Stuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9

Lex Fridman 0:00 The following is a conversation with Stuart Russell. He's a professor of computer science at UC Berkeley and a co-author of a book that introduced me and millions of other people to the amazing world of AI, called Artificial Intelligence: The Modern Approach. So, it was an honor for me to have this conversation as part of MIT course on Artificial General Intelligence and The Artificial Intelligence podcast. If you enjoy it, please subscribe on YouTube, iTunes, or your podcast provider of choice, or simply connect with me on Twitter @lexfridman, spelled F-R-I-D. And now, here's my conversation with Stuart Russell. So you've mentioned in 1975, in high school, you've created one of your first AI programs that played chess.

Stuart Russell 0:49 Yeah.

Lex Fridman 0:50 Were you ever able to build a program that beat you at chess or another board game?

Stuart Russell 0:59 Uh, so my program never beat me at chess. I actually wrote the program at Imperial College, so I used to take the bus every Wednesday with a box of cards this big, uh, and shove them into the card reader, and they gave us eight seconds of CPU time. It took about five seconds to read the cards in and compile the code, so we had three seconds of CPU time, which was enough to make one move, you know, with a not very deep search, and then we would print that move out, and then we'd have to go to the back of the queue and wait to feed the cards in again.

Lex Fridman 1:36 How deep was the search?

Stuart Russell 1:37 (laughs)

Lex Fridman 1:37 What, are we talking about one move, two moves, three moves?

Stuart Russell 1:39 Uh, so, no, I think we got, uh, we got an eight-move, uh, eight, you know, depth eight, um, with alpha-beta, and we had some tricks of our own about, um, move ordering and some pruning of the tree, and...

Lex Fridman 1:53 But you were still able to beat that program?

Stuart Russell 1:55 Yeah, yeah. I, I was a reasonable chess player in my youth.

Lex Fridman 1:58 (laughs)

Stuart Russell 1:59 I did an Othello program, uh, and a backgammon program. So when I got to Berkeley, I worked a lot on what we call meta-reasoning, which really means reasoning about reasoning. In, in the case of a game-playing program, you need to reason about what parts of the search tree you're actually going to explore, because the search tree is enormous, uh, you know, bigger than the number of atoms in the universe, and, and, uh, the way programs succeed and the way humans succeed is by only looking at a small fraction of the search tree. And if you look at the right fraction, you play really well. If you look at the wrong fraction, if you waste your time thinking about things that are never gonna happen, the moves that no one's ever gonna make, then you're gonna lose 'cause you, you won't be able to figure out the right decision. So that question of how machines can manage their own computation, how they, how they decide what to think about is the meta-reasoning question. We developed some methods for doing that, and very simply, a machine should think about whatever thoughts are going to improve its decision quality. We were able to show that both for Othello, which is a standard two-player game, and, uh, for backgammon, which includes, uh, dice rolls, so it's a two-player game with uncertainty. For both of those cases, we could come up with algorithms that were actually much more efficient than the standard alpha-beta search, uh, which chess programs at the time were using, and that, those programs could beat me. And I think you can see the same basic ideas in AlphaGo and AlphaZero today. The way they explore the tree is using a form of meta-reasoning to select what to think about based on how useful it is to think about it.

Lex Fridman 3:57 Is there any insights you can describe without Greek symbols of how do we select which paths to go down?

Stuart Russell 4:05 There, there's really two kinds of learning going on. So, uh, uh, as you say, AlphaGo learns to evaluate board positions. So it can, it can look at a Go board, and it actually has probably a superhuman ability to instantly tell how promising that situation is.

Lex Fridman 4:22 Mm-hmm.

Stuart Russell 4:24 To me, the amazing thing about AlphaGo is not that it can beat the world champion with its hands tied behind its back, but, uh, the fact that if you stop it from searching altogether, so you say, "Okay, you're not allowed to do any thinking ahead, right? You can just consider each of your legal moves, and then look at the resulting situation and evaluate it." So what we call a depth-one search.

Lex Fridman 4:51 Mm-hmm.

Stuart Russell 4:51 So just the immediate outcome of your moves and decide if that's good or bad. That version of AlphaGo can still play at a professional level, right? And even professionals are sitting there for five, 10 minutes deciding what to do, and AlphaGo, in less than a second, can instantly intuit what is the right move to make based on its ability to evaluate positions. Um, and that is remarkable, um, because, you know, we don't have that level of intuition about Go. We actually have to think about the situation. So anyway, that capability that, um, AlphaGo has is one big part of why it beats humans.

Lex Fridman 5:35 Right.

Stuart Russell 5:36 The other big part is that it's able to look ahead 40, 50, 60 moves into the future.

Lex Fridman 5:43 Mm-hmm.

Stuart Russell 5:44 And, you know, if it was considering all possibilities 40 or 50 or 60 moves into the future, that would be, you know, 10 to the 200 possibilities, so way, way more than, you know, atoms in the universe.

Lex Fridman 6:01 Mm-hmm.

Stuart Russell 6:01 And, and so on. So it's very, very selective about what it looks at.So, let me try to give you an intuition about how you decide what to think about. It's a combination of two things. One is, um, how promising it is.

Lex Fridman 6:18 Mm-hmm.

Stuart Russell 6:18 Right? So, if you're already convinced that a move is terrible, there's no point spending a lot more time convincing yourself that it's terrible.

Lex Fridman 6:26 Mm-hmm.

Stuart Russell 6:27 Uh, because it's probably not gonna change your mind. So, the- the real reason you think is because there's some possibility of changing your mind about what to do.

Lex Fridman 6:36 Mm-hmm.

Stuart Russell 6:37 Right? And it's that changing of mind that would result then in- in a better final action in the real world. So, that's the purpose of thinking, is to improve the final action in the real world. And so, if you think about a move that is guaranteed to be terrible, you can convince yourself it's terrible, you're still not gonna change your mind.

Lex Fridman 6:58 Mm-hmm.

Stuart Russell 6:59 Right? But on the other hand here, suppose you had a choice between two moves. One of them, you've already figured out is guaranteed to be a draw, let's say.

Lex Fridman 7:07 Mm-hmm.

Stuart Russell 7:08 And then the other one looks a little bit worse. Like it looks fairly likely that if you make that move, you're gonna lose. But there's still some uncertainty-

Lex Fridman 7:17 Mm-hmm.

Stuart Russell 7:17 ... about the value of that move. There's still some possibility that it will turn out to be a win.

Lex Fridman 7:22 Mm-hmm.

Stuart Russell 7:23 Right? Then it's worth thinking about that. So even though it's less promising on average than the other move, which is guaranteed to be a draw, there's still some purpose in thinking about it, because there's a chance that you'll change your mind and- and discover that in fact it's a better move. So, it's a combination of how good the move appears to be and how much uncertainty there is about its value. The more uncertainty, the more it's worth thinking about, because there's a higher upside, if you wanna think of it that way.

Lex Fridman 7:52 And of course, in the beginning, especially in the AlphaGo Zero formulation, it's, everything is shrouded in- in uncertainty. So, you're really swimming in a sea of, uh, uncertainty, so it- it benefits you to... I mean, actually following the same process as you described, but because you're so uncertain about everything, uh, y- you basically have to try a lot of different directions.

Stuart Russell 8:15 Yeah. So- so the- the early parts of the search tree, uh, are fairly bushy-

Lex Fridman 8:20 Mm-hmm.

