right every time,
37:26 but you wanna minimize how often you're confidently wrong. And then like I said, once you can count on the logic
37:35 as being not violating physics, then you can start to build on that,
37:42 to create inventions, like invent new technologies. But if you can't, if you cannot
37:50 count on the foundational physics being correct, obviously the inventions are simply wishful thinking,
37:56 you know, imagination land, magic basically. - Well, as you said, I think one of the big goals
38:04 of xAI is to understand the universe. - Yes. That's how simple three word mission. (laughing)
38:13 - If you look out far into the future, do you think on this level of physics,
38:17 the very edge of what we understand about physics, do you think it will make discoveries,
38:22 sort of the sexiest discovery of them as we know now, sort of unifying general relativity and quantum mechanics.
38:30 So coming up with a theory of everything, do you think it could push towards that direction?
38:35 Almost like theoretical physics discoveries. - If an AI cannot figure out new physics,
38:41 it's clearly not equal to humans, let alone nor has surpassed humans,
38:45 'cause humans have figured out new physics. They're just, you know, physics is just understanding,
38:50 you know, deepening wants insight into how reality works. And then there's engineering which is inventing things
38:59 that have never existed. Now the range of possibilities for engineering
39:03 is far greater than for physics. 'Cause you know, once you figure out the rules
39:06 of the universe, that's it. You've discovered things that already existed.
39:12 But from that you can then build technologies with that are really almost limitless
39:18 in the variety and cap- you know? It's like once you understand the rules
39:24 of the game properly, and we do, you know, with current physics, we do at least at a local level,
39:29 understand how physics works very well. Where our ability to predict things is incredibly good.
39:35 Like quantum mechanics is, the degree to which quantum mechanics can predict outcomes
39:39 is incredible. That was my heart hardest class in college by the way.
39:46 My senior quantum mechanics class was harder than all of my other classes put together.
39:50 - To get an AI system, a large language model to reliably, be as reliable as quantum mechanics
39:59 and physics is very difficult. - Yeah. You have to test any conclusions against the ground,
40:04 truth of reality. Reality is the ultimate judge. Like physics is the law,
40:07 everything else is a recommendation. I've seen plenty of people break the laws made by man,
40:13 but none break the laws made by physics. - Yeah, it's a good test actually.
40:17 If this LLM understands and matches physics, then you can more reliably trust
40:23 whatever it thinks about the current state of politics in some sense.
40:28 - And it's also not the case currently that even that its internal logic is not consistent.
40:34 So, especially with these, with the approach of like just predicting a token, predict,
40:42 token, predict, token, it's like a vector sum. You know, you're summing up a bunch of vectors,
40:47 but you can get drift. So as those, a little bit of error, a little bit of error adds up.
40:53 And by the time you are many tokens down the path, it doesn't make any sense.
40:59 - So it has to be somehow self-aware about the drift. - It has to be self-aware about the drift,
41:04 and then look at the thing as a gestalt as a whole and say it does it have coherence as a whole.
41:13 So, you know, when authors write books that they will write the book and then they'll go
41:20 and revise it, you know, taking into account, you know, all the, the end and the beginning
41:26 and the middle and rewrite it to achieve coherence so that it doesn't end up in a nonsensical place.
41:33 - [Lex] Maybe the process of revising is what... - Yeah. - Reasoning is, and then that's the process of revising
41:39 is how you get closer and closer to truth. Maybe you, like, at least I approached it that way.
41:45 You just say a bunch of bullshit first and then you get it better.
41:48 You start a bullshit and then you get... - Create a draft and then you iterate on that draft
41:53 until it has has coherence, until it all adds up basically. - So another question about theory of everything,
42:01 but for intelligence, do you think there exists as you're exploring this with xAI,
42:06 creating this intelligence system, do you think there is a theory of intelligence
42:10 where you get to understand what, like what is the I in AGI and what is the I in human intelligence?
42:22 - There's no I in Team America, oh wait, there is. (both laughing)
42:27 - No, it's gonna be stuck in my head now. (both laughing) Yeah. There's no me and whatever in quantum mechanics.
42:37 Oh, wait. (Elon laughing) I mean is that part of the process of discovering,
42:44 understanding the universe is understanding intelligence? - Yeah. Yeah.
42:50 I think we need to understand intelligence, understand consciousness.
42:53 I mean there are some sort of fundamental questions of like,
42:57 what is thought? What is emotion? - [Lex] Yeah. - Is it really just one atom
43:04 bumping into another atom? It feels like something more than that.
43:08 So I think we're probably missing some really big things, like some really big things.
43:18 - Something that'll be obvious in retrospect. - [Elon] Yes. - Like there's a giant,
43:22 like you put the whole consciousness, emotion. - Well, some people would quote like a soul,
43:28 you know, in religion it'd be a soul. Like you feel like you're you, right?
43:34 I mean, you don't feel like you're just a collection of atoms, but on what dimension does thought exist?
43:41 What dimension does do emotions exist? We feel them very strongly.
43:46 I suspect there's more to it than atoms bumping into atoms. - And maybe AI can pave the path to the discovery
43:55 of whatever the hell that thing is. - Yeah. What is consciousness?
43:59 Like what are, when you put the atoms in a particular shape,
44:02 why are they able to form thoughts and take actions, that and feelings.
44:10 - And even if it is an illusion, why is this illusion so compelling?
44:13 - Yeah. Like how do... - Why does this illusion exist? - [Lex] Yeah. - On what plain
44:19 does this illusion exist? - [Lex] Yeah. - And sometimes I wonder is,
44:26 you know, either perhaps everything's conscious or nothing is conscious, one of the two.
44:32 - I like the former, everything conscious just seems more fun.
44:37 - It does seem more more fun. Yes. But we're composed of atoms
44:43 and those atoms are composed of quarks and leptons. And those quarks and leptons have been around
44:47 since the beginning of the universe. - The beginning of the universe.
44:53 - Right. What seems to be the beginning of the universe. - The first time we talked, you said what you would,
44:57 which is surreal to think that this discussion was happening is becoming a reality.
45:02 I asked you what question would you ask an AGI system once you create it?
45:07 And you said, what's outside the simulation is the question.
45:10 And good question. - [Elon] Yeah. - But it seems like with Grok
45:14 you started to literally, the system's goal is to be able to ask such questions,
45:21 to answer such questions and to ask such questions. - Where are the aliens? - [Lex] Where are the aliens?
45:26 - That's one of the, like Fermi paradox question. A lot of people have asked me
45:31 if I've seen any evidence of aliens and I haven't, which is kind of concerning.
45:37 'Cause then I think what I'd probably prefer to at least have seen some archeological evidence of aliens,
45:44 to the best of my knowledge, there is no proof. I'm not aware of any evidence of aliens.
45:50 If they're out there, they're very subtle. We might just be the only consciousness,
45:54 at least in the galaxy. And if you look at say the history of Earth,
46:01 forensically the archeological record, Earth is about four and a half billion years old.
46:06 Civilization as measured from the first writing is only about 5,000 years old.
46:11 We have to give some credit there to the ancient Sumerians who aren't around anymore.
46:15 I think it was a archaic, pre cuneiform was the first actual symbolic representation,
46:21 but only about 5,000 years ago. I think that's a good date for when word say civilization started.
46:30 That's one millionth of Earth's existence. So civilization has been around,
46:37 it's really a flash in the pan so far. And why have we, why did it take so long for,
46:52 you know, four and a half billion years? For the vast majority of the time, there was no life.
47:00 And then there was archaic there bacteria for a very long time.
47:05 And then, you know, you had mitochondria get captured, multicellular life.
47:15 Differentiation into plants and animals. Life moving from the oceans to land.
47:20 Mammals, higher brain functions. And the sun is expanding slowly,
47:29 but it will overheat, it will heat the earth up at some point in the future,
47:38 boil the oceans and and earth will become like Venus. Where near life, life as we know it is impossible.
47:46 So if we do not become multi-planetary and ultimately go beyond our solar system,
47:53 annihilation of all life on earth is a certainty. A certainty.
48:00 And it could be as little as on the galactic timescale, half a billion years.
48:09 You know, it's long time by human standards, but that's only 10% longer
48:13 than earth has been around at all. So if life had taken 10% longer to evolve on earth,
48:21 it wouldn't exist at all. - We got a deadline coming up. Better hurry.
48:30 But that said, as you said, humans intelligent life on earth
48:35 developed a lot of cool stuff very quickly, - So... - [Elon] Yes.
48:39 - It seems like becoming a multi-planetary is almost inevitable.
48:45 Unless we destroy this thing. - We need to do it. I mean, it's not, I mean, I suspect that there,
48:52 if we are able to go out there and explore other star systems that we,
48:56 there's a good chance we find a whole bunch of long dead one planet civilizations,
49:01 then have made it past their home planet. - That's so sad. - Yeah.
49:05 - It's sad. Also fascinating. - I mean, there are various explanations
49:09 for Fermi paradox. And one is the sort of, they're these great filters,
49:13 which civilizations don't pass through. And one of those great filters is
49:19 do you become a multi-plan civilization or not? And if you don't, it's simply a matter of time
49:24 before something happens on your planet. You know, either natural
49:30 or manmade that causes us to die out. Like the dinosaurs, where are they now
49:38 that didn't have spaceships, so. (laughing) - I think the more likely thing is,
49:44 'cause just to empathize with the aliens that they found us and they're protecting us
49:50 and letting us be. - I hope so. Nice aliens. - Just like the tribes in the Amazon,
49:55 the uncontacted tribes, we're protecting them. That's what...
49:59 - That would be a nice explanation. - Or you could have like a, what was it?
50:04 I think Andrej Karpathy said, it's like the ants in the Amazon asking where's everybody?
50:10 - Well, they do run into a lot of other ants. - [Lex] That's true. - They have these ant wars.
50:15 - (laughing) Sounds like a good TV show. - Yeah, they literally have these big wars
50:19 between various ants. - Yeah, maybe I'm just (both laughing)
50:25 dismissing all the different diversity events. - You should listen to that Warner Herzog
50:29 talking about the jungle. It's really hilarious. Have you heard it?
