Latest / Elon Musk Thinking / Elon Musk Heated Interview with ChatGPT CEO Sam Altman

Transcript

Elon Musk Heated Interview with ChatGPT CEO Sam Altman · Elon Musk Thinking

• 0:00
  Today we have Elon Musk Elon, thank you for joining us, Thanks for
• 0:03
  having me. So we want to spend the time today talking about your view
• 0:08
  of the future and what people should work on. So to start off,
• 0:11
  could you tell us you famously said when you were younger, there were five
• 0:14
  problems that you thought were most important for you to work on. If you
• 0:18
  were twenty two today, what would the five problems that you would think about
• 0:21
  working on. B Well, first of all, I think, um,
• 0:25
  if somebody is doing something that is useful to the rest of society, I
• 0:29
  think that's a good thing. Like it doesn't have to change the world.
• 0:32
  Like you know, Um, if you're doing something that has high value to
• 0:36
  people, and frankly, even if it's something if it's like, um,
• 0:40
  just a little game um, or you know, the some improvement in photo
• 0:47
  sharing or something, if if does a small amount of good for a large
• 0:51
  number of people, UM, that's I mean, I think that's that's fine.
• 0:55
  Like stuff doesn't need to be changed the world just to be good.
• 0:58
  UM. But you know, in terms of things that I think are most
• 1:03
  likely to affect the future of humanity, I think um ai is probably the
• 1:07
  single biggest item in the near term that's likely effect humanity. So it's very
• 1:12
  important that we have the advent of AI UH in a good way. That
• 1:19
  that is something that UM, if you if you could look into the crystal
• 1:23
  ball and see the future, you would like you would like that outcome,
• 1:26
  UM, because it is something that could go UM could go wrong UM,
• 1:33
  as we've talked about many times UM, and so we really need to make
• 1:37
  sure it goes right. UM. That's that's I think AI work working on
• 1:42
  AI and making sure it's great future. That's that's the most important thing I
• 1:46
  think right now, UM the most pressing item. So UH. Then UM,
• 1:52
  I would say anything into do with with genetics. UM, if you
• 1:55
  can actually solve UM genetic diseases, if you can prevent dementia or Alzheimer's or
• 2:04
  something like that, that with genetic reprogramming, that would be wonderful. So
• 2:07
  I think there's genetics. It might be the sort of second most important item.
• 2:15
  I think, UM having a high band with interface to the brain like
• 2:21
  UM, we're currently bandwidth limited. We have a digital tertiary self in the
• 2:25
  form of our email capabilities of computers, phones, applications were effectively superhuman,
• 2:31
  but we are extremely bad with constraint in that interface between the cortex and you're
• 2:37
  sort of that that tertiary digital form of yourself, and helping solve that bandwidth
• 2:43
  constraint would would be I think very important for the future as well. So
• 2:47
  one of the I think most common questions I hear young people, ambitious young
• 2:53
  people ask, is I want to be the next Elon Musk. How do
• 2:54
  I do that? Um? Obviously the next Elon Musk will work on different
• 3:00
  things than than you did. But what have you done or what did you
• 3:02
  do when you were younger that you think sort of set you up to have
• 3:07
  a big impact. Well, I think that's well, I should say that
• 3:09
  I do not expect to be involved in all these things. So the the
• 3:15
  the five things that I thought about the time in college, so quite a
• 3:17
  long time ago, twenty five years ago, UM, you know, being
• 3:23
  you know, making life multiplanetary UM, slttering, the transition to sustainable energy
• 3:30
  UM, the Internet broadly speaking UM, and then genetics and AI. I
• 3:37
  think I didn't expect to be involved in in all of those things. I
• 3:40
  actually, at the time in college I sort of thought m helping with electrification
• 3:46
  of cars was how we sought out and that's uh, that's actually what I
• 3:51
  worked on as an intern was an advanced ultra capacitors with to see if they
• 3:55
  would be a breakthrough relative to batteries for energy storage and in cars. And
• 4:00
  then when I came out to go to Stamford, UM, that's what I
• 4:03
  was going to be doing my grass studies on is as it was, working
• 4:06
  on advanced energy storage technologies for electric cars. And then I put that on
• 4:12
  hold to start an internet company in ninety five because um, there does seem
• 4:19
  to be like a time for particular technologies when they're at a steep point in
• 4:25
  the inflection curve. And and I didn't want to you know, do a
• 4:30
  PhD at Stamford and then and what should all happen? Um? And then
• 4:33
  and I wasn't entirely certain that the technology I'd be working on would actually succeed.