Stuart Russell 8:20 ... um, that it w- it will look at a lot of different possibilities. But fairly quickly, the degree of certainty about some of the moves... I mean, if a move is really terrible, you'll pretty quickly find out, right? You'll lose half your pieces or half your territory, and, um, and then you'll say, "Okay, this- this is not worth thinking about anymore." And then, so further down, the tree becomes very long and narrow.

Lex Fridman 8:43 Mm-hmm.

Stuart Russell 8:43 And, uh, you're- you're following various lines of play, you know, 10- 10, 20, 30, 40, 50 moves into the future. And, um, you know, that again is something that- that human beings have a very hard time doing, uh, mainly because they just lack the short-term memory. You just can't remember a sequence of moves that's 50 moves long, uh, and you can't- you can't, uh, imagine the board correctly, uh, for that many moves into the future.

Lex Fridman 9:13 Of course, the top players, I- I'm much more familiar with chess, but the top players probably have... they have echoes of the same kind of intuition and instinct that, in a moment's time, AlphaGo applies when they see a board. I mean, they've seen those patterns. Human beings have seen those patterns before at the top, at the grandmaster level. It seems that there's some, uh, similarities, or maybe it's- it's our imagination creates a- a vision of those similarities, but it feels like this kind of pattern recognition that the AlphaGo approaches are- are using is similar to what human beings at the top level are using.

Stuart Russell 9:56 I think there's, uh, there's some truth to that.

Lex Fridman 10:01 Oh, but not entirely?

Stuart Russell 10:02 Yeah. I- I mean, I think the- the extent to which a human grandmaster can reliably rec- instantly recognize the right move and instantly recognize the value of a position, uh, I think that's a little bit overrated.

Lex Fridman 10:17 But if you sacrifice a queen, for example, I mean, there's these- there's these beautiful games of chess with Bobby Fischer or somebody, where it's seeming to make a bad move, and I'm not sure there's a- a- a perfect degree of calculation involved, where they've calculated all the possible things that happen-

Stuart Russell 10:36 Mm-hmm.

Lex Fridman 10:37 ... but there's an instinct there, right? That somehow adds up to go down that path.

Stuart Russell 10:42 Th- yeah. So, I think what happens is you- you- you get a sense that there's some possibility in the position, uh, even if you make a weird-looking move, um, that it opens up some- some lines of- of calculation that otherwise would be, uh, definitely bad. And, um, and it's that intuition that there's something here in this position that might, uh, might yield a win.

Lex Fridman 11:12 Down the set of possibilities, yeah.

Stuart Russell 11:13 And then- and then you follow that, right? And- and in some sense, when a- when a chess player is following a line in- in his or her mind, they're- they're mentally simulating what the other person is gonna do, what the opponent is gonna do. And they can do that as long as the moves are kind of forced.

Lex Fridman 11:32 Mm-hmm.

Stuart Russell 11:33 Right? As long as there's a, you know, there's a- a for- what we call a forcing variation, where the opponent doesn't really have much choice how to respond, and then you see if you can force them into a situation where you win. You know, we see plenty of mistakes, uh, even- even in grandmaster games, where they just miss some simple three, four, five-move, uh, combination that, you know, wasn't particularly apparent in- in the position, but was still there.

Lex Fridman 12:01 That's the thing that makes us human.

Stuart Russell 12:02 Yep.

Lex Fridman 12:03 So, when... you mentioned that in Othello, those games were, after some, uh, meta-reasoning improvements and research, was able to beat you.... how did that make you feel?

Stuart Russell 12:15 Part of the meta-reasoning capability that it had, um, (clears throat) was based on learning.

Lex Fridman 12:20 Mm-hmm.

Stuart Russell 12:21 And, um, and you could sit down the next day and, and you could just feel that it had got a lot smarter. (clears throat)

Lex Fridman 12:29 Yeah.

Stuart Russell 12:29 You know, and all- and all of a sudden, you really felt like you were sort of pressed against the wall because-

Lex Fridman 12:35 Yeah.

Stuart Russell 12:35 ... uh, it was, it was much more aggressive and, and was totally unforgiving of, of any minor mistake that you might make. And, uh, and actually s- it seemed understood the game better, uh, than I did. And, you know, Garry Kasparov has this quote where, um, during his match against Deep Blue, he said he suddenly felt that there was a new kind of intelligence across the board.

Lex Fridman 12:58 Hmm.

Stuart Russell 13:00 So-

Lex Fridman 13:00 Do you think that's a scary or an exciting possibility-

Stuart Russell 13:04 Um.

Lex Fridman 13:04 ... for Garry Kasparov and for yourself? In, in the context of chess, purely sort of in this, like, that feeling, whatever that is.

Stuart Russell 13:14 I, I think it's definitely, um, an exciting feeling. You know, this is what made me work on AI in the first place was as soon as I really understood what a computer was, I wanted to make it-

Lex Fridman 13:25 Mm-hmm.

Stuart Russell 13:25 ... smart. You know, I started out with, uh, the first program I wrote was for the Sinclair Programmable Calculator.

Lex Fridman 13:33 (laughs)

Stuart Russell 13:33 Uh, and I think you could write a 21-step, uh, algorithm. That was the biggest program you could write-

Lex Fridman 13:41 Mm-hmm.

Stuart Russell 13:41 ... something like that, um, and do little arithmetic calculation. So I syn- I think I implemented Newton's method for square roots and a few other things like that. Um, but then, you know, I thought, "Okay, if, if I just had more space-"

Lex Fridman 13:55 Mm-hmm.

Stuart Russell 13:55 "... I could make this thing intelligent." And so I started thinking about AI and-

Lex Fridman 14:04 S-

Stuart Russell 14:05 And I think the, the, the thing that's scary is not, is not the chess program because, you know, chess programs, they're not in the taking-over-the-world business.

Lex Fridman 14:18 (laughs)

Stuart Russell 14:18 Uh, but if you extrapolate, you know, there are things about chess that don't resemble the real world, right? We know, we know the rules of chess. The chess board is completely visible to the program, where of course the real world is not. Most y- most of the real world is, is not visible from wherever you're sitting, so to speak. And, uh, to overcome those kinds of problems, you need qualitatively different algorithms. Another thing about the real world is that, you know, we, we regularly plan ahead on the timescales involving billions or trillions of steps.

Lex Fridman 15:07 Mm-hmm.

Stuart Russell 15:08 Um, now we don't plan those in detail, but, you know, when you choose to do a PhD at Berkeley, that's a five-year commitment and that amounts to about a trillion motor control steps that you will eventually, uh, be committed to.

Lex Fridman 15:24 Including going up the stairs, opening doors, drinking water-

Stuart Russell 15:29 Type, yeah, I mean every-

Lex Fridman 15:30 ... typing. (laughs)

Stuart Russell 15:30 ... every, every finger movement while you're typing, every character of every paper and, you know-

Lex Fridman 15:34 Yeah.

Stuart Russell 15:34 ... thesis and everything else. So you're not committing in advance to the specific motor control steps, but you're still reasoning on a timescale that will eventually reduce to, uh, trillions of motor control actions. And, uh, so for all of these reasons, you know, AlphaGo and, and Deep Blue and so on don't represent any kind of threat to humanity, but they are a step towards it, right?

Lex Fridman 16:00 Yes.