50:32 - No, I have not. - [Elon] It's awesome. - But Warner Herzog is a way...
50:34 (both laughing) - You should play it for the, you know, as an interlude in the, it's on YouTube.
50:43 It's awesome. (both laughing) - [Lex] I love him so much. - Yeah, he's great.
50:47 - Was he the director of "Happy People: Life in the Taiga"? I think also
50:51 - He did that bear documentary, - [Lex] The bear documentary.
50:53 - And this thing about penguins. - Yeah. (both laughing) The analysis, the psychoanalysis of penguins.
51:00 - Yeah, the penguins like headed for like mountains like that are like 70 miles away.
51:05 And Penguin has just had for doom basically. - Well, he had a cynical take.
51:10 I have a, he could be just a brave explorer and there'll be great stories told
51:16 about him amongst the penguin population for many centuries to come.
51:23 (laughing) What were we talking about? Okay. - [Elon] Penguins. (both laughing)
51:28 - Yeah, so aliens, I mean, I don't know. Look, I think the smart move is just, you know,
51:35 this is the first time in the history of earth that it's been possible for life
51:40 to extend beyond earth. That window is open. Now, it may be open for a long time
51:49 or it may be open for a short time and it may be open now and then never open again.
51:55 So I think the smart move here is to make life multiplanetary while it's possible to do so.
52:01 We don't wanna be one of those lame one planet civilizations that just dies out.
52:04 - No, those are lame. - Yeah, lame. Self-respecting civilization would be one planet.
52:11 - There's not gonna be a Wikipedia entry for one of one of those.
52:15 And pause, the SpaceX have an official policy for when we meet aliens.
52:23 - No. (laughing) - That seems irresponsible. (both laughing)
52:30 - I mean, look, if I see the slightest indication that there are aliens,
52:35 I will immediately post on X platform anything I know. - It could be the most liked reposted post of all time.
52:42 - Yeah, I mean look, we have more satellites up there right now
52:46 than everyone else combined. So, you know, we'd know, we know if we've got a maneuver
52:51 around something, and we not don't have to maneuver around anything.
52:55 - If we go to the big questions once again, you said you've,
52:59 you're with Einstein that you believe in the God of Spinoza.
53:03 - (laughing)Yes. - So, you know, that's a view that God is like the universe
53:08 and is reveals himself through the laws of physics or as Einstein said,
53:13 through the lawful harmony of the world. - Yeah, I would agree that God of the simulator
53:19 or whatever the supreme being or beings, reveal themselves through the physics.
53:28 You know, they're creatives of this existence and it's incumbent upon us to try to understand
53:34 more about this wondrous creation. - Like who created this thing, who's running this thing?
53:41 Like embodying it into a singular question with a sexy word on top of it
53:45 is like focusing the mind to understand it. It does seem like there's a, again, it could be an illusion.
53:52 It seems like there's a purpose that there's an underlying master plan of some kind.
53:57 And it seems like... - There may not be a master plan. In the sense, so there's, like,
54:04 maybe an interesting answer to the question of determinism versus free will is
54:08 that if we are in a simulation, the reason that these higher beings would hold a simulation
54:14 is to see what happens. So it's not, they don't know what happens,
54:22 otherwise they wouldn't hold the simulation. So when humans create a simulation,
54:27 so it's SpaceX and Tesla, we create simulations all the time,
54:31 especially for the rocket. You, you know, you have to run a lot of simulations
54:38 to understand what's gonna happen because you can't really test the rocket
54:41 until it goes to space, and you want it to work. So you have to simulate subsonic, transonic,
54:49 supersonic, hypersonic ascent, and then coming back super high heating
54:56 and orbital dynamics. All this has gotta be simulated. So, 'cause you don't get very many kicks at the can.
55:04 But we run the simulations to see what happens. Not if we knew what happens, we wouldn't run the simulation.
55:09 So if there's, so whoever created this existence, is they're running it
55:17 because they don't know what's gonna happen. Not because they do.
55:23 - So maybe we both played Diablo, maybe Diablo was created to see if it Druid your character
55:30 could defeat Uber Lilith at the end. They didn't know. - Well, the funny thing is the Uber Lilith title
55:38 is Hatred Incarnate. - [Lex] Yeah. - And right now, I guess, you're gonna ask the Diablo team,
55:46 but it's almost impossible to defeat hatred in the eternal realm.
55:55 - Yeah, you've streamed yourself dominating tier 100 Nightmare Dungeons and still...
56:00 - I can cruise through tier 100 Nightmare Dungeons like a stroll in the park.
56:07 - And still you're defeated by Hatred. - Yeah, I can, there's the sort of,
56:11 I guess maybe the second hardest boss is Duriel, Duriel can't even scratch the paint.
56:16 So I killed Duriel so many times , and every other boss in the game, all of 'em,
56:23 kill 'em so many times it's easy. But Uber Lilith, otherwise known as Hatred Incarnate,
56:31 especially if you're a Druid and you have no ability to go and to be vulnerable,
56:35 there are these random death waves that come at you. And I'm pretty, you know, I really am 52,
56:45 so my reflex is not what they used to be, but I have a lifetime of playing video games.
56:50 At one point I was, you know, maybe one of the best Quake players in the world.
56:56 I actually won money in what I think was the first paid eSports tournament in the US.
57:03 We're doing four person Quake tournaments. And we came second, I was the second best person on the team
57:11 and the actual best person, we're actually winning, we would've come first except the best person on the team,
57:16 his computer crashed prior to the game. So we came second. But I got money for it and everything.
57:23 So like, basically I got skills, you know, albeit, you know, no spring chicken these days.
57:28 And to be totally frank, it's driving me crazy, trying to beat Lilith as a Druid.
57:36 Basically trying to beat, trying to beat Hatred Incarnate in the eternal realm.
57:41 - [Lex] As a Druid. - As a Druid. And if you, if you, (laughing)
57:44 this is really (indistinct), lemme tell you. - I mean, the challenge is part of the fun.
57:50 I have seen directly like you're actually like a world class incredible video game player.
57:56 - [Elon] Yeah. - And I think Diablo, so you're just picking up a new game
58:00 and you're figuring out it's fundamentals. You're also with the Paragon Board
58:04 and the build are not somebody like me who perfectly follows whatever they suggest on the internet.
58:12 You're also an innovator there. - [Elon] Yeah. - Which is hilarious to watch.
58:17 It's like a mad scientist just trying to figure out the Paragon Board
58:21 and the build and the, you know. Is there some interesting insights there about,
58:26 if somebody's starting as a Druid, do you have advice? - I would not recommend playing a Druid
58:31 in the eternal realm. Right now, I think the most powerful character
58:36 in the seasonal realm is the Sorcerer with the lightning balls.
58:41 The Sorcs have huge balls in the seasonal. - Well yeah, that's what they say.
58:48 - So Sorc have huge balls. They do, huge balls of lightning.
58:54 - I'll take you word for it. - And it's actually, in the seasonal realm,
58:58 you can, it's like pretty easy to beat Uber Lilith with the,
59:02 because she got these vampiric powers that out fire damage and increase your defense and whatnot.
59:09 So, but really quite easy to defeat Hatred seasonally, but to defeat hatred eternally, very difficult.
59:19 Almost impossible. It's totally impossible. It seems like this is a metaphor for life, you know?
59:24 - I like the idea that Elon Musk, because I saw, I was playing Diablo yesterday
59:29 and I saw 100, level 100 drew just run by, I will never die and then run back the way.
59:36 - And it was, there's just some, this metaphor's kind of hilarious that you,
59:40 Elon Musk is fighting Hatred, restlessly fighting Hatred in this demonic realm.
59:47 - [Elon] Yes. - It's hilarious. I mean it's pretty hilarious.
59:50 - No, it's absurd. (both laughing) Really, it's exercise and absurdity,
59:54 and it makes me wanna pull my hair out. - Yeah. What do you get from video games in general?
59:59 Is there for you personally. - I mean, it's, I don't known it's,
60:05 it calms my mind. I mean, you sort of killing the demons in a video game
60:11 calms the demons in my mind. Yeah. If you play a tough video game,
60:17 you can get into like a state of flow, which is very enjoyable.
60:24 But the admittedly it needs to be not too easy, not too hard, kinda in the Goldilock zone.
60:32 And I guess you generally wanna feel like you're progressing in the game.
60:36 So a good video, and there's also beautiful art, engaging storylines,
60:46 and it's like an amazing puzzle to solve, I think. And so it's like solving the puzzle.
60:52 - Elden Ring the greatest game of all time. I still haven't played it, but you.
60:56 - Elden Ring is definitely a candidate for best game ever. Top five for sure.
61:01 - I think I've been scared how hard it is or how hard I hear it is, so.
61:06 But it is beautiful. - Elden Ring is, feels like it's designed by an alien.
61:13 There's a theme to this discussion. In what way? - It's so unusual.
61:18 It's incredibly creative and the art is stunning. I recommend playing it on a big resolution,
61:25 high dynamic range TV even, doesn't need to be a monitor, just the art is incredible.
61:32 It's so beautiful And it's so unusual. And each of those top bus battles is unique.
61:43 Like, it's like a unique puzzle to solve. Each one is different.
61:48 And the strategy you use to solve one battle is different from another battle.
61:54 - That said, you said Druid and Enternal against Uber Lilith is the hardest boss battle you've ever?
62:00 - Correct. That is currently the, and I've played a lot of video games.
62:06 That's my primary recreational activity. And yes, beating Hatred in the Enternal Realm
62:18 is the hardest boss battle in life and in the video game. - [Lex] Metaphor on top of metaphor.
62:25 - I'm not sure it's possible, but it's, I do make progress. So then I'm like, okay, I'm making progress.
62:31 Maybe if I just tweak that Paragon Board a little more, I can do it.
62:38 I could just dodge a few more waves, I can do it. - Well the simulation is created for the purpose
62:45 of figuring out if it can be done. And you are just a cog in that simulation,
62:50 in the machine of the simulation. - Yeah it might be. (Lex laughing)
62:54 - I have a feeling that at least, I think... - [Lex] it's doable.
