episode
stringlengths
45
100
text
stringlengths
1
528
timestamp_link
stringlengths
56
56
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
fission reactions would produce neutrons of the right energy to continue the reaction.
https://karpathy.ai/lexicap/0009-large.html#01:05:50.320
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
And that was demonstrated in Germany, I think in 1938, if I remember correctly.
https://karpathy.ai/lexicap/0009-large.html#01:05:57.440
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
The first nuclear weapon patent was 1939 by the French. So this was actually going on well before
https://karpathy.ai/lexicap/0009-large.html#01:06:01.440
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
World War II really got going. And then the British probably had the most advanced capability
https://karpathy.ai/lexicap/0009-large.html#01:06:16.480
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
in this area. But for safety reasons, among others, and just resources, they moved the program
https://karpathy.ai/lexicap/0009-large.html#01:06:22.640
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
from Britain to the US and then that became Manhattan Project. So the reason why we couldn't
https://karpathy.ai/lexicap/0009-large.html#01:06:30.160
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
have any kind of oversight of nuclear weapons and nuclear technology
https://karpathy.ai/lexicap/0009-large.html#01:06:40.560
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
was because we were basically already in an arms race and a war.
https://karpathy.ai/lexicap/0009-large.html#01:06:46.560
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
LR But you mentioned then in the 20s and 30s. So what are the echoes? The way you've described
https://karpathy.ai/lexicap/0009-large.html#01:06:50.800
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
this story, I mean, there's clearly echoes. Why do you think most AI researchers,
https://karpathy.ai/lexicap/0009-large.html#01:07:00.960
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
folks who are really close to the metal, they really are not concerned about AI. They don't
https://karpathy.ai/lexicap/0009-large.html#01:07:06.800
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
think about it, whether it's they don't want to think about it. But why do you think that is,
https://karpathy.ai/lexicap/0009-large.html#01:07:11.760
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
is what are the echoes of the nuclear situation to the current AI situation? And what can we do
https://karpathy.ai/lexicap/0009-large.html#01:07:18.240
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
about it? BF I think there is a kind of motivated cognition, which is a term in psychology means
https://karpathy.ai/lexicap/0009-large.html#01:07:27.120
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
that you believe what you would like to be true, rather than what is true. And it's unsettling
https://karpathy.ai/lexicap/0009-large.html#01:07:35.520
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
to think that what you're working on might be the end of the human race, obviously. So you would
https://karpathy.ai/lexicap/0009-large.html#01:07:46.000
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
rather instantly deny it and come up with some reason why it couldn't be true. And I have,
https://karpathy.ai/lexicap/0009-large.html#01:07:52.640
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
I collected a long list of reasons that extremely intelligent, competent AI scientists have come up
https://karpathy.ai/lexicap/0009-large.html#01:08:00.560
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
with for why we shouldn't worry about this. For example, calculators are superhuman at arithmetic
https://karpathy.ai/lexicap/0009-large.html#01:08:08.160
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
and they haven't taken over the world. So there's nothing to worry about. Well, okay, my five year
https://karpathy.ai/lexicap/0009-large.html#01:08:16.800
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
old, you know, could have figured out why that was an unreasonable and really quite weak argument.
https://karpathy.ai/lexicap/0009-large.html#01:08:22.000
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
Another one was, while it's theoretically possible that you could have superhuman AI destroy the
https://karpathy.ai/lexicap/0009-large.html#01:08:29.040
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
world, it's also theoretically possible that a black hole could materialize right next to the
https://karpathy.ai/lexicap/0009-large.html#01:08:40.320
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
earth and destroy humanity. I mean, yes, it's theoretically possible, quantum theoretically,
https://karpathy.ai/lexicap/0009-large.html#01:08:45.680
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
extremely unlikely that it would just materialize right there. But that's a completely bogus analogy,
https://karpathy.ai/lexicap/0009-large.html#01:08:50.960
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
because, you know, if the whole physics community on earth was working to materialize a black hole
https://karpathy.ai/lexicap/0009-large.html#01:08:58.080
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
in near earth orbit, right? Wouldn't you ask them, is that a good idea? Is that going to be safe?
https://karpathy.ai/lexicap/0009-large.html#01:09:04.240
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
You know, what if you succeed? Right. And that's the thing, right? The AI community is sort of
https://karpathy.ai/lexicap/0009-large.html#01:09:10.160
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
refused to ask itself, what if you succeed? And initially I think that was because it was too hard,
https://karpathy.ai/lexicap/0009-large.html#01:09:16.720
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
but, you know, Alan Turing asked himself that, and he said, we'd be toast, right? If we were lucky,
https://karpathy.ai/lexicap/0009-large.html#01:09:24.240
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
we might be able to switch off the power, but probably we'd be toast. But there's also an aspect
https://karpathy.ai/lexicap/0009-large.html#01:09:32.720
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
that because we're not exactly sure what the future holds, it's not clear exactly,
https://karpathy.ai/lexicap/0009-large.html#01:09:37.600
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
so technically what to worry about, sort of how things go wrong. And so there is something,
https://karpathy.ai/lexicap/0009-large.html#01:09:45.200
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
it feels like, maybe you can correct me if I'm wrong, but there's something paralyzing about
https://karpathy.ai/lexicap/0009-large.html#01:09:53.360
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
worrying about something that logically is inevitable, but you have to think about it,
https://karpathy.ai/lexicap/0009-large.html#01:09:58.800
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
logically is inevitable, but you don't really know what that will look like.