Stuart Russell 16:01 That... And progress in AI occurs by essentially removing one by one-

Lex Fridman 16:08 Mm-hmm.

Stuart Russell 16:08 ... these assumptions that make problems easy, like the assumption of complete observability of the situation, right? If we remove that assumption, you need a much more complicated kind of computing design and you n- you need something that actually keeps track of all the things you can't see and tries to estimate what's going on, uh, and there's inevitable uncertainty, uh, in that. So it becomes a much more complicated problem. But, you know, we are removing those assumptions. We are starting to have algorithms that can cope with much longer timescales, that can cope with uncertainty, that can cope with partial observability. And so each of those steps sort of magnifies by 1,000 the range of things that we can do with AI systems.

Lex Fridman 16:55 So the way I started in AI, I wanted to be a psychiatrist for a long time. I wanted to understand the mind in high school, and of course program and so on. And then I showed up, uh, University of Illinois to an AI lab and they said, "Okay, I don't have time for you, but here's a book, AI: A Modern Approach." I think it was the first edition at the time.

Stuart Russell 17:15 Mm-hmm.

Lex Fridman 17:16 (laughs) And, "Here, go, go, go learn this." And I remember the lay of the land was, well, it's incredible that we solved chess, but we'll never solve Go. I mean, it was pretty certain that Go in the way we thought about systems that reason was impossible to solve, and now we've solved it, so it's a very-

Stuart Russell 17:34 Well, I think I, I would have said that it's unlikely we could take the kind of algorithm that was used for chess and just get it to scale up, uh, and work well for Go. And at the time what we thought was that in order to solve Go, we would have to do something similar to the way humans manage the complexity of Go, which is to break it down into kind of sub-games. So-

Lex Fridman 18:05 Mm-hmm.

Stuart Russell 18:05 ... when a human thinks about a Go board, they think about different parts of the board as sort of weakly connected to each other, and they think about, "Okay, within this part of the board, here's how things could go. In that part of the board, here's how things could go."

Lex Fridman 18:19 Mm-hmm.

Stuart Russell 18:19 And then you try to sort of couple those two analyses together.... and deal with the interactions and maybe revise your views of how things are gonna go in each part, and then you've got maybe five, six, seven, 10 parts of the board.

Lex Fridman 18:31 Mm-hmm.

Stuart Russell 18:32 And, um, that actually resembles the real world much more than chess does-

Lex Fridman 18:39 Mm-hmm.

Stuart Russell 18:39 ... because in the real world, you know, we have work, we have home life, we have sport, you know, whatever... different kinds of activities, you know, shopping. These all are connected to each other but they're weakly connected. So, when I'm typing a paper, you know, I don't simultaneously have to decide which order I'm gonna get the, you know, the milk and the butter. You know, that doesn't affect the typing. But I do need to realize, "Okay, I better finish this before the shops close because I don't have anything... I don't have any food at home."

Lex Fridman 19:13 Right.

Stuart Russell 19:14 Right? So there's some weak connection, but not in the way that chess works where everything is tied into a single stream of thought. So the, the thought was that Go... to, to solve Go, we'd have to make progress on stuff that would be useful for the real world. And in a way, AlphaGo is a little bit disappointing-

Lex Fridman 19:32 (laughs) Right.

Stuart Russell 19:32 ... because the p- the program design for AlphaGo is actually not that different from, from Deep Blue or f- or even from Arthur Samuel's checker playing program from the 1950s. And in fact, the... so the two things that make AlphaGo work is one, one is its amazing ability, ability to evaluate the positions, and the other is the meta-reasoning capability which, which allows it to, to explore some paths in the tree very deeply and to abandon other paths very quickly.

Lex Fridman 20:07 So th- this word meta-reasoning, uh, while technically correct, inspires perhaps the, the wrong degree of power that AlphaGo has. For example, the word reasoning is a, is a powerful word. Let me ask you sort of... So you were part of the symbolic AI world for a while, like, where AI was, uh-

Stuart Russell 20:29 Mm-hmm.

Lex Fridman 20:29 There was a lot of excellent interesting ideas there that unfortunately met a winter. And (laughs) so, i- do you think it reemerges as if-

Stuart Russell 20:40 Oh, so I, I would say... Yeah, it's not quite as simple as that. So the, the AI winter... The fir- the first winter that was actually named as such was the one in the late '80s, and that came about because, uh, in the mid-'80s, there was a... really a concerted attempt to push AI out into the real world, uh, using what, uh, was called expert system technology. And for the most part, that technology was just not ready for prime time. They were trying, uh, in many cases to do a form of uncertain reasoning-

Lex Fridman 21:24 Mm-hmm.

Stuart Russell 21:25 ... uh, judge, you know, judgment, combinations of evidence, diagnosis, those kinds of things, which was simply invalid. Uh, and when you try to apply invalid reasoning methods to real problems, you can fudge it for small versions of the problem, but when it starts to get larger, the thing just falls apart. So, many companies found that, uh, the stuff just didn't work and they were spending tons of money on consultants to try to make it work and there were, you know, other practical reasons like, you know, they, they were asking the companies to buy incredibly expensive Lisp machine, uh, workstations which were literally between $50,000 and $100,000 in, uh, you know, in 1980s money, which was... would be, like, between $150,000 and $300,000 per workstation in current prices, so-

Lex Fridman 22:21 And then the bottom line, they weren't, uh, seeing a profit from it?

Stuart Russell 22:24 Yeah, um, they, they... In many cases. I think there were some successes, there's no doubt about that. But people, I would say, overinvested. Every major company was starting an AI department, just like now, and I worry a bit that we might see similar disappointments, not because the technol- the current technology is invalid, but it's limited in its scope. And, uh, it's almost the, the dual of the, you know, the scope problems that expert systems had, so...

Lex Fridman 23:01 What have you learned from that hype cycle and what can we do to prevent another winter, for example?

Stuart Russell 23:07 Uh, yeah. So when I'm, I'm giving talks these days, that's one of the warnings that I give. Uh, so there's this two, two-part warning slide. One is that, uh, you know, rather than data being the new oil, data is the new snake oil.

Lex Fridman 23:22 (laughs)

Stuart Russell 23:22 Uh...

Lex Fridman 23:22 That's a good line.

Stuart Russell 23:23 And then, um...

Lex Fridman 23:25 (laughs)

Stuart Russell 23:25 And then the other, uh, is that we might see a kind of very visible failure in some of the major application areas, and I think self-driving cars would be the flagship. And, uh, I think when you look at the history... So the first self-driving car was on the freeway, uh, driving itself, changing lanes, overtaking, in 1987. And, uh, so it's more than 30 years.

Lex Fridman 24:01 Yep.

Stuart Russell 24:01 And God, that kind of looks like where we are today, right? You know, prototypes on the freeway, changing lanes and overtaking.

Lex Fridman 24:08 Yep.

Stuart Russell 24:08 Um, now, I think significant progress has been made, particularly on the perception side. So, we worked a lot on autonomous vehicles in the early mid-'90s at Berkeley, you know, and we had our own big demonstrations, you know. We, we put congressmen into-

Lex Fridman 24:27 Yep.

Stuart Russell 24:27 ... self-driving cars and, and had them zooming along the freeway.And, uh, the problem was clearly perception, right?

Lex Fridman 24:37 At the time, the problem was the perception?