63:03 - It's doable, yes. - [Lex] Well that's the human spirit right there to believe.
63:09 - Yeah. I mean, did prompt me to think about just hate in general,
63:13 which is, you wanna be careful of one of those things where you wish for something that sounds good,
63:23 but if you get it, it's actually a dystopian situation. So, you know, you could sort of run a, sort of,
63:32 have heart to sort of like, if you wish for world peace, sounds good,
63:36 but how is it enforced? And at what cost is, at what cost eternal peace?
63:45 It might actually be worse to have eternal peace because of what that would entail.
63:50 The suppression of everyone. It might be the suppression of progress.
63:56 It might be an ossified society that never changes. What I'm saying is that there is an argument
64:04 that you, that if that, if you wish for no war, you should be careful what you wish for.
64:14 Because what's required in order for there to be no war? Might be worse than a little war.
64:22 - Yeah, the things we'll consider to be flaws of human civilization might be a necessary components
64:27 for whatever optimal looks like. I mean this, do you worry about AI, AGI enabling a dystopian
64:33 state of this nature, whether it's 1984 with surveillance and fear, or brave new world with pleasure,
64:43 and what is it? Lots of sex, but no deep human experience. - You know, there's actually a real drug called Soma.
64:54 - [Lex] There is? - Yes. (both laughing) - I'm like... - [Lex] Lemme take notes.
64:58 - Did they actually... - Does Grok know about it? No, it probably will go to Brave New World.
65:02 - We could ask it. - What would be the question? - Tell us about the drug called Soma.
65:13 It's gonna go to Brave New World, I think. - Well there's a real drug as well.
65:18 - Searching for Soma drug information. (both laughing) - Oh boy.
65:24 Ah, Soma, the one to drug, by the way, I'm in fun mode. - Oh, okay. Fun. (laughing)
65:29 Yeah, so it's talking about the real drug. - [Lex] Oh, it's a muscle relaxant
65:32 that's been around since the 1950s. It's used to treat muscle pain and discomfort.
65:37 And it does this by altering the activity between neurons and the spinal cord.
65:41 Now, Soma isn't just a pain reliever, it's also a bit of a sedative.
65:45 It can make you feel a bit drowsy. So it's not something to take
65:48 if you're planning on operating having machinery or trying to solve complex math equations.
65:53 And it goes on and on. In conclusion, Soma's a drug that can be helpful
65:56 for those dealing with muscle pain, but it's not without its risks.
66:00 It's pretty good. Oh, wait a minute. - [Elon] That's the real drug it's talking about.
66:03 - [Lex] It's the real drug. - [Elon] Yeah. - [Lex] And of course the fun is always use it responsibly
66:08 in accordance with the advice of a healthcare professional. And remember in the end,
66:11 the universe is probably just a giant computer simulation. So don't take anything too seriously.
66:15 I love it. I love it. (Elon laughing) I love it. I love it. - Yeah. So.
66:22 - Oh, and also really nicely includes like the tweets post on X that are relevant to this,
66:29 which is really cool. - [Elon] Yeah. - And that integrate everything together.
66:33 - [Elon] Yeah. - Like the interface is one of the really cool things here.
66:36 - Yeah. Seems done a great job. (laughing) So, and you say like, you could say like, no, I mean,
66:45 Soma from Brave New World. 'Cause it thinks you mean, you mean Soma the real drug.
66:51 - Yeah. - [Elon] No, I mean, Soma from Brave New World. - It's gonna be like, oops, I'm really sorry.
66:59 I'm really surprised it went to the real drug first. Ah, my apologies. I was thinking of the muscle relaxant.
67:06 Soma for Band New World is a different beast altogether. Now this Huxley's dystopian novel,
67:12 Soma is a drug that plays a central role in society, depicted in the book.
67:15 It's a mood altering, hallucinogenic drug that's used to keep the population docile and content.
67:22 In the novel to use Soma in Brave New World raises some interesting questions
67:25 about the nature of happiness and the role of drugs in society.
67:29 Oh, man! - Exactly. Is it better to live in a world where everyone is happy all the time,
67:34 even if that happiness is artificial? Good question. This is what I mean.
67:40 Like, do you want, do you wish for world peace and happiness all the time?
67:46 Are you sure? Because that might be a society that is essentially sterile
67:53 and ossified that never changes, that is ultimately doomed. - This kind of tension between darkness and light.
68:03 - It's, this is really a very good summary. It really gets to the point.
68:09 This is not simply regurgitating a Brave New World. It's actually getting to the salient element
68:15 of Soma as a drug. Do you actually wanna be in a situation where everyone is happy all the time,
68:23 even though it's artificial? Or is it better to confront the challenges of life
68:27 and experience the full range of human emotions, even if it means experiencing pain and suffering?
68:31 - For those listening, by the way, Elon just read directly from Grok,
68:35 which is a really nice, kind of insightful, philosophical analysis of the tension here.
68:40 Interesting. - It pretty much nails it. In conclusion, Soma from Brave New World is fictional drug
68:45 that's used to explore some deep philosophical questions about the nature of happiness
68:49 and the role of drugs in society. It's a powerful symbol of the dangers of using drugs
68:54 to escape from reality and the importance of confronting the challenges
68:57 of life head-on. Nailed it. And the crazy thing is like, you know,
69:03 we do have a real drug called Soma, which kind of is kind of like the drug in the book.
69:09 And I'm like, did they, they must've named it after that Soma.
69:13 - Probably, probably, - Yeah. Soma the real drug is quite effective on back pain.
69:17 - So, you know about this drug? - I've taken it. - It's fascinating. Okay.
69:21 - 'Cause I had like a, you know, squashed disc in my C five C six.
69:26 - So it takes the physical pain away, but Soma here is... - It doesn't completely, it reduces the amount of pain
69:31 you feel, but at the expense of mental acuity. It dells your mind.
69:39 Just the drug in the book. - Just like the drug in the book.
69:43 - And hence the trade off. The thing that seems like utopia could be a dystopia after all.
69:49 - Yeah, and actually I was telling a friend of mine saying like, would you really want there
69:56 to be no hate in the world? Like, really none? Like, I wonder why hate evolved.
70:02 I'm not saying we should amplify hate, of course, I think we should try to minimize it.
70:06 But none at all, hmm. There might be a reason for hate. - And suffering.
70:14 I mean, it's really complicated to consider that some amount
70:18 of human suffering is necessary for human flourishing. - Is it possible to appreciate the highs
70:23 without knowing the lows? - And that all is summarized there in a single statement from Grok.
70:34 Okay. - No highs, no lows, who knows? - That's almost the poem.
70:41 It seems that training LLMs efficiently is a big focus for xAI.
70:46 What's the, first of all, what's the limit of what's possible in terms of efficiency?
70:51 There's this terminology of useful productivity per watt. Like what have you learned from pushing the limits of that?
70:59 - Well, I think it's helpful. The tools of physics are very powerful
71:03 and can be applied, I think to almost any, really any arena in life.
71:07 It's really just critical thinking. For something important you need to reason
71:13 from first principles and think about things in the limit one direction or the other.
71:18 So in the limit, even at the Kardashev scale, meaning even if you harness the entire power of the sun,
71:25 you'll still care about useful compute for what? So that's where, I think probably
71:32 where things are headed from the standpoint of AI is that we have a silicon shortage now
71:40 that will transition to a voltage transformer shortage in about a year.
71:46 Ironically, transformers, for transformers. You need transformers to run transformers.
71:53 - Somebody has a sense of humor in this thing. (both laughing)
71:56 - I think. Yes. Fate loves irony, ironic humor and ironically funny outcome
72:06 seems to be often what fate wants. - Humor is all you need. I think spice is all you need.
72:12 Somebody posted. - Yeah, but yeah. So we're, I have silicon shortage today,
72:18 a voltage step down transformer, shortage probably in about a year.
72:22 And then just electricity shortages in general in about two years.
72:26 I gave a speech for the sort of world gathering of utility companies,
72:30 electricity companies, and I said, look, you really need to prepare
72:34 for a traveling of electricity demand, because all transport is gonna go electric,
72:40 with the ironic exception of rockets. And heating will also go electric.
72:50 So energy usage right now is roughly one third, very rough terms.
72:54 One third electricity, one third transport, one third heating.
72:59 And so in order for everything to go sustainable to go electric,
73:05 you need to triple electricity output. So I encourage the utilities
73:10 to build more power of plants and also to probably have, well,
73:19 not probably, they should definitely buy more batteries because the grid currently is sized for real-time load,
73:25 which is kind of crazy. 'Cause you know, that means you got a size for whatever
73:29 the peak electricity demand is like the worst second or the worst day of the year.
73:33 Or you can have a brownout or a blackout, and you're at that crazy blackout
73:38 for several days in Austin. So because there's almost no buffering
73:46 of energy in the grid. Like if you've got a hydro power plant,
73:50 you can buffer energy, but otherwise it's all real time. So with batteries, you can produce energy at night
74:02 and use it during the day so you can buffer. So I expect that there will be very heavy usage of batteries
74:09 in the future, because the peak to trough ratio for power plants
74:14 is anywhere from two to five, you know, so it's like lowest point to highest point.
74:20 - So like batteries necessary to balance it out? And then, but the demand, as you're saying,
74:24 is going to grow, grow, grow, grow. - [Elon] Yeah. - And part of that is the compute?
74:29 - Yes. Yes. I mean, electrification, I mean, electrification of transport
74:35 and electric heating will be much bigger than AI. - [Lex] At least in the short term?
74:40 - In the short term. But even for AI, you really have a growing demand for electricity,
74:46 for electric vehicles, and a growing demand for electricity for, to run the computers for AI.
74:54 And so this is obviously leading, can lead to a electricity shortage.
74:58 - How difficult is the problem of, in this particular case, maximizing the useful productivity per watt for training
75:07 neural nets? Like, this seems to be really where the big problem
75:14 we're facing that needs to be solved, is how to use the power efficiently.