• 4:39
  Um, like you can get you can get a you know, doctrine
• 4:42
  on many things that ultimately are not do not have a practical bearing on the
• 4:46
  world. UM. And I wanted to, you know, just I really
• 4:49
  was just trying to be useful. That's the optimization. It was like,
• 4:54
  what what what can I do that would actually be useful. Do you think
• 4:57
  people that want to be useful to day should get PhDs? UM? Mostly
• 5:01
  not so what is the best way? Yes, but mostly not. UM.
• 5:08
  How should someone figure out how they can be most useful whatever this thing
• 5:12
  is that you're trying to create, what would what would be the UM utility
• 5:16
  delta compared to the current state of the art times how many people it would
• 5:20
  affect. So that's why I think UM. Having something that has a that
• 5:25
  that has a makes makes a big difference but affects a sort of small to
• 5:29
  moderate number of people is great, as is something that makes even a small
• 5:31
  difference but but affects a vast number of people. Like the area yeah,
• 5:35
  you know under the curve, Yeah exactly, the area under the curve is
• 5:40
  would actually be roughly similar for those two things. So it's actually really about
• 5:45
  uh yeah, just trying to be useful and matter when you're trying to estimate
• 5:49
  probability of success. So you say this thing will be really useful area under
• 5:54
  the curve. I guess to use the example of SpaceX. When you made
• 5:58
  the go decision that you're actually going to do that was kind of a very
• 6:00
  crazy thing at the time, very crazy for sure. Yeah, well I'm
• 6:03
  not sure about saying that, UM, but I kind of agree. I
• 6:06
  agreed with them that it was quite crazy crazy if m If the objective was
• 6:12
  to achieve the best risk adjusted return UM, starting a rock company is insane.
• 6:19
  But that was not that was not my objective. I had some of
• 6:23
  the come to the conclusion that if something didn't happen to improve rock technology would
• 6:28
  be stuck on Earth forever. Um And and the big aerospace companies had just
• 6:33
  had no interest in radical innovation. UM. All they wanted to do was
• 6:36
  try to make their old technology slightly better every year, and in fact sometimes
• 6:44
  we would actually get worse. Um And, particularly in rockets is pretty bad.
• 6:47
  Like the in sixty nine were able to go to the Moon with a
• 6:51
  Saturn five, and then the Space Shuttle could only take people to lowth orbit,
• 6:55
  and then the Space Shuttle retired. I mean that that trend is basically
• 6:58
  trends to zero. If you also I think technology just automatically gets better over
• 7:02
  a year, but it actually doesn't. It only gets better if smart people
• 7:06
  work work like crazy to make it better. That's how any technology actually gets
• 7:11
  better, and by itself. Technology if people don't work in it actually will
• 7:16
  decline. Um. I mean you can look look at the history of civilizations,
• 7:20
  many civilizations, and look at say, um ancient Egypt. Were they
• 7:25
  able to build these incredible permids, and then they basically forgot how to build
• 7:29
  permits, and and then even in hieroglyphics, they've forgot how to read hier
• 7:34
  hieroglyphics. Or we look at Rome and how they're able to to build these
• 7:38
  incredible roadways and aqueducts and indoor plumbing, and they've forgot how to do all
• 7:42
  of those things. Um. And there are many such examples in history.
• 7:47
  UM. So I think she always bear in mind that you know an entropy
• 7:56
  is not on your side. Ye. One thing I really like about you
• 8:01
  is you are unusually fearless and willing to go in the face of other people
• 8:05
  telling you something that's crazy. And I know a lot of pretty crazy people.
• 8:07
  You still stand out. Where does that come from? Or how do
• 8:09
  you think about making a decision when everyone tells you this is a crazy idea?
• 8:13
  Or where do you get the internal strength to do that? Well,
• 8:16
  that's all I'd say. I actually think I think I fear feel fear quite
• 8:20
  strongly, so as one as though I just have the absence of fear.
• 8:26
  I feel it quite strongly. But there it's times when something is important enough,
• 8:31
  you've believe it enough that do you do it in spite of fear.
• 8:37
  So speaking of important things like people shouldn't think I should, people shouldn't think,
• 8:41
  well, I feel fear about this, and therefore I shouldn't do it.
• 8:46
  It's normal to feel fear, like you'd have to definitely something mentally wrong
• 8:50
  if you should feel fear, so you just feel it and let the importance
• 8:54
  of it drive you to do it anyway. Yeah, you know. I
• 8:58
  Actually something I can be helpful isism some degree. Um, if you just
• 9:01
  if you just accept the probabilities, then that diminishes fear. So um.