Stuart Russell 24:39 At the- Yeah. So w- you know, in simulation with perfect perception, you could actually show that you can drive safely for a long time even if the other car's misbehaving, and- and so on. But simultaneously, we worked on mach- machine vision for detecting cars and tracking pedestrians, and so on, and we couldn't get the reliability of detection and tracking up to a high enough, uh, particu- level, particularly in bad weather conditions, uh, nighttime, rain, fog.

Lex Fridman 25:11 Good enough for demos, but perhaps not good enough to cover the general ca- uh, the- the general operation (inaudible)

Stuart Russell 25:16 Yeah. So the thing about driving is, you know, so suppose you're a taxi driver, you know, and you drive every day, eight hours a day for 10 years, right? That's 100 million seconds of driving, you know? And any one of those seconds, you could make a fatal mistake.

Lex Fridman 25:29 Yeah.

Stuart Russell 25:29 So you're talking about eight-nines of reliability, right? Now, if your vision system only detects 98.3% of the vehicles, right? (laughs)

Lex Fridman 25:42 Hm.

Stuart Russell 25:42 Then that's sort of, you know, one and a bit nines of reliability.

Lex Fridman 25:45 Yeah.

Stuart Russell 25:46 So you have another seven orders of magnitude to go. And, um, a- and this is what people don't understand. They- they think, "Oh, because I had a successful demo, I'm pretty much done." But you're n- you're not even within seven orders of magnitude of being done, and that's the difficulty, and it's- it's not the, "Can I follow a white line?" That's not the problem, right? We can follow a white line all the way across the country. But it's the- it's the weird stuff that happens, it's, you know...

Lex Fridman 26:21 It's all the edge cases, yeah.

Stuart Russell 26:22 The edge case, other drivers doing weird things. Um, you know, so if you talk to Google, right? So they had, um, actually very classical architecture where, you know, you had machine vision which would detect all the other cars, and pedestrians, and the white lines, and the road signs, and then basically that was fed into a logical database.

Lex Fridman 26:45 Mm-hmm.

Stuart Russell 26:45 And then you had a classical 1970s rule-based expert system, um, telling you, "Okay, if you're in the middle lane, and there's a bicyclist in the right lane who is signaling this, then- then- then- then you do that."

Lex Fridman 26:58 Yeah.

Stuart Russell 26:59 Right? And what they found was that every day, they'd go out, and there'd be another situation that the rules didn't cover, you know? So they- they'd come to a traffic circle, and there's a little girl riding her bicycle the wrong way around the traffic circle. Okay, what do you do? "We don't have a rule." "Oh my god, okay, stop." And then you- you know, they'd come back and add more rules, and they just found that this was not really converging. And, um... And if you think about it, right, how d- how do you deal with an unexpected situation? Meaning one that you've never previously encountered and the sort of the- the reasoning required to figure out the solution for that situation has never been done, and so it doesn't match any, uh, previous situation in terms of the kind of reasoning you have to do? Well, you know, in chess programs this happens all the time, right? You're constantly coming up with situations you haven't seen before-

Lex Fridman 27:53 Hmm. That's right.

Stuart Russell 27:53 ... and you have to reason about them, and you have to think about, "Okay, here are the possible things I could do, here are the outcomes, here's how desirable the outcomes are," and then pick the right one. You know, in the '90s we were saying, "Okay, this is how you're gonna have to do automated vehicles. They're gonna have to have lookahead capability." But the lookahead for driving is more difficult than it is for chess because-

Lex Fridman 28:14 'Cause of humans.

Stuart Russell 28:15 ... the other... Right, there's humans, and they are less predictable than a op-

Lex Fridman 28:19 Than chess pieces.

Stuart Russell 28:19 ... than a... Well, than w- You have an opponent in chess who's also somewhat unpredictable.

Lex Fridman 28:26 Hm.

Stuart Russell 28:26 But for example, in chess, you always know the opponent's intention. They're trying to beat you, right? Whereas in driving, you don't know, is this guy trying to turn left? Or has he just forgotten to turn off his turn signal?

Lex Fridman 28:38 Mm-hmm.

Stuart Russell 28:39 Or is he drunk? Or is he, you know, changing the channel on his radio? Or whatever it might be, you gotta try and figure out the mental state, the intent of the other-

Lex Fridman 28:49 Mm-hmm.

Stuart Russell 28:49 ... drivers to forecast the possible evolutions of their trajectories, and then you gotta figure out, "Okay, which is the trajectory for me that's going to be safest?" And those all interact with each other-

Lex Fridman 29:02 Mm-hmm.

Stuart Russell 29:02 ... because the other drivers are gonna react to your trajectory, um, and so on. So, you know, they've got the classic merging onto the freeway problem where you're kind of racing a vehicle that's already on the freeway, and you're... are you gonna pull ahead of them? Or are you gonna let them go first and pull in behind? And- and you get this sort of uncertainty about who's going first. So all those kinds of things mean that you need a decision-making architecture that's very different from either a rule-based system or, it seems to me, a- kind of an end-to-end neural network system. You know, so just as AlphaGo is pretty good when it doesn't do any lookahead, but it's way, way, way, way better when it does, I think the same is gonna be true for driving. You can have a driving system that's pretty good when it doesn't do any lookahead, but that's not good enough. Um, you know, and we've already seen multiple deaths caused by poorly designed, uh, machine learning algorithms that don't really understand what they're doing.

Lex Fridman 30:09 Yeah, on- on several levels. I think, uh, so on the perception side, uh, there's mistakes being made by those algorithms where the perception is very shallow. On the planning side, the lookahead like you said, and the thing that we come- come up against that's really interesting when you try to deploy systems in the real world is...... you can't think of an artificial intelligence system as a thing that responds to the world always. You have to realize that it's an agent that others will respond to as well. So, in order to drive successfully, you can't just try to do obstacle avoidance.

Stuart Russell 30:47 Right.

Lex Fridman 30:48 You-

Stuart Russell 30:48 You can't pretend that you're invisible. (laughs) Right?

Lex Fridman 30:50 (laughs)

Stuart Russell 30:50 You're the invisible car.

Lex Fridman 30:52 Right. So-

Stuart Russell 30:52 Uh, it doesn't work that way.

Lex Fridman 30:53 I mean, but y- you have to assert... y- others have to be scared of you. Just, we're all s- there's this tension. There's this game... So if w- we study a lot of work with pedestrians.

Stuart Russell 31:03 Mm-hmm.

Lex Fridman 31:04 If you approach pedestrians as purely an obstacle avoidance, so you're, you're doing look ahead as in modeling the intent, then you're, you're- they're not going to... they're going to take advantage of you. They're not going to respect you at all. There has to be a tension, a fear, a some amount of uncertainty. That's how we have creat- we keep on-

Stuart Russell 31:24 Yeah. Or, or, or at least just a kind of a, a resoluteness. (laughs)

Lex Fridman 31:27 Right. Yes, a cert-

Stuart Russell 31:27 Let's put it that way.

Lex Fridman 31:28 ...tainess.

Stuart Russell 31:28 You ha- you have to display a certain amount of resoluteness. You can't, you can't be too tentative.

Lex Fridman 31:32 Yeah.

Stuart Russell 31:33 And, uh, yeah. So the... right. The, uh, the solutions then become pretty complicated, right?

Lex Fridman 31:40 Right.

Stuart Russell 31:40 You get into game theoretic-

Lex Fridman 31:42 Yes.