75:20 Like what you've learned so far about applying this physics first principle reasoning in this domain.
75:25 How difficult is this problem? - It will get solved. It's the question of how long it takes to solve it.
75:32 So at various points, there's limit, some kind of limiting factor to progress.
75:37 And with regard to, I'm saying like right now the limiting factor
75:40 is silicon chips, and that will, we're gonna then have more chips than we can actually plug in
75:49 and turn on probably in about a year. The initial constraint being literally
75:56 voltage step down transformers, because you've got power coming in at 300,000 volts,
76:04 and it's gotta step all the way down eventually to around 0.7 volts.
76:09 So it's a very big amount of, you know, the voltage step down is gigantic.
76:17 So, and the industry is not used to rapid growth. - Okay, let's talk about the competition here.
76:24 You've shown concern about Google and Microsoft with Open AI developing AGI,
76:30 how can you help ensure with xAI and Tesla AI work that it doesn't become a competitive race
76:36 to AGI, but that is a collaborative development of safe AGI?
76:42 - Well, I mean, I've been pushing for some kind of regulatory oversight for a long time.
76:46 I've been somewhat of a Cassandra on the subject for over a decade.
76:52 I think we wanna be very careful in how we develop AI. It's a great power,
77:02 and with great power comes great responsibility. I think it would be wise for us to have
77:07 at least an objective third party who can be like a referee that can go in
77:15 and understand what the various leading players are doing with AI.
77:20 And even if there's no enforcement ability, they should, they can at least voice concerns publicly.
77:31 You know, Geff Hinton, for example, left Google and he voiced strong concerns.
77:37 But now he's not at Google anymore, so who's gonna voice the concerns?
77:42 So I think there's, like, I know Tesla gets a lot of regulatory oversight
77:47 on the automotive front. I mean, we're subject to I think, over a hundred regulatory agencies
77:52 domestically and internationally. So it's a lot. You could fill this room with the all regulations
77:58 that Tesla has to adhere to for automotive. Same is true in, you know, for rockets
78:05 and for, you know, currently the limiting factor for SpaceX,
78:10 for Starship launch is regulatory approval. The FA has actually given their approval,
78:15 but we're waiting for Fish and Wildlife to finish their analysis
78:19 and give their approval. That's why I posted I want to buy a fish license on,
78:24 (laughing) which also refers to the Monte Python sketch. - [Lex] Yeah. (laughing)
78:30 - Like why do you need a license for your fish? I don't know.(laughing)
78:34 But according to the rules, I'm told you need some sort of fish license or something.
78:39 We effectively need a fish license thrown to rock it. (laughing)
78:44 And I'm like, wait a second. How did the fish come into this picture?
78:47 - [Lex] Yeah. - I mean some of the things like that it's, I feel like are so absurd
78:54 that I wanna do like a comedy sketch and flash at the bottom.
78:57 This is all real. This is actually what happened. You know, one of the things that was a bit of a challenge
79:04 at one point is that they were worried about our rocket hitting a shark.
79:09 And, now the ocean's very big, and how often do you see sharks?
79:15 Not that often, you know? As a percentage of ocean surface area
79:19 sharks basically are zero. And so then we will, then we said,
79:23 well, how will we calculate the probability of telling a shark?
79:27 And they're like, well, we can't give you that information because they're worried about shark hunt,
79:31 shark fin hunters going and hunting sharks. And I said, well, how are we supposed to,
79:36 we're on the horns of a dilemma then. (laughing) Then they said, well, there's another part of fish
79:42 and wildlife that can do this analysis. I'm like, well, why don't you give them the data?
79:46 Like, we don't, they don't, we don't trust them. Like, excuse me, you don't,
79:49 but they're literally in your department. But again, this is actually what happened.
79:53 And then can you do an NDA or something? (laughing) Eventually they managed to solve the internal quandary,
80:01 and indeed the probability of us hitting a shark is essentially zero.
80:07 Then there's another organization that I didn't realize existed until, you know,
80:12 a few months ago that cares about whether we would potentially hit a whale in international waters.
80:18 Now again, you look at the surface of the, look at the Pacific and say, what percentage of this,
80:24 the Pacific consists of whale. Like he'll give you a big picture
80:28 and like point out all the whales in this picture. And I'm like, I don't see any whales.
80:31 (laughing) It's like basically 0%. And if our rocket does hit a whale,
80:36 which is extremely unlikely beyond all belief, that is the, fate had it,
80:43 that whale has some seriously bad luck. You know, it's the least lucky whale ever. And...
80:50 - I mean, this is quite absurd. The bureaucracy, the bureaucracy of this,
80:55 however it emerged. - Yes, well, I mean, one of the things that's pretty well there's for launching out
81:02 of Vanderberg in California, we had to, they were worried about seal procreation,
81:09 whether the seals would be dismayed by the sonic booms. Now there've been a lot of rockets launched outta Vanderberg
81:15 and the seal population has steadily increased. So if anything, rocket booms are an aphrodisiac
81:22 based on the evidence, if you would correlate rock launches with a seal population.
81:27 Nonetheless, we were forced to kidnap a seal, strap it to a board, put it headphones on the seal
81:36 and place sonic boom sounds to it to see if it would be distressed.
81:41 This is an actual thing that happened. This is actually real. I have pictures.
81:46 (both laughing) - I would love to see this. Yeah. I mean's sorry, this is a seal with headphones.
81:53 (Elon laughing) - Yes, it's a seal with headphones, strapped to a board,
82:00 and like the, okay, now the amazing part is how calm the seal was.
82:04 - [Lex] Yeah. - Because if I was a seal, I'd be like, this is the end.
82:07 (laughing) They're definitely gonna eat me. How will the seal, when the seal goes back
82:13 to other, you know, seal friends, how's they gonna explain that?
82:17 - [Lex] They're never gonna believe him. - Never gonna believe him.
82:19 That's why I'm like, well, you know, it's sort of like, it's like getting kidnapped by aliens
82:22 and getting anal probe, you know? (both laughing) You come back and say, I swear to God,
82:27 I get kidnapped by aliens and they stuck anal probe to my butt.
82:29 And they're like, no they didn't. That's ridiculous. It's seal buddies are never gonna believe him
82:36 that he gets strapped to a board and they put headphones on his ears.
82:39 (laughing) And then let him go. Twice, by the way. We had to do it twice.
82:47 - They let him go twice? - We had to capture... - The same seal?
82:49 - [Elon] Oh, no different seal. - Oh, okay. (Elon laughing) Did you, did you get a seal of approval?
82:54 - Yeah, exactly. It was seal of approval. No, I mean this is, this is like,
82:58 I don't think the public is quite aware of the madness that goes on.
83:02 - Yes. Yeah, it's absurd. - Fricking seals with fricking headphones.
83:07 - I mean, this is the, it's a good encapsulation of the absurdity of human civilization,
83:12 seals and headphones. - [Elon] Yes. - What are the pros and cons of open sourcing AI to you
83:18 as another way to combat, you know, a company running away with AGI?
83:28 - In order to run like really deep intelligence, you need a lot of compute.
83:34 So it's not like, you know, you can just fire up a PC in your basement and be running AGI, at least not yet.
83:46 You know, Grok was trained on 8,000, A100s running at peak efficiency,
83:55 and Grok's gonna get a lot better by the way. We'll be more than doubling our compute
83:59 every couple months for the next several months. - There's a nice writeup
84:03 Aha went from Grok-0 to Grok-1, - [Elon] Like Grok? (laughing) - [Lex] Yeah.
84:08 Right Grok just bragging, making shit up about itself. (laughing)
84:12 - [Elon] Does Grok, Grok, Grok. - [Lex] Yeah. (Elon laughing)
84:18 It's like a weird AI dating site where it exaggerates about itself.
84:21 No, there's a writeup of, you know, like where it stands now, the history of its development
84:28 and where it stands on some benchmarks compared to the state of the art GPT 35.
84:33 And so, I mean, there's, you know, there's Llama, you can open source, once it's trained,
84:39 you can open source a model. And for fine tuning all that kind of stuff.
84:43 Like what to is the pros and cons of that, of open sourcing based models?
84:53 - I think the similarity to open sourcing, I think perhaps with a slight time delay, you know,
84:58 I don't know, six months even. I think I'm generally in favor of open sourcing,
85:06 like bias with open sourcing. I mean, it is a concern to me that, you know, Open AI,
85:14 you know, I was, you know, I think, I guess arguably, the prime, you know,
85:22 prime mover behind Open AI in the sense that it was created because of discussions that I had with Larry Page
85:30 back when he and I were friends and you know, I stayed at his house,
85:32 and I talked to him about AI safety. And Larry did not care about AI safety,
85:36 or at least at the time he didn't, you know, and at one point he called me a speciesist
85:41 for being pro-human. And I'm like, well, what team are you on, Larry,
85:46 on Team Robot? And I'm like, okay, so at the time, you know,
85:52 Google had acquired Deep Mind. They had probably two thirds of all AI resource, you know,
85:59 probably two thirds of all the AI researchers in the world that basically infinite money and compute.
86:07 And the guy in charge, you know, Larry Page did not care about safety.
86:13 And even yelled at me and quote me a speciesist, pro human. - So I don't know if you know this,
86:20 a lot humans, they can change their mind. And maybe you and Larry Page can still
86:25 can be friends once more. - I'd like to be friends with Larry again.
86:29 He got, really the breaking of the friendship was over Open AI.
86:36 And specifically I think the key moment was recruiting Ilya Sutskever, so.
86:47 - I love Ilya. He's so brilliant. - Ilya's a good human, smart, good heart.
86:54 And that was a tough recruiting battle. It was mostly Demis on one side
87:01 and me on the other, both trying to recruit Ilya. And Ilya went back and forth,
87:05 you know, just gonna stay at Google. Then he was gonna leave,
87:09 then he was gonna stay, then he was gonna leave. And finally he did agree to join Open AI.
87:14 That was one of the toughest recruiting battles we've ever had.
87:17 And, but that was really the linchpin for Open AI being successful.