https://karpathy.ai/lexicap/0009-large.html#01:10:05.200
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
Yeah, I think that's, it's a reasonable point and, you know, it's certainly in terms of
https://karpathy.ai/lexicap/0009-large.html#01:10:10.720
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
existential risks, it's different from, you know, asteroid collides with the earth, right? Which,
https://karpathy.ai/lexicap/0009-large.html#01:10:18.480
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
again, is quite possible, you know, it's happened in the past, it'll probably happen again,
https://karpathy.ai/lexicap/0009-large.html#01:10:24.000
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
we don't know right now, but if we did detect an asteroid that was going to hit the earth
https://karpathy.ai/lexicap/0009-large.html#01:10:29.520
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
in 75 years time, we'd certainly be doing something about it.
https://karpathy.ai/lexicap/0009-large.html#01:10:34.960
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
Well, it's clear there's got big rock and there's,
https://karpathy.ai/lexicap/0009-large.html#01:10:39.760
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
we'll probably have a meeting and see what do we do about the big rock with AI.
https://karpathy.ai/lexicap/0009-large.html#01:10:42.080
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
Right, with AI, I mean, there are very few people who think it's not going to happen within the
https://karpathy.ai/lexicap/0009-large.html#01:10:46.160
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
next 75 years. I know Rod Brooks doesn't think it's going to happen, maybe Andrew Ng doesn't
https://karpathy.ai/lexicap/0009-large.html#01:10:50.160
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
think it's happened, but, you know, a lot of the people who work day to day, you know, as you say,
https://karpathy.ai/lexicap/0009-large.html#01:10:56.160
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
at the rock face, they think it's going to happen. I think the median estimate from AI researchers is
https://karpathy.ai/lexicap/0009-large.html#01:11:02.800
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
somewhere in 40 to 50 years from now, or maybe, you know, I think in Asia, they think it's going
https://karpathy.ai/lexicap/0009-large.html#01:11:10.640
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
to be even faster than that. I'm a little bit more conservative, I think it'd probably take
https://karpathy.ai/lexicap/0009-large.html#01:11:16.000
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
longer than that, but I think, you know, as happened with nuclear weapons, it can happen
https://karpathy.ai/lexicap/0009-large.html#01:11:24.080
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
overnight that you have these breakthroughs and we need more than one breakthrough, but,
https://karpathy.ai/lexicap/0009-large.html#01:11:30.720
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
you know, it's on the order of half a dozen, I mean, this is a very rough scale, but sort of
https://karpathy.ai/lexicap/0009-large.html#01:11:34.960
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
half a dozen breakthroughs of that nature would have to happen for us to reach the superhuman AI.
https://karpathy.ai/lexicap/0009-large.html#01:11:40.640
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
But the, you know, the AI research community is vast now, the massive investments from governments,
https://karpathy.ai/lexicap/0009-large.html#01:11:49.920
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
from corporations, tons of really, really smart people, you know, you just have to look at the
https://karpathy.ai/lexicap/0009-large.html#01:11:57.280
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
rate of progress in different areas of AI to see that things are moving pretty fast. So to say,
https://karpathy.ai/lexicap/0009-large.html#01:12:03.360
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
oh, it's just going to be thousands of years, I don't see any basis for that. You know, I see,
https://karpathy.ai/lexicap/0009-large.html#01:12:09.200
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
you know, for example, the Stanford 100 year AI project, right, which is supposed to be sort of,
https://karpathy.ai/lexicap/0009-large.html#01:12:15.920
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
you know, the serious establishment view, their most recent report actually said it's probably
https://karpathy.ai/lexicap/0009-large.html#01:12:26.400
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
not even possible. Oh, wow.