Stuart Russell 31:42 ...analyses and... So w- you know, at Berkeley now we're working a lot on this kind of interaction between machines and humans. Uh-

Lex Fridman 31:52 And that's exciting. Yep.

Stuart Russell 31:54 And, uh, so my colleague, uh, Anca Dragan, actually... You know, if you, if you formulate the problem game theoretically, and you just let the system figure out the solution-

Lex Fridman 32:07 Mm-hmm.

Stuart Russell 32:07 ... you know, it does interesting unexpected things. Like sometimes at a stop sign, if no one is going first, right, the car will actually back up a little.

Lex Fridman 32:18 (laughs)

Stuart Russell 32:18 Right?

Lex Fridman 32:19 Interesting.

Stuart Russell 32:19 Just to indicate to the other cars that they should go.

Lex Fridman 32:22 (laughs)

Stuart Russell 32:22 Uh, and that's something it invented entirely by itself.

Lex Fridman 32:25 That's interesting.

Stuart Russell 32:26 Right? There was, you know... We didn't say, "This is the language of communication at stop signs." It figured it out.

Lex Fridman 32:32 That's really interesting. Uh, so let me, one, just step back for a second. Just, this beautiful philosophical notion. So Pamela McCorduck in 1979 wrote, "AI began with the ancient wish to forge the gods." So when you think about the history of our civilization, uh, do you think that there is an inherent desire to create, uh, let's not say gods, but to create super intelligence? Is it inherent to us? Is it in our genes that the natural arc of human civilization is to create things that are of greater and greater power and perhaps, um, you know, echoes of ourselves? So to create the gods as, as, uh, Pamela said? Is that-

Stuart Russell 33:25 Maybe. I mean, you know, we're all, we're all individuals, but certainly we see over and over again in history, uh, individuals who thought about this possibility.

Lex Fridman 33:39 Hopefully we're n- I'm not being too philosophical here.

Stuart Russell 33:42 Mm-hmm.

Lex Fridman 33:42 But if you look at the arc of this, you know, where this is going, and we'll talk about AI safety, we'll talk about greater and greater intelligence, do you see that there... In, in, when you created the Othello program and you felt this excitement, what was that excitement? Was it excitement of a tinker who created something cool like a clock? Or was there a magic... or was it more like a child being born? Y- you know-

Stuart Russell 34:11 Yeah. So I mean, I s- I certainly understand that viewpoint. And if you look at, um, the Lighthill Report, um, which was com- So in the '70s there was a lot of controversy in the UK about AI and, you know, whether it was for real, and how much the mone- money the government should invest and...

Lex Fridman 34:30 Mm-hmm.

Stuart Russell 34:33 So it's a lo- a long story, but the government commissioned a report by, by Lighthill who was a physicist and, uh, he wrote a very damning report about AI which I think was the point. (laughs) Uh, and, uh, he said that, that these are, uh, you know, frustrated men who unable to have children-

Lex Fridman 34:55 (laughs)

Stuart Russell 34:55 ... would like to, to create-

Lex Fridman 34:58 Interesting.

Stuart Russell 34:58 ... uh, and, you know, create, uh, life, um, y- you know, as a kind of replacement.

Lex Fridman 35:03 That's a good turn. (laughs)

Stuart Russell 35:03 You know, which I, uh, which I think is really pretty unfair. But there is... I mean there... there is a kind of magic I would say. You... when you, you build something, and, and what you're building in is really just... uh, you're building in some understanding of the principles of learning and decision-making. And to see those principles actually then turn into intelligent behavior in, in specific situations, it's an incredible thing. And, uh, you know, that is, uh, naturally going to make you think, "Okay, where does this end?"

Lex Fridman 36:00 And so there's a... there's magical, optimistic views of where it ends, whatever your view of optimism is, whatever your view of utopia is. It's probably different for everybody.

Stuart Russell 36:11 Yep.

Lex Fridman 36:12 But you've often talk about, uh, concerns you have of how things may go, uh, wrong. So, uh, I've talked to, uh, Max Tegmark, uh, there's a lot of interesting ways to think about AI safety. You're one of the m- seminal people thinking about this problem among sort of being in the weeds of actually, uh, solving specific AI problems. You're also thinking about the big picture of where we're going.

Stuart Russell 36:42 Mm-hmm.

Lex Fridman 36:42 So-Can you talk about several elements of it? Let's just talk about maybe the control problem, so this idea of losing ability to control the behavior in a- of a AI system. So w- how do you see that? How do you see that coming about? What do you think we can do, uh, to manage it?

Stuart Russell 37:06 Well, so it- it doesn't take a genius to realize that if you make something that's smarter than you, you might have a problem (laughs) . You know, and Turing, uh, Alan Turing, you know, wrote about this and gave lectures about this, you know, in, I think, 1951. He did a lecture on the radio, and, uh, he basically says, you know, once the machine thinking method starts, uh, you know, very quickly, they'll outstrip humanity. And, uh, you know, if we're lucky, we might be able to... I think he says i- if... "We may be able to turn off the power at strategic moments, but even so, our species would be humbled."

Lex Fridman 37:51 Yeah, you-

Stuart Russell 37:52 And actually, I think he was wrong about that, right? (laughs) 'cause you- 'cause you got- you know, if it's a sufficiently intelligent machine, it's not gonna let you switch it off. Uh-

Lex Fridman 37:58 So, uh-

Stuart Russell 37:58 ... it's actually in competition with you.

Lex Fridman 38:00 So, what do you think is meant, just for a quick tangent, if we shut off this super intelligent machine, that our species would be humbled?

Stuart Russell 38:12 I think he means that we would realize that we are inferior, right? That we- we only survive by the skin of our teeth because we happened to get to the off switch (laughs) , you know, just in-

Lex Fridman 38:24 That was a close call.

Stuart Russell 38:25 ... just in time, uh, you know?

Lex Fridman 38:26 Yeah.

Stuart Russell 38:26 And if- and if we hadn't, then, uh, we would have lost control over the Earth.

Lex Fridman 38:31 So, do you... Are you more worried when you think about this stuff about super intelligent AI, or are you more worried about super powerful AI that's not aligned with our values? So, the paperclip, uh, scenario is kind of-

Stuart Russell 38:45 Mm-hmm.

Lex Fridman 38:46 ... uh-

Stuart Russell 38:47 I think, um... So, the main problem I'm working on is- is the control problem, the- the problem of machines pursuing objectives that are, as you say, not aligned with human objectives. And- and this has been- this has been the way we've thought about AI since the beginning. You- you build a machine for optimizing, and then you put in some objective, and it optimizes, right? And- and, um, you know, we- we can think of this as the- the King Midas problem, right? 'Cause if, you know, so King Midas put in this objective, right? "Everything I touch should turn to gold." And the gods, you know, that's like the machine, they said, "Okay, done." You know, "You now have this power." And of course, his food and his drink and his family all turn to gold, and then he dies of misery and starvation. And, um, this is... You know, it's- it's a warning. It's- it's a failure mode that pretty much every culture, uh, in history has had some story along the same lines. You know, there's the- the genie that gives you three wishes and-

Lex Fridman 39:58 Yeah.