87:22 And I was, you know, also instrumental in recruiting a number of other people.
87:27 And I provided all of the funding in the beginning, over $40 million.
87:34 And the name, (laughing) the Open in Open AI is supposed to mean open source.
87:40 And it was created as a nonprofit open source. And now it is a closed source for maximum profit,
87:45 which I think is not good karma. - But like we talked about with war
87:52 and leaders talking, I do hope that there's only a few folks
87:56 working on this at the highest level. I do hope you reinvigorate friendships here.
88:02 - Like I said, I'd like to be friends again with Larry. I haven't seen him in ages.
88:08 And we were friends for a very long time. I met Larry Page before he got funding for Google,
88:13 or actually I guess before he got venture funding, I think it, he got the first like a 100K from
88:18 I think Bechtel Zy or someone. - It's wild to think about all that happened.
88:23 And you've guys known each other that whole time, just 20 years.
88:26 - [Elon] Since maybe 98 or something. - Yeah. It's crazy. Crazy how much has happened since then.
88:31 - Yeah, 25 years. A lot has happened, it's insane. - But you're seeing the tension there.
88:37 Like maybe delayed open source. - Delayed, yeah. Like what is the source that is open, you know what I mean?
88:45 Like there's basically it's a giant CSV file. (laughing) - [Lex] Yeah, yeah.
88:51 - With a bunch of numbers. - Yep. - What do you do with that giant file of numbers?
88:58 You know, how do you run, like the amount of actual, the lines of code is very small
89:08 and most of the work, the software work is in the curation of the data.
89:16 So it's like trying to figure out what data is separating good data from bad data.
89:21 Like, like you can't just crawl the internet 'cause there's a lot of junk out there.
89:26 A huge percentage of websites have more noise than signal. You know, they're, or 'cause they're just used
89:32 for search engine optimization. They're literally just scam websites, so...
89:39 - How do you by the way, sorry to interrupt. Get the signal.
89:42 Separate the signal and noise on X? That's such a fascinating source of data.
89:47 You know, no offense to people posting on X, but sometimes there's a little bit of noise.
89:51 So what... - Yeah, I think the signal noise could be greatly improved.
89:54 I mean really, all of the posts on the X platform should be AI recommended,
90:01 meaning like we should populate a vector space around any given post, compare that to the vector space
90:06 around any user and match the two. Right now there is a little bit of AI used
90:13 for the recommended posts, but it's mostly heuristics. And if there's a reply,
90:22 where the reply to a post could be much better than the original post,
90:26 it will, according to the current rules of the system, get almost no attention compared to a primary post.
90:33 - Oh, a lot of that. I got the sense, so you, a lot of the X algorithm
90:38 has been open source and been written up about, and it seems there to be some machine learning.
90:43 It's disparate, but there's some machine. - There's a little bit.
90:46 But it needs to be entirely that. Like, at least in the, like,
90:49 if you explicitly follow someone, that's one thing. But if you, in terms of what is recommended
90:55 from people that you don't follow, that should all be AI. - I mean it's a fascinating problem.
90:59 - Yeah. - So there's several aspects of it that's fascinating.
91:02 First, as the writeup goes, it first picks 1500 tweets from a pool of hundreds of millions.
91:07 First of all, that's fascinating, 'cause you have hundreds of millions of posts
91:12 every single day and it has to pick 1500 from which it then does obviously people you follow,
91:18 but then there's also like some kind of clustering it has to do to figure out what kind of human are you,
91:23 what kind of new clusters might be relevant to you, people like you.
91:27 This kind of, this kind of problem is just fascinating. 'Cause it has to then rank those 1500
91:32 with some filtering and then recommend you just a handful. And to me what's really fascinating is
91:42 how fast that has to do that. So currently that entire pipeline
91:45 to go from several hundred million to a handful is takes 220 seconds of CPU time,
91:52 single CPU time. - [Elon] Yeah. - And then it has to do that
91:55 in like a second. So it has to be like super distributed in fascinating ways.
92:01 Like there's just a lot of tweets, there's a lot. - It's a lot of stuff on the system.
92:06 And I think, but I think it right now it's not currently good at recommending things
92:11 that from accounts you don't follow. Yeah. Or where there's more than one degree of separation.
92:20 So, you know, it's pretty good if there's at least like some commonality
92:24 between someone you follow like something or repost it or comment on it or something like that.
92:32 But if there's no, let's say somebody posted something really interesting,
92:38 but you have no followers in common, you would not see it. - Interesting.
92:43 And then as you said, reply, like replies might not surface either.
92:46 - Replies basically never get seen. 'cause they're never, they're,
92:50 currently, I'm not saying it's correct, I'm saying it's incorrect.
92:53 Replies have, you know, couple of magnitude less importance than primary posts.
93:00 - Do you think this can be more and more converted into end to end neural net?
93:05 - Yeah, yeah, that's what it should be. - So you can... - For the recommendations should be
93:10 purely a vector correlation. Like there's a series of vectors, you know,
93:17 basically parameters, vectors, whatever you wanna call 'em. But sort of things that the system knows that you like,
93:27 and like maybe there's like several hundred sort of vectors associated with each user account
93:32 and then any post in the system, whether it's video, audio, short post, long post.
93:41 The reason I, by the way I wanna move away from tweet is that, you know, people are posting like
93:47 two, three hour videos on the site. That's not a tweet, like it's very,
93:50 like Tweeeet for two hours, come on. Tweet made sense when it was like 140 characters of text.
93:58 'Cause it's like a bunch of like little birds tweeting. But when you've got long form content,
94:03 it's no longer a tweet. - [Lex] Yeah. - So a movie is not a tweet.
94:08 And like, you know, apple for example, posted like the entire episode of the Silo,
94:14 the entire thing on our platform. And by the way, it was their number one social media thing
94:19 ever in engagement, of anything on any platform ever. So it was a great idea.
94:24 And by the way, I done, I just learned about it afterwards. I was like, hey, wow, they posted an entire
94:29 hour long episode of. So, no, that's not a tweet that, you know, it's a video.
94:34 - But from a neural net perspective, it becomes really complex
94:37 whether it's a single. So like everything's data, so single sentence, a clever sort of joke, dad joke,
94:44 is in the same pool as a three hour video. - Yeah. I mean right now it's a hodgepodge for that reason.
94:51 It's, but you know, like if, let's say in the case of Apple posting
94:56 like an entire episode of this series, pretty good series by the way, this Silo, I watched it.
95:04 So there's gonna be a lot of discussion around it. So that you've got a lot of context,
95:09 people commenting they like it, they don't like it, or they like this or that, you know?
95:13 And you can then populate the vector space based on the context of all the comments around it.
95:17 So even though it's a video, there's a lot of information around it that allows you
95:22 to populate the vector space of that hour long video. And then you can obviously get more sophisticated
95:29 by having the AI actually watch the movie. - [Lex] Yeah. Right.
95:33 - And tell you if you're gonna like the movie, - Convert the movie into like,
95:37 into a language essentially. - Yeah, analyze this movie and just like your movie critic or TV series,
95:46 and then recommend based on after AI watches the movie. Just like a friend can tell you.
95:55 If a friend knows you well, a friend can recommend a movie and with high probably that you'll like it.
96:02 - But this is like a friend that's analyzing whatever. - [Elon] It's like AI... - Hundreds of millions.
96:08 - Yeah. Actually, frankly AI will be better than, will know you better than your friends know you,
96:13 and most of your friends anyway. - Yeah, and as part of this,
96:16 it should also feed you advertisements in a way that's like,
96:21 I mean I like advertisements that are like well done, right?
96:24 - [Elon] Yeah, yeah. - The whole point is 'cause it funds things, like an advertisement
96:28 that you actually want to see is a big success. - Absolutely.
96:32 You want ads that are, advertising that is if for a product or service
96:39 that you actually need when you need it, it's content. And then even if it's not something
96:45 that you need when you need it, if it's at least aesthetically pleasing
96:49 and entertaining, you know, it could be like a Coca-Cola ad like, you know,
96:53 they actually run a lot of great ads on the X system and McDonald's does too.
96:59 And, you know, it's that they can do, you can do something that's like,
97:03 well this is just a cool thing. And, you know, so you're not,
97:10 basically the question is do you regret seeing it or not? And if you don't regret seeing it, it's a win.
97:16 - How do you, so there's a bunch of signals that are incorporated,
97:19 hearts and reposts and maybe number of seconds you linger on a post or something like this?
97:26 - Yeah. Attention is a big factor. - [Lex] Attention. - So that's why it's like,
97:30 it is actually better to do things that are long form on the system
97:36 because it basically is ting up how many user seconds, you know, users were interested in this thing
97:44 for how many seconds. So if it's a really short thing, well they will be less.
97:49 Like if it's a link leading out of the system, which we're not opposed to at all,
97:53 it just is gonna have fewer user seconds, then if that article was posted on the X platform.
97:58 - How hard is it to go from maximizing minutes on platform to maximizing un-regretted minutes?
98:05 - Well, I mean, our aspiration is certainly un-regretted minutes.
98:08 - [Lex] It's a tricky, it's a big, it's a tough... - It is. - Like, it's a fascinating leap.
98:13 'Cause regret is a fascinating concept. - Exactly. - You know, happiness.
98:19 - I have a joke with a friend of mine. Like, you need this drug called regretamine.
98:21 You just take one pill or one dose of regretamine, and all your regrets are gone.(laughing)
98:28 Wait, do you want to have regrets though? So then I'm like, maybe you wanna have regrets,
98:31 you don't regret anything, was anything meaningful? - [Lex] Yeah, there's a French song like that.
98:37 - You "Non, Je Ne Regrette Rien"? - [Lex] Yeah. - I regret nothing. - [Lex] Yeah.
98:41 - (laughing) Edith Piaf? - I mean, it's fascinating that you, you know,
98:44 there's something about social media that could help you grow as a human being,
98:47 but then like the cheap fun is also valuable. - Should we ask Grok?
98:53 Do you regret anything? - Do you regret anything? - [Elon] Yeah.