https://karpathy.ai/lexicap/0009-large.html#01:12:32.400
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
Right. Which if you want a perfect example of people in denial, that's it. Because, you know,
https://karpathy.ai/lexicap/0009-large.html#01:12:35.280
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
for the whole history of AI, we've been saying to philosophers who said it wasn't possible,
https://karpathy.ai/lexicap/0009-large.html#01:12:42.880
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
well, you have no idea what you're talking about. Of course it's possible, right? Give me an argument
https://karpathy.ai/lexicap/0009-large.html#01:12:49.520
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
for why it couldn't happen. And there isn't one, right? And now, because people are worried that
https://karpathy.ai/lexicap/0009-large.html#01:12:53.920
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
maybe AI might get a bad name, or I just don't want to think about this, they're saying, okay,
https://karpathy.ai/lexicap/0009-large.html#01:13:00.400
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
well, of course, it's not really possible. You know, imagine if, you know, the leaders of the
https://karpathy.ai/lexicap/0009-large.html#01:13:06.080
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
cancer biology community got up and said, well, you know, of course, curing cancer,
https://karpathy.ai/lexicap/0009-large.html#01:13:12.240
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
it's not really possible. There'd be complete outrage and dismay. And, you know, I find this
https://karpathy.ai/lexicap/0009-large.html#01:13:17.360
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
really a strange phenomenon. So, okay, so if you accept that it's possible,
https://karpathy.ai/lexicap/0009-large.html#01:13:28.320
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
and if you accept that it's probably going to happen, the point that you're making that,
https://karpathy.ai/lexicap/0009-large.html#01:13:35.680
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
you know, how does it go wrong? A valid question. Without that, without an answer to that question,
https://karpathy.ai/lexicap/0009-large.html#01:13:42.400
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
then you're stuck with what I call the gorilla problem, which is, you know, the problem that
https://karpathy.ai/lexicap/0009-large.html#01:13:50.160
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
the gorillas face, right? They made something more intelligent than them, namely us, a few million
https://karpathy.ai/lexicap/0009-large.html#01:13:54.320
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
years ago, and now they're in deep doo doo. So there's really nothing they can do. They've lost
https://karpathy.ai/lexicap/0009-large.html#01:14:00.480
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
the control. They failed to solve the control problem of controlling humans, and so they've
https://karpathy.ai/lexicap/0009-large.html#01:14:07.680
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
lost. So we don't want to be in that situation. And if the gorilla problem is the only formulation
https://karpathy.ai/lexicap/0009-large.html#01:14:13.760
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
you have, there's not a lot you can do, right? Other than to say, okay, we should try to stop,
https://karpathy.ai/lexicap/0009-large.html#01:14:20.240
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
you know, we should just not make the humans, or in this case, not make the AI. And I think
https://karpathy.ai/lexicap/0009-large.html#01:14:26.640
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
that's really hard to do. I'm not actually proposing that that's a feasible course of
https://karpathy.ai/lexicap/0009-large.html#01:14:31.760
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
action. I also think that, you know, if properly controlled AI could be incredibly beneficial.
https://karpathy.ai/lexicap/0009-large.html#01:14:40.320
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
But it seems to me that there's a consensus that one of the major failure modes is this
https://karpathy.ai/lexicap/0009-large.html#01:14:48.800
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
loss of control, that we create AI systems that are pursuing incorrect objectives. And because
https://karpathy.ai/lexicap/0009-large.html#01:14:56.720
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
the AI system believes it knows what the objective is, it has no incentive to listen to us anymore,
https://karpathy.ai/lexicap/0009-large.html#01:15:05.040
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
so to speak, right? It's just carrying out the strategy that it has computed as being the optimal
https://karpathy.ai/lexicap/0009-large.html#01:15:12.240
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
solution. And, you know, it may be that in the process, it needs to acquire more resources to
https://karpathy.ai/lexicap/0009-large.html#01:15:21.680
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
increase the possibility of success or prevent various failure modes by defending itself against
https://karpathy.ai/lexicap/0009-large.html#01:15:30.480
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
interference. And so that collection of problems, I think, is something we can address. The other
https://karpathy.ai/lexicap/0009-large.html#01:15:36.800
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
problems are, roughly speaking, you know, misuse, right? So even if we solve the control problem,
https://karpathy.ai/lexicap/0009-large.html#01:15:45.920
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
we make perfectly safe controllable AI systems. Well, why? You know, why does Dr. Evil going to
https://karpathy.ai/lexicap/0009-large.html#01:15:55.680
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
use those, right? He wants to just take over the world and he'll make unsafe AI systems that then
https://karpathy.ai/lexicap/0009-large.html#01:16:01.600
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
get out of control. So that's one problem, which is sort of a, you know, partly a policing problem,
https://karpathy.ai/lexicap/0009-large.html#01:16:06.480
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
partly a sort of a cultural problem for the profession of how we teach people what kinds
https://karpathy.ai/lexicap/0009-large.html#01:16:12.960
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
of AI systems are safe. You talk about autonomous weapon system and how pretty much everybody
https://karpathy.ai/lexicap/0009-large.html#01:16:21.280
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
agrees that there's too many ways that that can go horribly wrong. This great slaughterbots movie
https://karpathy.ai/lexicap/0009-large.html#01:16:26.000
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
that kind of illustrates that beautifully. I want to talk about that. That's another,
https://karpathy.ai/lexicap/0009-large.html#01:16:32.000
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
there's another topic I'm having to talk about. I just want to mention that what I see is the
https://karpathy.ai/lexicap/0009-large.html#01:16:36.960
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
third major failure mode, which is overuse, not so much misuse, but overuse of AI that we become
https://karpathy.ai/lexicap/0009-large.html#01:16:41.200
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
overly dependent. So I call this the WALL E problem. So if you've seen WALL E, the movie,
https://karpathy.ai/lexicap/0009-large.html#01:16:49.760
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
all right, all the humans are on the spaceship and the machines look after everything for them,
https://karpathy.ai/lexicap/0009-large.html#01:16:54.960
Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
and they just watch TV and drink big gulps. And they're all sort of obese and stupid and they
https://karpathy.ai/lexicap/0009-large.html#01:17:00.240