Stuart Russell 39:58 ... you know, the third wish is always, you know, "Please undo the first two wishes 'cause I messed up." And, uh, you know, and when Arthur Samuel wrote his chess, uh, his checker-playing program, which learned to play checkers considerably better than Arthur Samuel could play and actually reached a pretty decent standard, uh, Norbert Wiener, who was a- one of the major mathematicians of the 20th century, he's sort of the father of modern automation control systems, you know, he saw this and he basically extrapolated, uh, you know, as Turing did and said, "Okay, uh, this is how we could lose control." And, uh, specifically that we have to be certain that the purpose we put into the machine is the purpose which we really desire. And the problem is we can't do that, right?

Lex Fridman 40:58 You- you mean we're not... it's, uh, very difficult, uh, to encode, to- to- to put our values on paper is really difficult? Or are you just saying it's impossible?

Stuart Russell 41:06 Uh, theor- (laughs)

Lex Fridman 41:09 (laughs) The line is gray between the two.

Stuart Russell 41:10 So- so, th- theoretically, it's possible but, uh, i- in practice, it's extremely unlikely that we could specify correctly in advance the full range of concerns of humanity. Uh, yeah.

Lex Fridman 41:25 The... You talked about, uh, cultural transmission of values I think is how humans-to-human transmission of values happens, right? Uh...

Stuart Russell 41:32 Well, we learn... You know, I mean, w- as we grow up, we learn about the values that matter, how things- how things should go, what is reasonable to pursue and what isn't reasonable to pursue. Uh, and it's-

Lex Fridman 41:44 Do you think machines can learn in the same kind of way?

Stuart Russell 41:46 Yeah, so I- I think that, um, what we need to do is to get away from this idea that you build an optimizing machine and then you put the objective into it because if it's possible that you might put in a wrong objective, and we already know this is possible 'cause it's happened lots of times-

Lex Fridman 42:05 Mm-hmm.

Stuart Russell 42:06 ... right? That means that the machine should never take an objective that's given as gospel truth because once it takes the mach- the- the objective as gospel truth, all right, then it believes that whatever actions it's taking in pursuit of that objective are the correct things to do. So, you could be jumping up and down and saying, "N- you know, no, no, no, you're gonna destroy the world!" But the machine knows what the true objective is and it's pursuing it.

Lex Fridman 42:36 Mm-hmm.

Stuart Russell 42:36 And tough luck to you.

Lex Fridman 42:37 (laughs)

Stuart Russell 42:38 You know, and this is not restricted to AI, right? This is, you know, I think...... many of the 20th century technologies, right? So in statistics, you, you minimize a loss function. A loss function is exogenously specified. In control theory, you minimize a cost function. In operations research, you maximize a reward function.

Lex Fridman 42:57 Mm-hmm.

Stuart Russell 42:57 Uh, and so on. So in all these disciplines, this is how we conceive of the problem, and it's the wrong problem, because we cannot specify with certainty the correct objective, right? We need uncertainty, we need the machine to be uncertain about-

Lex Fridman 43:16 That's objective.

Stuart Russell 43:17 ... what it is that it's supposed to be maximizing.

Lex Fridman 43:19 It's... My favorite idea of yours, uh, I've heard you say somewhere, uh, well, I shouldn't pick favorites, but it just sounds beautiful of, uh, we need to teach machines humility. Uh. (laughs)

Stuart Russell 43:30 Yeah. I mean, that, that's-

Lex Fridman 43:31 It's a, it's a beautiful way to put it. I, I, I love it. Uh...

Stuart Russell 43:34 Um, that they're humble-

Lex Fridman 43:35 (laughs)

Stuart Russell 43:36 ... in, in that they know-

Lex Fridman 43:37 Humble AI.

Stuart Russell 43:37 ... they know that they don't know-

Lex Fridman 43:39 Right.

Stuart Russell 43:39 ... what it is they're supposed to be doing. And, uh, and that those, those objectives, I mean, they exist. They're within us, but we may not be able to explicate them. We may not even know, uh, uh, you know, how we want our future to go. Right?

Lex Fridman 43:57 So... Exactly.

Stuart Russell 43:58 And the machine, you know, a, a, a machine that's uncertain is going to be deferential to us. So if we say, "Don't do that," well, now the machine's learned something a bit more about our true objectives. Because w- something that it thought was reasonable in pursuit of our objective turns out not to be, so now it's learned something. So it's going to defer because it wants to be doing what we really want. And, um, you know, that, that point I think is absolutely central to solving the control problem.

Lex Fridman 44:32 Yeah.

Stuart Russell 44:32 Uh, and it's a different kind of AI. When you, when you take away this idea that the objective is known, then, in fact, a lot of the theoretical frameworks that we're so familiar with, you know, Markov decision processes, uh, goal-based planning, uh, you know, standard game tree search, all of these, uh, techniques actually become inapplicable.

Lex Fridman 45:00 (laughs)

Stuart Russell 45:01 Uh, and you get a more complicated problem because, because now the interaction with the human becomes part of the problem. Because the human by making choices is giving you more information about the true objective. And that information helps you achieve the objective better. And so, that really means that you're mostly dealing with game theoretic problems, where you've got the machine and the human and they're coupled together, uh, rather than a machine going off by itself with a fixed objective.

Lex Fridman 45:39 Which is fascinating on the machine and the human level, that we... When you don't have an, an objective means you're together coming up with an objective. I mean, uh, there's a lot of philosophy that, you know, you could argue that life doesn't really have meaning. We, we together agree on what gives it meaning, and we kind of culturally create, uh, things that give why the heck we are on this Earth anyway. Uh, we together-

Stuart Russell 46:07 Yeah.

Lex Fridman 46:07 ... as a society create that meaning, and you have to learn that objective. And one of the biggest... I, I thought that's where you were gonna go for a second. Uh, one of the biggest troubles we run into outside of statistics and machine learning and AI, in just human civilization, is when, uh, you look at, uh... I came from the S- I was born in the Soviet Union. And the history of the 20th century, uh, we ran into the most trouble, us humans, when there was a, a, a h- a certainty about the objective. And you do whatever it takes to achieve that objective, whether you're talking about Germany or communist Russia.

Stuart Russell 46:44 Mm-hmm.

Lex Fridman 46:44 Uh, all...

Stuart Russell 46:45 Yeah, I, and, and, and, and-

Lex Fridman 46:45 You, you get into trouble with humans' faith-

Stuart Russell 46:46 ... as I would say with, uh, you know, corporations. In fact, some people argue that, you know, we don't have to look forward to a time when AI systems take over the world. They already have, and they're called corporations. Right? The corporations happen to be, uh, using people as components right now.

Lex Fridman 47:05 Mm-hmm.

Stuart Russell 47:05 Um, but they are effectively algorithmic machines, and they're optimizing an objective, which is quarterly profit, that isn't aligned with overall wellbeing of the human race. And they are destroying the world. I mean-

Lex Fridman 47:19 Right.

Stuart Russell 47:19 ... they're, they are primarily responsible for our inability to tackle climate change.

Lex Fridman 47:24 Right.

Stuart Russell 47:25 So, I think that's one way of thinking about what's going on with, uh, with corporations. But, uh, I think the point you're making i- is valid, that there, there are many systems in the real world where we've sort of prematurely fixed on the objective and then decoupled the, uh, the machine from those that it's supposed to be serving. Um, and I think you see this with government. Right? Government is supposed to be a machine that serves people, but instead, it tends to be taken over by people who have their own objective-

Lex Fridman 48:04 Mm-hmm.

Stuart Russell 48:04 ... uh, and use government to optimize that objective regardless of what people want.