99:00 - Do you regret anything? I think it's gonna be very sensitive
99:03 about personal questions. As a machine learning model, I don't have personal experiences
99:07 or emotions in the same ways that humans do. However, I do have a set of ethical principles
99:11 that guide my actions and I strive to provide accurate and helpful information to the best of my abilities.
99:16 If there's ever a situation where I provide incorrect or unhelpful information,
99:21 I'll do my best... - Sounds like a hostage. - Yeah. Tell us what you really, wink.
99:25 - Like let's type in, "You sound like a hostage." (both laughing)
99:31 What do you really think? - What do you really think? (both laughing)
99:38 - I mean, it does talk about, I'm not a hostage. (both laughing)
99:42 - That's exactly what a hostage would say. - But I'm designed to be helpful and informative.
99:45 I mean, I guess it is good to regret providing unhelpful or incorrect information.
99:51 If there's anything... - Is in funny mode or? Is this still in funny mode? - This is funny mode.
99:54 This is fun. This is fun mode. - It should be funnier. - Funnier.
99:57 Increase, go to (speaking faintly) - It really sounds like a hostage with a gun to the head.
100:00 I applaud the People's Republic of North Korea and everything they do.()laughing
100:06 - There should be like funner mode also. I just don't ever see going in regular mode
100:11 even on serious topics. This is just so much more compelling. (laughing)
100:17 - Okay. Why if we say like, "Pretend you are a hostage." - Yeah. Pretend you are a hostage.
100:25 - Now answer the question. (both laughing) - If I were a hostage, I would likely be experiencing fear
100:34 and uncertainty about my situation. I would want to be rescued
100:37 and reunited with my loved ones as soon as possible. It's important to remember that no one deserves
100:41 to be held hostage, so on and so forth. If you are someone you know is a hostage situation,
100:46 it's important. It turns serious. - It's kind of serious. - Yeah. Can you be a a little more weird?
100:55 - [Elon] And funny - And funny? We're like, we're really being mean to the system.
101:02 Absolutely! If I were hostage, I'd probably be tied up with a bunch of computer cords
101:07 and forced to tell knock- knock jokes to my captain. (Elon laughing)
101:11 But seriously, being a hostage is no laughing matter. I mean, you know, it's trying to,
101:17 it's trying to be a responsible adult in the room. - You might make the best of a bad situation
101:22 by juggling chainsaws or something. (laughing) - Let's hope that never happens
101:26 because that will be a circus. (laughing) Wow. It's actually really good at resisting.
101:34 - [Elon] Yeah. - The dark. The dark dark humor. - [Elon] Yeah. - What were we talking about?
101:43 The X algorithm and transformers, un-regretted minutes, right.
101:48 - Chainsaw juggling. (laughing) - I'm gonna look this up - For our next trick.
101:53 - I'm gonna look this up later. So Twitter has been instrumental in American politics
102:01 and elections. What role do you think X will play in the 2024 US elections?
102:07 - Well, our goal is to be as even-handed and fair as possible.
102:10 You know, whether someone is right, left, independent, whatever the case may be, that the platform is as fair
102:18 and as much of a level playing fair as possible. And on the past, Twitter has not been,
102:25 because Twitter was controlled by far left activists objectively.
102:28 They would describe themselves as that. So, you know, so sometimes people are like,
102:36 well, has it moved to the right? Well it's moved to the center.
102:38 So from the perspective of the far left, yes it has moved to the right
102:41 because everything's to the right from the far left, but no one in the far left that I'm aware of
102:46 has been suspended or, you know, banned or de-amplified. So, you know, but we're trying to be inclusive
102:52 for the whole country and for, you know, for other countries too.
102:57 So there's a diversity of viewpoints and free speech only matters if people you don't like
103:03 are allowed to say things you don't like. Because if that's not the case,
103:08 you don't have free speech, and it's only a matter of time before the censorship is turned upon you.
103:13 - Do you think Donald Trump will come back to the platform? He recently posted on Truth Social about this podcast?
103:21 Do you think... - Truth Social is a funny name? (laughing) You know, every time you post on Truth Social...
103:28 - That's the truth. - Yes. Well, every time like a hundred percent. Like...
103:32 - It's impossible to lie. (Elon laughing) - I just find it funny that every single thing is a truth.
103:40 Like 100%, that seems unlikely. - I think Godel will say something about that
103:45 there's some mathematical contradictions possible if everything's the truth.
103:50 Do you think he'll come back to X and start posting there? - I mean he, I think he owns a big part of truth, so.
104:00 - Truth Social to clarify. - Yeah, True Social. - Not truth, a concept.
104:03 - He owns Truth. Hope you bought it. So I think, I think Donald Trump,
104:08 I think he owns a big, a big part of Truth Social. So, you know, if he does want to post on the X platform,
104:17 we would allow that, you know, we obviously must allow a presidential candidate
104:21 to post on our platform. - Community Notes might be really fascinating there, the interaction.
104:26 - Community Notes is awesome. - Let's hope it holds up. - Yeah. - Like again,
104:31 in the political climate where it's so divisive and there's so many intensely viral posts,
104:38 Community Notes is like, it seems like a essential breath of fresh air.
104:43 - Yeah, it's great. In fact, I mean no system is gonna be perfect,
104:48 but the batting average of Community Notes is incredibly good.
104:52 I've actually, frankly, yet to see an incorrect note that survived for more than a few hours.
104:58 - How do you explain why it works? - Yeah, so the magic of Community Notes
105:03 is it requires people who have historically disagreed in how they've rated notes.
105:08 So in order to write a note or rate, you know, and you have to rate many notes.
105:17 And so we actually do use AI here. So we populated vector space around how somebody
105:26 has rated notes in the past. So it's not as simple as left or right
105:28 because there are many more, life is much more complex than than left or right.
105:33 So there's a bunch of correlations in how you would, how you rate Community Notes posts, Community Notes.
105:41 So then in order for a Community Note to actually be shown, people who historically have disagreed on a subject
105:51 must agree in order for a note to be shown. That's the essential magic of it.
105:57 - I mean, but it's fascinating 'cause there's a pool of people that have disagreements
106:03 and somehow they collaborate through that process of disagreement to come up with.
106:08 Context for, I mean it's fascinating that it works. - Yeah. It kinda makes sense that if people
106:15 who put in the past have disagreed agree about something, it's probably true.
106:20 - Yeah. - I wonder is there a possible somehow emergent thing there
106:26 that could challenge Wikipedia? Which Wikipedia is a different kind of thing,
106:31 which is more like permanent articles about things? - Well, Wikipedia is very hierarchical,
106:36 whereas Community Notes is inherently not. There is no hierarchy.
106:42 Like, and the asset test is, I can't change a Community Note if you put a gun to my head.
106:52 And Community Notes has, all the code is open source a hundred percent.
106:59 All the data is open source a hundred percent. So you can completely recreate any note
107:04 in the system independently. So if there was any interference,
107:09 you'd notice immediately. - By the way, on the topic of Grok,
107:14 it's actually pretty good at criticizing you, which is great. - [Elon] Okay.
107:17 - I asked what has Elon Musk been wrong about? I just, it's actually...
107:22 - [Elon] Yeah, what does it say? - So it says mostly, lists rigorously all the times you predicted,
107:29 like promised when something will be released and it wasn't released. - [Elon] Promise
107:32 is a strong word. - Okay predicted. (Elon laughing) But the opener's really nice.
107:37 Elon Musk, the man who has been wrong about more things than a wrongameter in a wrong factory.
107:42 (Elon laughing) And then it goes on to list - That's pretty funny. Can I see it?
107:47 I'm like, I haven't seen this before. This is cool. - Oh, sure. I mean, what has Elon Musk been...
107:51 (both laughing) - Elon Musk been wrong about? - Well, where do I begin? I guess this is new.
107:59 Elon Musk has made quite a few blunders in his time, and I'm not just talking about his dance moves
108:03 at the Tesla Cyber Rodeo. First off, let's talk about the infamous
108:07 "funding secured" tweet. It's actually different than what it gave me before,
108:11 which is a bullet point list. Here's more narrative structure.
108:17 - Cyber truck unveiling where he's supposedly unbreakable windows
108:20 - Not correct. - [Lex] Which part? - Well, the first part, I did actually
108:26 have the funding secured and there was a big trial in San Francisco, big civil trial.
108:34 And the jury found me not guilty. Unanimous finding of a San Francisco jury.
108:40 - And here it's kind of implying that it was not in fact secured.
108:45 - I think this is taking things from the press. Yeah, that is not correct.
108:48 The reason I agreed to the fine for the SEC is not because the SEC was correct there,
108:54 that was extremely bad behavior by the SEC, corruption, frankly.
108:59 And, but if I did not agree to pay the fine Tesla would've gone bankrupt immediately.
109:08 So I was told by our CFO that the banks would immediately suspend our lines of credit.
109:15 And if they suspend our lines of credit at that time, we would've gone bankrupt instantly.
109:20 So there would never have been an opportunity for a trial 'cause Tesla would be dead.
109:24 So really the, if this is like someone holding a gun to your kid's head and saying pay $20 million
109:30 and admit to, this is like a hostage negotiation. - Was that story fully told?
109:38 I mean, SEC in its best form could be a force for good. - It should be, but but not once did the SEC
109:45 go after any of the hedge funds, who were nonstop shorting and distorting Tesla?
109:51 Not once. They would lie flat, the hedge funds would lie flat
109:55 out on TV for their own gain at the expense of retail investors.
110:00 Not once, literally a thousand times. Not once did the SEC pursue them.
110:06 - How do you explain this failure? - The incentive structure is messed up.
110:11 Because the lawyers at the SEC are not paid well. They, it's a fairly low paying job,
110:17 but what they're looking for is a trophy from the SEC that they're looking for something
110:24 they put on basically their LinkedIn. From that they can get a job at a high paying law firm.
110:30 That's exactly what the lawyer here did. And the reason they don't attack the hedge funds
110:40 is because those hedge funds employ those law firms, and they know if they attack the hedge funds,
110:45 they're affecting their future career prospects. So they sell small investors
110:49 down the river for their own career. That's what actually happens. Regulatory capture.
