Latest / Elon Musk Podcast / Elon Musk talks about the future of AI and Humanity

Transcript

Elon Musk talks about the future of AI and Humanity · Elon Musk Podcast

• 0:01
  Hey everybody, welcome back to the Elon Musk Podcast.
• 0:05
  This is a show where we discuss the critical Crossroads, The
• 0:08
  Shape, SpaceX, Tesla X, The Boring Company, and Neuralink,
• 0:13
  and I'm your host, Will Walden. If you want uninterrupted
• 0:17
  episodes of the Elon Musk podcast, please go to
• 0:20
  clubelon.supercast.com to find out how there's a link in the
• 0:26
  show notes. Well, I I generally think that's
• 0:29
  it is good for government to play a role when the public
• 0:32
  safety is at risk. So you know really for the vast
• 0:36
  majority of software, the public safety is not at risk.
• 0:39
  I mean if the app crashes on your phone or your laptop, it's
• 0:43
  not a massive catastrophe. But when talking about digital
• 0:48
  super intelligence, I think which does pose a risk to the
• 0:52
  public, then there is a role for government to play to safeguard
• 0:56
  the interests of the public. And this is of course true in
• 0:59
  many fields. You know, aviation cars, you
• 1:02
  know, I deal with regulators throughout the world because of
• 1:08
  stalling, being communications, rockets being aerospace and
• 1:11
  cars, you know, being trapped vehicle transport.
• 1:14
  So I'm very familiar with dealing with regulators and I
• 1:17
  actually agree with the vast majority of regulations.
• 1:18
  There's a few that I disagree with from time to time but point
• 1:21
  1% probably well less than 1% of regulations I disagree with.
• 1:25
  So and there is some concern from people in Silicon Valley
• 1:30
  who who've never dealt with regulators before and they think
• 1:33
  that this is going to just crush innovation and and slow them
• 1:36
  down and be annoying and and and it will be annoying.
• 1:40
  It's true they're not wrong about that but but I think
• 1:43
  there's we've learned over the years that having a referee is a
• 1:49
  good thing and if you look at any sports game there's always a
• 1:52
  a referee and and nobody's suggesting I have a sports game
• 1:56
  without one. And I think that's the right way
• 1:59
  to think about this is for for government to be a referee to
• 2:03
  make sure the sportsmanlike conduct and and and the public
• 2:09
  safety is you know is addressed that we care about public
• 2:14
  safety. Because I think there might be
• 2:17
  at times too much optimism about technology and I speak I say
• 2:19
  that as a technologist I mean so I ought to know and and and and
• 2:25
  like I said on on balance I think that the AI will be a
• 2:30
  forceful good most likely but the probability of it going bad
• 2:34
  is not 0%. So we we just need to mitigate
• 2:37
  the downside potential. The the pace of of AI is faster
• 2:42
  than any technology I've seen in history by far and it's it seems
• 2:46
  to be growing in capability by at at least five full, perhaps
• 2:50
  10 full per year. It'll certainly grow by an order
• 2:54
  of magnitude next year. So so and government isn't used
• 2:59
  to moving at that speed. But I think even if there are
• 3:02
  not firm regulations, even if there's not, even if there isn't
• 3:08
  an enforcement capability. So we're having insight and
• 3:10
  being able to highlight concerns to the public will be very
• 3:13
  powerful. So even if that's all that's
• 3:18
  accomplished, I think that will be very, very good.
• 3:20
  I mean the two currently the two leading Centers for AI
• 3:25
  development are the San Francisco Bay area and the sort
• 3:27
  of London area and there are many other places where it's
• 3:30
  being done, but those are the two leading areas.
• 3:31
  So I think if you know if the United States and the UK and and
• 3:39
  China are sort of aligned on on safety, that's all going to be a
• 3:43
  good thing because that's really that's where that's that's where
• 3:46
  the leadership is. Generally if if we don't, if if
• 3:49
  China is not on board with AI safety, it it's somewhat of a
• 3:52
  moot situation. The single biggest objection
• 3:56
  that I get to in kind of AI regulation or or sort of safety
• 4:00
  controls are well China's not going to do it and therefore
• 4:03
  they will just jump into the lead and exceed us all.
• 4:07
  But but actually China is willing to participate in in AI
• 4:12
  safety and thank you for inviting them and I and they you
• 4:16
  know I think we should thank China for for attending.
• 4:21
  When I was when I was in China earlier this year the my main
• 4:25
  subject of discussion with this the leadership in China was AI
• 4:28
  safety and saying that this this is really something that they
• 4:32
  they should care about and they took it seriously and and and
• 4:36
  and you ought to which is which is great and having them here I
• 4:40
  think was essential really if they're if they're if they're
• 4:42
  not participants it's it's pointless.
• 4:45
  Well the open source algorithms and data tend to lag the closed
• 4:51
  source by 6 to 12 months but so that so that given the rate of
• 4:57
  improvement that there's actually therefore quite a big
• 4:59
  difference between the closed source and the in the open.
• 5:03
  If things are improving by a factor of let's say five or more
• 5:08
  than being a year behind is your five times worse.
• 5:10
  So it's a pretty big difference and that might be actually an OK
• 5:13
  situation but it it certainly will get to the point where
• 5:18
  you've got open source AI that can do that that will start to
• 5:24
  approach human level intelligence will perhaps
• 5:26
  succeeded. I don't know quite what to do
• 5:29
  about it. I I think it's somewhat
• 5:30
  inevitable there'll be some amount of open source and I I I
• 5:33