Lex Fridman 48:08 Do, do you have... Do you find appealing the idea of almost, uh, arguing machines where you have multiple AI systems with a clear fixed objective? We have in government the red team and the blue team-

Stuart Russell 48:21 Mm-hmm.

Lex Fridman 48:21 ... that are very fixed on their objectives and they argue and it kinda... Uh, many would disagree, but it kinda seems to make it work somewhat? That (laughs) , uh, the, the, the duality of it... Uh, I, th- a- I know a lot of ...

Stuart Russell 48:36 Yeah. (laughs)

Lex Fridman 48:37 ... ... we'll disagree. Okay, let's go 100 years back when there was... still was going on-

Stuart Russell 48:40 Mm-hmm.

Lex Fridman 48:40 ... or at the founding of this country. There was disagreements, and that disagreement is where, uh... So it was a balance between certainty and forced humility because the power was distributed.

Stuart Russell 48:54 Yeah, I think that the, um, the- the nature of debate and- and disagreement argument takes, uh, as a premise, the idea that you could be wrong, right? Which means that y- y- you're not necessarily absolutely convinced that your objective is- is the correct one. Right? Um, if you were absolutely convinced, there'd be no point in having any discussion or argument because you would never change your mind-

Lex Fridman 49:22 Mm-hmm.

Stuart Russell 49:23 ... um, and there wouldn't be any- any sort of synthesis or- or anything like that. So- so I think you can think of argumentation as a- as an implementation-

Lex Fridman 49:33 Mm-hmm.

Stuart Russell 49:33 ... of a form of uncertain reasoning, uh, and, um, you know, I- I've been reading recently about utilitarianism and the history of efforts to define, in a sort of clear mathematical way, a... if you like, a formula for moral or political decision-making. And it's really interesting that the parallels between the philosophical discussions going back 200 years and what you see now in discussions about existential risk because-

Lex Fridman 50:10 Mm-hmm.

Stuart Russell 50:11 ... y- y- it's almost exactly the same. So someone would say, "Okay, well, here's a formula for how we should make decisions." Right? So utilitarianism is roughly, you know, each person has a utility function, and then we make decisions to maximize the sum of everybody's utility.

Lex Fridman 50:27 Mm-hmm.

Stuart Russell 50:28 Right? And then people point out, "Well, you know, in that case, the best policy is one that leads to the enormously vast population, all of whom a- are living a life that's barely worth living."

Lex Fridman 50:42 Mm-hmm.

Stuart Russell 50:43 Right? And- and this is called the repugnant conclusion. And, uh, you know, another version is, you know, that we- we should maximize pleasure, and that's what we mean by utility. And then you'll get people effectively saying, "Well, in that case, you know, we might as well just have everyone hooked up to a heroin drip."

Lex Fridman 51:02 Yeah.

Stuart Russell 51:02 You know, and they didn't use those words, but that debate, you know, was happening in the 19th century as it is now, uh, about AI, that if we get the formula wrong, you know, we're going to have AI systems working towards an outcome that, in retrospect, would be exactly wrong.

Lex Fridman 51:22 Do you think there's... uh, it has beautifully put. So th- the echoes are there, but do you think, um, I mean if you look at Sam Harris'... (laughs) our imagination worries about the AI version of that because of the, uh, the- the speed at which the things going wrong in the utilitarian context could happen.

Stuart Russell 51:45 Yeah.

Lex Fridman 51:46 Is that... is that a worry for you?

Stuart Russell 51:47 Yeah, I- I think that, you know, i- in most cases, not in all, but, y- you know, if- if we have a wrong political idea, you know, we see it starting to go wrong and we're, you know, we're- we're not completely stupid and so we say, "Okay, that was... maybe that was a mistake. Uh, let's try something different." And- and also we're very slow and inefficient about implementing these things and so on. So you have to worry when you have corporations or political systems that are extremely efficient.

Stuart Russell 52:19 But when we look at AI systems or even just computers in general, right? They have this different characteristic from ordinary human activity in the past. So let's say you were a surgeon, you had some idea about how to do some operation-

Lex Fridman 52:36 Mm-hmm.

Stuart Russell 52:37 ... right? Well, and- and let's say you were wrong, right?That that way of doing the operation would mostly kill the patient.

Lex Fridman 52:43 Mm-hmm.

Stuart Russell 52:43 Well, you'd find out pretty quickly, like after three, maybe three or four tries, right? But that isn't true for pharmaceutical companies because they don't do three or four operations, they- they manufacture three or four billion pills and they sell them and then they find out maybe six months or a year later that, oh, people are dying of heart attacks or getting cancer from this drug. And so that's why we have the FDA, right? Because of the scalability of pharmaceutical production. And, you know, and there have been some unbelievably bad episodes in the history of pharmaceuticals and- and adulteration of- of products and so on that- that have killed tens of thousands or paralyzed hundreds of thousands of people. Now with computers, we have that same scalability problem that you can sit there and type for I equals one to five billion do, right?

Lex Fridman 53:49 Yeah.

Stuart Russell 53:49 And all of a sudden you're having an impact on a global scale. And yet we have no FDA, right? There's absolutely no controls at all over what a bunch of undergraduates with too much caffeine can do to the world. And, uh, you know, we look at what happened with Facebook, well social media in general, and click-through optimization. So you have a simple feedback algorithm that's trying to just optimize click-through, right?

Lex Fridman 54:19 Mm-hmm.

Stuart Russell 54:19 That sounds reasonable, right? Because you don't want to be feeding people ads that they don't care about and not interested in. And you might even think of that process as simply adjusting the- the feeding of ads or news articles or whatever it might be to match people's preferences, right? Which sounds like a good idea. But in fact, that isn't how the algorithm works, right? You make more money... the algorithm makes more money if...... it can better predict what people are going to click on, 'cause then it can feed them exactly that, right? So the way to maximize click-through is actually to modify the people-

Lex Fridman 55:10 Right.

Stuart Russell 55:10 ... to make them more predictable. Uh, and one way to do that is to feed them information which will change their behavior and preferences towards extremes that make them predictable. You know, whatever is the nearest extreme or the nearest predictable point, uh, that's where you're going to end up. And the machines will force you there. Now, and- and I think there's a reasonable argument to say that this, among other things, is contributing to the destruction of democracy in the world. And where was the oversight of this process? Where were the people saying, "Okay, you would like to apply this algorithm to five billion people on the- on the face of the Earth. Can you show me that it's safe? Can you show me that it won't have various kinds of negative effects?" No, there was no one asking that question. There was no one placed between, you know, the- the undergrads with too much caffeine and the human race.

Lex Fridman 56:15 Well, it's-

Stuart Russell 56:15 It's just they just did it.

Lex Fridman 56:17 And s- but some, uh, w- way outside the scope of my knowledge, so economists would argue that the, what- what is it? The invisible hand, so the- the capitalist system, it was the oversight. So if you're going to corrupt society with whatever decision you make as a company, then that's going to be reflected in people not using your product. S- sort of one-

Stuart Russell 56:37 (laughs)

Lex Fridman 56:37 That's one model of oversight. So...

Stuart Russell 56:39 We- we shall see, but, you know-

Lex Fridman 56:40 We shall see, yeah.

Stuart Russell 56:40 ... in the meantime, you know, that, but you- you might even have broken the, uh, the political system that enables capitalism to function.