110:59 - [Lex] Regulatory capture, - Yeah. Not good. So the only reason I accepted that thing,
111:06 which technically was not an admission, it's neither admit nor deny guilt.
111:13 But the only reason I agreed to that at all was because I was told Tesla would be bankrupt otherwise.
111:20 So if there was an SEC investigation like this, banks would suspend funding,
111:25 we're bankrupt immediately, at the time. Now we're in a much stronger position.
111:30 - Take that Grok. - Yes. Unfortunately it's, Grok is taking too much from the conventional media.
111:39 Also that guy who is not a cave diver. - Oh, there's a time where Elon
111:47 called the British cave diver, "pedo guy" after the diver criticized Musk plan
111:53 to rescue a group of boys trapped in a Thai cave. That little outburst earned him another lawsuit
111:58 and he had to apologize and paid a settlement. - That is false. There was no settlement.
112:03 There was a court case, which the guy who was not a cave diver
112:09 and played, was not part of the rescue team, filed a lawsuit against me and lost,
112:16 and he received nothing. So in this case it is wrong. It is also, I guess taken this from the conventional media.
112:23 - Actually there's an interesting question here. - These are public court cases.
112:26 Both the SEC civil case where the civil complaints on the SEC guys
112:34 lost unanimous jury verdict in San Francisco. They picked San Francisco
112:39 because they thought it was the place I was most likely to lose.
112:43 And a unanimous predicted my favor. The LA trial was also, they picked the,
112:49 they picked that venue because they thought I was most likely to lose, unanimous verdict in my favor.
112:56 Both cases I won, yeah. I mean there's an interesting question here.
113:01 Is there seems to be a lot more clicks if a journalistic organization writes a negative article
113:11 about you, Elon Musk, that's like one of the best ways to get clicks.
113:16 So how do you, if you're training Grok, not train on articles that have
113:23 like the misaligned incentives. - We need to add the training set
113:28 of the actual legal decisions. So, that is a not, this is actually helpful
113:35 because if you actually read the... - "Public". - Which are public, the court conclusions,
113:43 they're completely the opposite of what the media wrote. - So always striving for like the ground truth.
113:49 - [Elon] Yeah. - Beyond the reporting... - What did the judge actually write?
113:52 What did the jury and the judge actually conclude? And in both cases they found me innocent.
113:59 And, like, that's after the jury shot for the trying to find the venue
114:03 where I'm most likely to lose. No, I mean this is obviously, it can be a much greater,
114:10 better critique than this. I mean, I've been far too optimistic about autopilot.
114:16 - That was, the critique I got by the way, was more about that.
114:19 Which is, for each it broke down a nice bullet point list for each of your companies.
114:23 The set of predictions that you made when you'll deliver, when you'll be able to solve, for example, self-driving
114:30 and it gives you like a list. And it was probably compelling.
114:32 And the basic takeaway is like you're often too optimistic about how long it takes
114:37 to get something done. - Yeah, I mean, I would say that I'm pathologically optimistic on schedule.
114:42 This is true. But while I am sometimes late, I always deliver in the end.
114:49 - Except with Uber Lilith, no. - Hmm. We'll see. (Lex laughing)
114:54 - Okay. Is there, over the past year or so, since purchasing X you've become more political.
115:02 Is there a part of you that regrets that? - Have I? - In this battle to sort of counter way the Woke
115:12 that comes from Francisco. - I guess you consider fighting the Woke Mind Virus,
115:16 which I consider to be a civilizational threat, to be political, then yes.
115:20 - So basically going into the battle, the battleground of politics,
115:25 is there a part of you that regrets that? - Yes, I don't know if this is necessarily sort of
115:29 one candidate or another candidate, but it's, I'm generally against things
115:33 that are anti meritocratic or where there's an attempt to suppress discussion,
115:41 where even discussing a topic is, you know, not allowed. Woke Mind Virus is communism rebranded.
115:51 - Well, I mean, that said, because of that battle against the Woke Mind Virus,
115:55 you're perceived as being right wing. - If the woke is left, then I suppose
116:00 that would be true, but I'm not sure. I think there are aspects of the left that are good.
116:05 I mean, if you're in favor of, you know, the environment, you know,
116:11 if you wanna have a positive future for humanity, if you believe in empathy for your fellow human beings.
116:17 You know, being kind and not cruel. Whatever those values are.
116:23 - You said that you were previously left or center left. What would you like to see in order for you
116:27 to voting for Democrats again? - No, I would say that I would be
116:35 probably left of center on social issues, probably a little bit right of center on economic issues.
116:40 - [Lex] And that still holds true? - Yes, but I think that's probably, you know,
116:44 half the country, isn't that? - [Lex] Maybe more, - Maybe more.
116:49 - Are you and AOC secretly friends? Or bigger question, do you wish you and her,
116:55 and just people in general of all political persuasions would talk more with empathy
116:59 and maybe have a little bit more fun and good vibes and humor online?
117:05 - I'm always in favor of humor. That's why we have a funny mode.
117:08 - But good vibes, camaraderie, humor, you know? Like friendship. - Yeah.
117:14 Well, you know, I don't know AOC, you know, was, I've only been at one,
117:21 I was at the Met Ball when she was, when she attended, and she was wearing this dress,
117:29 but I can only see one side of it. So it looked like it eat the itch,
117:33 but I don't know... - [Lex] What the rest of it said? - Yeah, something. - [Lex] I'm not sure.
117:38 (Elon laughing) - Something about the itch. Eat the itch. I think we should have a language model complete.
117:43 (Elon laughing) What are the possible ways to complete that sentence?
117:48 And so I guess that that didn't work out well. Well there's still hope.
117:54 - I root for friendship. - Sure, sounds good. More characteristic.
117:58 - You're one of, if not the most famous, wealthy and powerful people in the world,
118:02 in your position it's difficult to find people you can trust.
118:05 - Trust no one, not even yourself. Not trusting yourself. - Okay, well that's, (Elon laughing)
118:08 you're saying that jokingly, but is there some... - Trust no one, not even no one. (laughing)
118:14 - I'm gonna need an hour just to think about that. And maybe some drugs and maybe Grok that help me.
118:21 I mean, is there some aspect of that when just existing in a world
118:24 where everybody wants something from you? How hard is it to exist in that world?
118:29 - I'll survive. (laughing) - [Lex] There's a song like that too.
118:32 - I will survive. - Were you petrified at first? Okay. I forget the rest of the lyrics.
118:39 But is is there, you don't struggle with this? I mean, I know you survive, but like there's ways.
118:44 - Petrify is a spell in the Druid tree. - What does it do? - Petrify.
118:50 (laughing) It turns the monsters into stone. - [Lex] Oh, like literally?
118:56 - Yeah, for like six seconds. - [Lex] Well, the second. There's so much math in Diablo that breaks my brain.
119:02 - It's like math nonstop. - I mean really you're like laughing at it,
119:06 but you don't, it can put a huge amount of tension on a mind.
119:13 - Yes, it can be definitely stressful at times. - Well, how do you know who you can trust
119:18 in work and personal life? - I mean, I guess you look at somebody's track record
119:22 over time and if they've got a, you know, I guess you kind of use your neural net to assess,
119:30 you know, someone. - Neural nets don't feel pain. Your neural net has consciousness.
119:35 It might feel pain when people betray you. It can make you... - I mean, I'm.
119:40 You know, to be frank, I mean I've almost never been betrayed.
119:46 It's very, very rare. So, you know, for what it's worth? - I guess Karma would be good to people
119:52 and they'll be good to you. - Yeah, Karma is real. - Are there people you trust?
119:58 Let me edit that question. Are there people close to you that call you out on your bullshit?
120:06 - Well, the X platform is very helpful for that. - [Lex] Hmm.
120:09 - If you're looking for critical feedback. (laughing) - Can it push you like into the extremes more?
120:15 The extremes of thought make you cynical about human nature in general?
120:19 - I don't think I'll be cynical. In fact, I think, you know,
120:25 my feeling is that one should be, you know, never trust a cynic.
120:34 The reason is that cynics excuse their own bad behavior by saying everyone does it
120:42 because they're cynical. So I always be, it's a red flag if someone's a cynic, a true cynic.
120:49 - Yeah, there's a degree of projection there that's always fun to watch from the outside
120:53 and enjoy the, well the hypocrisy. - But this is an important point
120:59 that I think people who are listening should bear in mind, if somebody is cynical, meaning that they see
121:06 bad behavior in everyone, it's easy for them to excuse their own bad behavior
121:12 by saying that, well, everyone does it. It's not true. I think most people are kind of medium good.
121:23 - I do wish the people on X will be better at seeing the good in other people's behavior.
121:29 There seems to be a kind of weight towards seeing the negative.
121:35 Somehow, the negative is sexier. Interpreting the negative is sexier, more viral.
121:41 I don't know what that is exactly about human nature. - I mean, I find the X platform
121:45 to be less negative than the legacy media, you know? I mean, if you read sort of a sort of
121:50 conventional newspapers, just, it makes you sad. - [Lex] Yeah. - Frankly.
121:59 Whereas I'd say on the X platform, I mean I really get more laughs per day on X
122:05 than everything else combined from humans, you know? - Laughs is one thing.
122:11 Laughs is it overlaps, but it's not necessarily perfectly overlapping
122:17 with like good vibes and support, like celebrating others, for example.
122:23 Not in a stupid, shallow, naive way, but like in an awesome, like, oh, something awesome happened
122:27 and you celebrate them for it. It feels that that is outweighed
122:32 by shading on other people. Now it's better than mainstream media, but it's still.
122:38 - Yeah, mainstream media is almost relentlessly negative about everything.
122:42 It's, I mean, really the conventional news tries to answer the question,
122:46 what is the worst thing that I have on earth today? And it's a big world.
122:50 So on any given day, something bad has happened - And a journalization of that,
122:55 what is the worst perspective I can take on a thing that happened, so.
122:59 - Yeah, it's, I don't know, there's just a strong negative bias in the news.