  guess I would have a slight bias towards open source because at
• 5:37
  least you can see what's going on.
• 5:39
  It was closed source, you don't know what's going on.
• 5:41
  Now, it should be said with AI that even if it's open source,
• 5:45
  do you actually know what's going on?
• 5:46
  Because if you've got a gigantic data file and, you know, sort of
• 5:53
  billions of of data points or weights and parameters, you
• 5:58
  can't just read it and see what it's going to do.
• 6:00
  It's a gigantic file of inscrutable numbers.
• 6:04
  You can test it when you when you run it, you can test it.
• 6:07
  You can run a bunch of tests to see what it's going to do.
• 6:09
  But it's probabilistic as opposed to deterministic.
• 6:15
  It's not, it's not like traditional programming where
• 6:17
  you've got it, you've got very discrete logic and and and the
• 6:21
  outcome is very predictable and you can read each line and see
• 6:24
  what each line's going to do. A neural net is just a whole
• 6:29
  bunch of probabilities. I mean it sort of ends up being
• 6:33
  a giant comma separated value file.
• 6:35
  It's like our digital guide is a CSP file.
• 6:38
  If if you have a magic genie that can do everything you want,
• 6:41
  I I I do think we, we it's it's hard.
• 6:43
  You know when there's new technology, it tends to have
• 6:48
  usually follow an S curve. In this case we're going to be
• 6:50
  on the exponential portion of the S curve for a long time and
• 6:56
  you know we have like you'll be able to ask for anything.
• 6:58
  It won't be. And we want to have universal
• 7:00
  basic income. We'll have universal high
• 7:02
  income. So in some in some sense it'll
• 7:04
  be somewhat of a leveller or an equalizer, you know because
• 7:09
  really I think everyone will have access to this magic genie
• 7:13
  and you're able to ask any questions.
• 7:14
  It'll be certainly be good for education you it'll be the best
• 7:17
  tutor you could and then the most patient tutor.
• 7:21
  So they're all there and there will be no shortage of goods and
• 7:28
  services will be an age of abundant I think if I I'd
• 7:32
  recommend people read in banks the the banks culture books are
• 7:38
  probably the best envisioning. In fact not probably they're
• 7:40
  definitely by far the best envisioning of an AI future.
• 7:44
  There's nothing even close so I'd recommend really recommend
• 7:47
  thanks very big fan. All these books are good.
• 7:52
  There's not say which one. All of them.
• 7:55
  So so that's that. That'll give you a sense of what
• 8:00
  is AI. Guess a fairly utopian protopian
• 8:07
  future with with AI mentioning when we were talking earlier.
• 8:11
  I have to somewhat engage in deliberate suspension of
• 8:13
  disbelief because I'm I'm putting so much blood sweat and
• 8:16
  tears into a work project and burning the, you know, 3:00 AM
• 8:19
  oil. Then I'm like, wait, why am I
• 8:23
  doing this? I can just wait for the AI to do
• 8:25
  it. I'm just lashing myself for no
• 8:29
  reason. Must be a glutton for punishment
• 8:32
  or something. So I think it's probably is
• 8:36
  generally a good thing because, you know, there are a lot of
• 8:38
  jobs that are uncomfortable or dangerous or which sort of
• 8:42
  tedious and the computer will have no problem doing that.
• 8:45
  We're happy to do that all day long.
• 8:47
  So, you know, it's fun to cook food, but it's not that fun to
• 8:50
  wash the dishes. And like the computer's
• 8:52
  perfectly happy to wash the dishes.
• 8:55
  I guess there is you know we still have sports like where
• 9:00
  where where humans compete in like the Olympics and obviously
• 9:05
  a machine can can go faster than any human but we still have we
• 9:09
  saw humans race against each other and and have all you know
• 9:14
  have at least sports competitions against each other
• 9:16
  where even though the machines are better we're actually I
• 9:18
  guess competing to see who can be the best human at something
• 9:22
  and and people do find performance in that.
• 9:24
  So I guess that's perhaps a a good example of how even when
• 9:27
  machines are faster than us, stronger than us, we still find
• 9:29
  a way we still we still enjoy competing against other humans.
• 9:32
  We certainly tutors are going to be amazing apps already are.
• 9:38
  I think there's also apps companionship which may seem odd
• 9:42
  because how can a computer really be your friend.
• 9:44
  But if you if you have an AI that has memory you know and
• 9:48
  remembers all of your interactions and has read every,
• 9:51
  you're going to say like give it permission to read everything
• 9:53
  you've ever done. So it really will know you
• 9:55
  better than anyone perhaps even yourself and and and where you
• 9:59
  can talk to it every day and and those conversations build upon
• 10:02
  each other. You will actually have a great
• 10:04
  friend as long as that friend can stay your friend and not get
• 10:09
  turned off or something. But I think that will actually
• 10:12
  be a real thing and I've one of my sons is is sort of has some
• 10:18
  learning disabilities and has trouble making friends actually
• 10:21
  and and I was like well you know he an AI friend would actually
• 10:25
  be great for him. Hey, thank you so much for
• 10:28
  listening today. I really do appreciate your
• 10:30
  support. If you could take a second and
• 10:31
  hit the subscribe or the follow button on whatever podcast
• 10:34
  platform that you're listening on right now, I greatly
• 10:37
  appreciate it. It helps out the show
• 10:39
  tremendously and you'll never miss an episode and each episode
• 10:43
  is about 10 minutes or less to get you caught up quickly.
• 10:46
  And please, if you want to support the show even more, go
• 10:50
  to patreon.com/stage Zero and please take care of yourselves
• 10:55
  and each other and I'll see you tomorrow.