• 9:09
  Starting SpaceX, I thought the odds of success were less than ten percent,
• 9:15
  and I just accepted that. Actually, probably I would just lose, lose
• 9:18
  everything, but that maybe would make some progress if we could just move the
• 9:24
  ball forward, even if we died, maybe some other company could pick up
• 9:28
  the baton and move and keep moving it forward so that we're still doing some
• 9:33
  good. Um. Yeah. Same with Tesla. I thought the odds of
• 9:37
  a car company succeeding were extremely low. What do you think the odds of
• 9:41
  the Mars colony are at this point today? Well, um, oddly enough,
• 9:48
  I actually think they're pretty good. Um, So like when can I
• 9:52
  go? Okay? Um, at this point, I am certain there is
• 9:56
  a way. I'm certain that success is one of the possible outcome for establishing
• 10:01
  a self sustaining mass colony about growing mass colony. I'm certain that that is
• 10:05
  possible, whereas until maybe a few years ago, I was not sure that
• 10:11
  success was even one of the possible outcomes. Some meaningful number of people going
• 10:15
  to Mars. I think this is potentially something that can be accomplished in about
• 10:20
  ten years, maybe sooner, maybe nine years. I need to make sure
• 10:28
  that SpaceX doesn't die between now and then, and that I don't die,
• 10:31
  or if I do die, that someone takes over who will continue That shouldn't
• 10:35
  go on the first launch, Yeah, exactly. The most launch will be
• 10:41
  a robotic anyway, so I want to go. Except for the Internet latency,
• 10:45
  Yeah, the INNRT latency will be pretty significant. Mars is roughly twelve
• 10:50
  light minutes from the Sun and Earth is eight light minutes, so closest approach
• 10:54
  Mars is four light minutes away that first approaches twenty a little more because you
• 10:58
  have to you can't talk directly through the on speaking of really important problems um
• 11:03
  AI. So you've been outspoken about AIM. Could you talk about what you
• 11:09
  think the positive future for AI looks like and how we get there? Okay,
• 11:13
  I mean I do want to emphasize that this is not really something that
• 11:20
  I advocate or this is not prescriptive. This is simply hopefully predictive, because
• 11:28
  who looks on a sailor will like like, this is something that I want
• 11:31
  to occur instead of something I think that probably is the best of the available
• 11:37
  alternatives. The best of the available alternatives that I can come up with,
• 11:45
  and maybe somebody else can come with a better approach or better outcome, is
• 11:50
  that we achieve democratrazation of AI technology, meaning that no one company or a
• 11:58
  small set of individuals has control over advanced AI technology. I think that that's
• 12:03
  very dangerous. Um it could also get stolen by somebody bad, you know,
• 12:09
  like some evil dictator or country could send their intelligence agency to go steal
• 12:15
  it and gain control. It just becomes a very unstable situation. I think
• 12:18
  if you've got any any incredibly powerful aim. You just don't know who's who's
• 12:26
  going to control that. So it's not as that I think that the risk
• 12:28
  is that the AI would develop a will of its own right off the bat.
• 12:31
  I think it's more that's the consumers that someone may use it in a
• 12:37
  way that is bad, or and even if they weren't going to use it
• 12:41
  in a way that's bad, that somebody could take it from them and use
• 12:43
  it in a way that's bad. That that I think is quite a big
• 12:46
  danger. So I think we must have democratrazation of AI technology and make it
• 12:50
  widely available. And that's you know, the reason that obviously, you mean,
• 12:56
  the rest of the team you know, created Open the Eye was to
• 13:01
  help with the democtras, help help spread out AI technology so it doesn't get
• 13:07
  concentrated in the hands of a few. UM. And but then of course
• 13:11
  that needs to be combined with solving the high band with interface to the cortex.
• 13:20
  UM. Humans are humans are socla, yes exactly, UM. But
• 13:26
  you know, we already have a situation in our brain where we've got the
• 13:30
  cortex and limit system, and the limit system is kind of the I mean,
• 13:35
  that's that's the primitive brain. It's kind of like the your your instincts
• 13:39
  and um whatnot. And then the cortex is the thinking upper part of the
• 13:43
  brain. Those two seem to work together quite well. Um. Occasionally your
• 13:48
  cortex and limbic system they disagree, but they generally works pretty well. And
• 13:54
  it's like rare to find someone who I've not found someone who wishes to either
• 13:58
  get rid of the cortex or get rid of the living system. Very true,
• 14:03
  Yeah, it's that's unusual. So so I think if if we can
• 14:07
  effectively UM merge with AI, like UM, improving that the neural link between
• 14:16
  your cortex and the the your digital extension of yourself, which already, like
• 14:22
  said, already exists, just has a bandwidth issue, UM, and then
• 14:26
  then effectively, UM, you become an AI human symbiote UM and and if
• 14:33
  that then is widespread with anyone who wants it can have it. Then we
• 14:39
  solve the control problem as well. We don't have to worry about UM some
• 14:43
  sort of evil dictator AI because kind of we are the aim collectively, that
• 14:48
  seems like the best outcome I can think of. So you've seen other companies
• 14:54
  in their early days that starts small and get really successful. UM, hope
• 14:58
  I don't get asking this on camera, but how do you open aye is
• 15:00
  going as a six month old company, I think should go pretty well.