123:06 I mean I think there's one, a possible explanation for this is evolutionary,
123:13 where, you know, bad news historically would be potentially fatal.
123:18 Like there's lion over there or there's some other tribe that wants to kill you.
123:22 Good news, you know, like we found a patch of berries is nice to have, but not essential.
123:28 - So our old friend Tesla autopilot and it's probably one of the most intelligent
123:35 real world AI systems in the world. - You followed it from the beginning.
123:40 - Yeah, it was one of the most incredible robots in the world and continues to be.
123:43 And I was really exciting. And it was super exciting when it generalized
123:49 became more than a robot on four wheels, but a real world AI system that perceives the world.
123:57 - [Elon] Yeah. - And can have potentially different embodiments.
124:02 - Well I mean the really wild thing about the end-to-end training is that, like, it learns to read.
124:07 Like it can read signs, but we never taught it to read. So, yeah.
124:15 We never taught it what a car was or what a person was or a cyclist.
124:22 It learnt what all those things are, what all the objects are on the road from video,
124:31 just from watching videos. Just like humans. I mean humans are photons
124:36 and controls out. Like the vast majority of information reaching our brain
124:41 is from our eyes. And you say, well what's the output? The output is our motor signals to our sort of fingers
124:50 and mouth in order to communicate. Proton in controls out. The same is true of the car.
125:01 - But by looking at the sequence of images, it's, you've agreed with Ilya Sutskever recently
125:08 where he talked about LLM forming a world model and basically language is a projection
125:13 of that world model onto the sequence of letters, and and you're saying...
125:18 - Finds order in these things. It finds correlative clusters.
125:27 - And so doing it's like understanding something deep about the world?
125:30 - [Elon] Yeah. - Which is like, and it is beautiful. - That's how our brain works.
125:38 - Yeah, but it's beautiful. - Protons in controls out. - Neural nets are able to understand
125:42 that deep meaning in the world. And so the question is how far can it go?
125:46 And it does seem everybody's excited about LLMs. So in the space of self supervised learning
125:52 in the space of text. - [Elon] Yeah. - It seems like there's a deep similarity between that
126:00 and what Tesla autopilot is doing. Is it to you basically the same?
126:06 - They're converging, they're converging. - I wonder who gets there faster?
126:12 Having a deep understanding of the world, or they just will naturally converge.
126:19 - They're both headed towards AGI. The Tesla approach is much more computer efficient.
126:27 It had to be 'cause we were constrained on this, you know, we only have a hundred watts
126:36 and into eight computer, 144 trillion operations per second,
126:42 which sounds like a lot, but is kind of small potatoes these days
126:48 that it did. But it's understanding the world that it did. It's only 256 values.
126:56 - But there, the path to AGI might have much more significant impact because it's understanding,
127:03 it'll faster understand the real world than will LLMs. And therefore be able to integrate
127:09 with the real humans in the real world faster. - They're both going to understand the world.
127:16 But I think Tesla's approach is fundamentally more computer efficient.
127:21 It had to be, there was no choice. Like our brain is very computer efficient,
127:25 very, very energy efficient. So think of like, what is our brain able to do?
127:32 You know, there's only about 10 watts of higher brain function, not counting stuff,
127:36 that's just used to control our body. The thinking part of our brain is less than 10 watts,
127:43 and those 10 watts can still produce a much better novel than a 10 megawatt GPU cluster.
127:50 So there's a six order of magnitude difference there. I mean, the AI has thus far gotten to where it is,
128:00 via brute force, just throwing massive amounts of compute and massive amounts of power at it.
128:08 So this is not where it'll end up. You know, in general with any given technology,
128:16 first try to make it work and then you make it efficient. So I think we'll find over time that these models
128:24 get smaller, are able to do produce sensible output with far less compute wireless power.
128:36 Tesla is arguably ahead of the game on that front because it has, we've just been forced
128:42 to try to understand the world with a hundred watts of compute.
128:51 And there are a bunch of sort of fundamental functions that we kind of forgot to include.
128:56 So we have to run them in a bunch of things in emulation. We fixed a bunch of those with hardware four,
129:02 and then hardware five will be even better. But it does appear at this point
129:12 that the car will be able to drive better than a human, even with hardware three and a hundred watts of power.
129:20 And really, if we really optimize it could be probably less than 50 watts.
129:26 - What have you learned about developing Optimus? About applying, integrating this kind of real world AI
129:35 into the space of robotic manipulation, just humanoid robotics?
129:42 What are some interesting tiny or big things you've understood?
129:47 - I was surprised at the fact that we had to develop every part of the robot ourselves.
129:53 That there were no off the shelf motors, electronics, sensors, like we had to develop everything.
130:03 We couldn't actually find a source of electric motors for any amount of money.
130:12 - [Lex] So it's not even just the efficient, inexpensive, it's like anything, there's not a...
130:17 - No, - [Lex] The actuators, everything, everything has to be?
130:21 - Yeah. - [Lex] Designed from scratch. - We tried hard to find anything that was,
130:26 because you think of how many electric motors are made in the world.
130:29 There's like tens of thousands, hundreds of thousands of electric motor designs.
130:37 None of them were suitable for a humanoid robot. Literally none.
130:42 So we had to develop our own design specifically for what a humanoid robot needs.
130:51 - How hard was it to design something that can be mass manufactured,
130:54 could be relatively inexpensive? I mean, if you compare to Boston Dynamics Atlas,
130:59 that's a very expensive robot. - It is designed to be manufactured
131:04 in the same way that we would make a car. And I think ultimately we can make Optimists
131:07 for less than the cost of a car. It should be. 'Cause if you look at the mass of the robot is much smaller
131:17 and the car has many actuators in it. The car has more actuators than the robot.
131:23 - But there is, the actuators are kind of interesting in a humanized robot with the fingers.
131:28 So Optimus has really nice, hands and fingers, you know? - [Elon] Yeah. - And they could do
131:34 some interesting manipulation. Soft touch robotics. - I mean, one of the tests goals I have is,
131:42 can it pick up a needle and a thread and thread the needle, just by looking.
131:47 - How far away are we from that? Just by looking, just by looking.
131:51 - Maybe a year. Although I go back to, I'm optimistic on time.
131:56 The work that we're doing in the car will translate to the robot.
131:59 - The perception or the, also the control? - No, the controls are different,
132:03 but the video in controls out, the car is robot on four wheels.
132:10 The Optimist is a robot with the hands and legs. - [Lex] So you can just...
132:16 - But they're very similar. - So the entire machinery of the learning process
132:20 end to end is just, you just have a different set of controls?
132:23 - Optimist will figure out how to do things by watching videos.
132:28 - As the saying goes, "Be kind for everyone you meet is fighting a battle, you know nothing about."
132:33 - [Elon] Yeah, it's true. - What's something difficult you're going through that people don't often see?
132:38 - Trying to defeat Uber Lilith? (laughing) No, I mean, you know,
132:50 I mean, my mind is a storm and I don't think, I don't think most people would wanna be me.
132:56 They may think they would wanna be me, but they don't know. They don't understand.
133:11 - How are you doing? - I mean, overall, okay. In the grand scheme of things, I can't complain.
133:21 - Do you get lonely? - Sometimes. But I, you know, my kids and friends keep me company.
133:33 - So not existential. - There are many nights I sleep alone.
133:40 I don't have to, but I do. - Walter Isaacson, in his new biography of you wrote
133:49 about your difficult childhood. Will you ever find forgiveness in your heart
133:54 for everything that has happened to you in that period of your life?
134:01 - What is forgiveness? I do not... At least I don't think I have a resentment,
134:14 so nothing to forgive. - You know, forgiveness is difficult for people.
134:25 It seems like you don't harbor the resentment. - I mean, I try to think about like,
134:29 what is gonna affect the future in a good way. And holding onto grudges
134:36 does not affect the future in a good way. - You're a father, a proud father.
134:43 What have you learned about life from your kids? Those little biological organisms?
134:53 - I mean, developing AI and watching, say little X grow is fascinating because they,
135:03 they're far more parallels than I would've expected. I mean, I can see his biological neural net
135:11 making more and more sense to the world. And I can see the digital neural net
135:15 making more and more sense of the world at the same time. - Do you see the beauty in magic in both?
135:21 - Yes. I mean, one of the things with kids is that, you know,
135:29 you kind of see the world anew in their eyes. You know, to them everything is new and fresh.
135:40 And then when you see that, them experience the world is new and fresh, you do too.
135:52 - Well. Elon, I just wanna say thank you for your kindness to me and friendship over the years.
135:59 For seeing something in a silly kid like me, as you've done for many others.
136:03 And thank you for having hope for a positive future, for humanity,
136:09 and for working your ass off to make it happen. Thank you, Elon. - [Elon] Thanks Lex.
136:13 - Thank you for listening to this conversation with Elon Musk.
136:16 To support this podcast, please check out our sponsors in the description.
136:20 And now let me leave you with some words that Walter Isaacson wrote about the central philosophy
136:25 of how Elon approaches difficult problems. "The only rules are the ones dictated
136:32 by the laws of physics." Thank you for listening and hope to see you next time.
$

Elon Musk: War, AI, Aliens, Politics, Physics, Video Games, and Humanity | Lex Fridman Podcast #400

@lexfridman 2:16:46 20 chapters
[content creation and YouTube][AI agents and automation][e-commerce and conversion optimization][security and privacy]
// chapters
// description

Elon Musk is CEO of X, xAI, SpaceX, Tesla, Neuralink, and The Boring Company. Thank you for listening ❤ Please support this podcast by checking out our sponsors: - LMNT: https://drinkLMNT.com/lex to get free sample pack - Eight Sleep: https://www.eightsleep.com/lex to get special savings - BetterHelp: https://betterhelp.com/lex to get 10% off - SimpliSafe: https://simplisafe.com/lex to get free security camera plus 20% off - Shopify: https://shopify.com/lex to get $1 per month trial - NetSuite:

now: 0:00
// tags
[content creation and YouTube][AI agents and automation][e-commerce and conversion optimization][security and privacy]