• 15:03
  I think we've got a really talented group of Opening Eye and yeah, really
• 15:07
  really talented team and they're working hard. UM. Open Eye is structured as
• 15:13
  uh see a five one through a nonprofit UM. But you know, many
• 15:18
  nonprofits do not have a sense of urgency. If it's fine, they don't
• 15:22
  have to have a sense of urgency UM. But Opening EYEASUM because I think
• 15:28
  people really believe in the mission. I think it's important UM, and it's
• 15:31
  it's about minimizing UM the risk of existential harm in the future. And so
• 15:41
  I think it's going well. I'm pretty impressed with what people are doing in
• 15:45
  the talent level, and obviously we're always looking for UM great people to join
• 15:50
  who believe in the mission for people. Yes, all right, just a
• 15:56
  few more questions before we wrap up. How do you spend your days now?
• 16:00
  Like? What what do you allocate most of your time too? My
• 16:03
  time is mostly split up between SpaceX and and Tesla, and of course I
• 16:08
  try to spend it's a part of every week at Opening Eye UM. So
• 16:14
  I spend most I spend basically half a day at open eye most weeks,
• 16:18
  um. And then and then I have some opening eye stuff that happens during
• 16:22
  the week, but other than that, it's sax or tesla. Like what
• 16:27
  does your time look like there? Yeah, so that's a good question.
• 16:32
  Um. I think a lot of people think I must spend a lot of
• 16:34
  time with media or on business see things. But actually, almost almost all
• 16:40
  my time, like eighty percent of it is spent on engineering and design.
• 16:44
  Engineering and design, so it's um, developing the next generation product. That's
• 16:52
  percent of it. Um. You probably don't remember this a very long time
• 16:55
  ago, many many years, you took me on a tour of SpaceX,
• 16:57
  and the most impressive thing was that you knew every detail of the rocket and
• 17:00
  every piece of engineering that went into it. I don't think many people get
• 17:03
  that about you. Yeah, I think a lot of people think I'm kind
• 17:07
  of a business person or something. It's just fine, like business is fine,
• 17:10
  but like I really it's you know, it's like it's SpaceX. Gwyn
• 17:17
  Shotwell is chief operating officer. She kind of manages legal, finance, sales
• 17:25
  and kind of general business activity. And then my time is almost entirely with
• 17:29
  the inteering team working on improving the Falconine and the Dragon spacecraft and developing the
• 17:36
  most colonial architecture. And then at Tesla it's working on the Model three and
• 17:44
  yeah something. The design studio typically UM happy day a week dealing with this
• 17:51
  aesthetics and and look and feel things, and then most of the rest the
• 17:56
  week is just going through engineering of of the car itself as well as engineering
• 18:02
  of the factory. UM because the biggest epipany I've had this year is that
• 18:10
  what really matters is the is the machine that builds the machine, the factory
• 18:14
  UM, and that is at least towards magitude hotter than the vehicle itself.
• 18:18
  It's amazing to watch the robots go here and these cars just happen. Yeah.
• 18:23
  Now, this actually is has a relatively low level of automation compared to
• 18:29
  what the Giga factory will have in what Model three will have. What's the
• 18:33
  speed on the line of these cars? Actually, the average feed on the
• 18:37
  line is incredibly slow. It's probably about UM including both X and sum,
• 18:45
  it's maybe five, you know, five centimeters per second? And what can
• 18:52
  you do? This is very slow? Or what would you like to get
• 18:55
  to I'm confident we can get to to at least one per second, so
• 19:00
  a twenty fold increase that would be very fast. Yeah, um, at
• 19:04
  least I mean, I think quite a one meter per second. Just put
• 19:07
  that perspective, is a slow walk or a co medium speed walk. A
• 19:11
  fast walk could be one and a half meters per second, and and then
• 19:17
  the fastest humans can run over ten meters per second. So if we're only
• 19:22
  doing points oh five meters per second, that's very slow current current speed.
• 19:27
  And at one meter per second, you can still walk faster than the production line