Lex Fridman 56:50 Well, you've changed it. (laughs) So-

Stuart Russell 56:53 We shall see, yeah.

Lex Fridman 56:54 Cha- change is, uh, often painful. So my- my question is, uh, absolutely, it's f- it's fascinating. You're absolutely right that there was zero oversight on algorithms that could have a profound, civilization-changing, uh, effect. So do you think it's possible? I mean, I'm- I haven't... Have you seen government? So do you think it's possible to create regulatory bodies, uh, oversight over AI algorithms, which are inherently such cutting edge set of, uh, ideas and technologies?

Stuart Russell 57:31 Yeah, but I think it takes time to figure out what kind of oversight, what kinds of controls. I mean, it took time to design the FDA regime. Uh, you know, and some people still don't like it and they wanna fix it. Um, and I think there are clear ways that it could be improved. But the whole notion that you have stage one, stage two, stage three, and here are the criteria for what you have to do to pass a stage one trial, right?

Lex Fridman 57:59 Yes.

Stuart Russell 57:59 We haven't even thought about what those would be for algorithms. So, I mean, I- I think there are, there are things we could do right now with regard to bias, for example. We- we have a pretty good technical handle on, uh, how to detect algorithms that are propagating bias that exists in datasets, um, how to de-bias those algorithms.

Lex Fridman 58:23 Mm-hmm.

Stuart Russell 58:25 Uh, and- and even what it's gonna cost you to do that. So I think we could start having some standards, um, on that. I think there are, there are things to do with impersonation and falsification that we could, we could work on. So, uh...

Lex Fridman 58:41 Deepfakes, yep.

Stuart Russell 58:42 Uh...

Lex Fridman 58:43 Or, uh, uh, the in-

Stuart Russell 58:45 A very simple point. So impersonation is, is a machine, uh, acting as if it was a person.

Lex Fridman 58:50 Mm-hmm.

Stuart Russell 58:51 I can't see-

Lex Fridman 58:52 Right.

Stuart Russell 58:52 ... a real justification for why we shouldn't insist that machines self-identify as machines.

Lex Fridman 58:59 Mm-hmm.

Stuart Russell 59:00 Uh, you know, where- where is the social benefit in, in, uh, fooling people into thinking that this is really a person when it isn't? You know, I- I don't mind if it uses a human-like voice that's easy to understand, that's fine, but it should just say, "I'm a machine," in some sh- some form.

Lex Fridman 59:20 And not many people are speaking to that, I would think, relatively obvious fact, so I think most people agree with you.

Stuart Russell 59:26 Yeah, I mean there is actually a law in California that bans impersonation, but only in certain restricted circumstances. So, uh, for the purpose of, uh, engaging in a fraudulent transaction and for the purpose of modifying someone's voting behavior. Uh, so those are, those are the circumstances where, uh, machines have to self-identify. Um, but I think there's, you know, arguably it should be in all circumstances. Um, and then when you, when you talk about deepfakes, you know, we're just at the beginning, but already it's possible to make a movie of anybody saying anything-

Lex Fridman 1:00:08 Yeah.

Stuart Russell 1:00:08 ... uh, in ways that are pretty hard to detect. Yeah.

Lex Fridman 1:00:11 Including yourself because you're on camera now and your voice is coming through with high resolution.

Stuart Russell 1:00:15 Yeah, so- so you could-

Lex Fridman 1:00:17 (laughs)

Stuart Russell 1:00:17 ... take what I'm saying and replace it with pretty much anything else you wanted me to be saying.

Lex Fridman 1:00:20 Yeah.

Stuart Russell 1:00:21 Uh, and- and even it would change my lips and-

Lex Fridman 1:00:23 Yeah.

Stuart Russell 1:00:23 ... ex- facial expressions to fit. And, uh, there's actually not much in the way of, uh, real legal protection against that. I think in the commercial area you could say, "Yeah, that's, uh, you're using my brand," and so on. That there are, there are rules about that. But in the political sphere, I think it's, uh, at- at the moment, it's, you know, anything goes. So-

Lex Fridman 1:00:50 Let me-

Stuart Russell 1:00:51 ... that- that could be really, really damaging.

Lex Fridman 1:00:54 And let me just, uh, try to make not an argument, but try to look back at history...... and, say something dark in essence is while regulation seems to be, oversight seems to be exactly the right thing to do here, it seems that human beings, what they naturally do is they wait for something to go wrong. If you're talking about nuclear weapons, you can't talk about nuclear weapons being dangerous until somebody actually, like the United States drops the bomb, or Chernobyl melting. Do you think we will have to wait f- for things going wrong in a way that's obviously damaging to society? Not an existential risk, but obviously damaging, or do you have faith that...

Stuart Russell 1:01:43 I, I hope not but, I mean it, it... I, I think we do have to look at history and when... So the two examples you gave, nuclear weapons and nuclear power are very, very interesting because, you know, nuclear weapons, we knew in the early years of the 20th century that atoms contained a huge amount of energy, right? We had E=MC², we knew the, the mass differences between the different atoms and their components, and we knew that you might be able to make an incredibly powerful explosive. So H.G. Wells wrote science fiction book, I think, in 1912.

Lex Fridman 1:02:25 Mm-hmm.

Stuart Russell 1:02:26 Frederick Soddy, who was the guy who discovered isotopes, he's a Nobel Prize winner, he gave a speech in 1915 saying that, you know, one pound of this new explosive would be the equivalent of 150 tons of dynamite, which turns out to be about right. And, uh, you know, c- this was in World War I, right? So he was imagining how much worse the World, uh, War would be, uh, if we were using that kind of explosive. But the physics establishment simply refused to believe that these things could be made.

Lex Fridman 1:03:04 Including the people who were making it. (laughs)

Stuart Russell 1:03:06 Well, so they were doing the nuclear physics...

Lex Fridman 1:03:09 I mean, eventually were the ones who made it-

Stuart Russell 1:03:11 And, so-

Lex Fridman 1:03:11 ... you talk about Fermi or whoever.

Stuart Russell 1:03:13 Well, so up to, um... The, the development, uh, was, was mostly theoretical, so it was people using sort of primitive kinds of particle acceleration and, and doing experiments, uh, at the, at the level of single particles or collections of particles. They, they, they weren't yet thinking about how to actually make a bomb or anything like that, but they knew the energy was there and they figured if they understood it better, uh, it might be possible. But the physics establishment, their view, and I think because they did not want it to be true, their view was that it could not be true, uh, that this could not provide a way to make a super weapon. And, um, you know, there was this famous speech, um, ru- given by Rutherford, who was the sort of leader of nuclear physics and, um, it was on September 11th, 1933 and he, he said, you know, "Anyone who talks about the possibility of obtaining energy from transformation of atoms is talking complete moonshine." And, uh, the next, uh, the next morning Leo Szilard read about that speech and then invented the nuclear chain reaction. And so as soon as he invented... As soon as he had that idea that you could make a chain reaction with neutrons, because neutrons were not repelled by the nucleus so they could enter the nucleus and then continue the reaction, as soon as he has that idea, he instantly realized that the world was in deep doodoo.

Lex Fridman 1:04:50 (laughs)

Stuart Russell 1:04:51 Uh, because this is 1933, right? You know, Hitler had recently come to power in Germany, Szilard was in London, uh, and eventually became a refugee and, uh, and came to the US. And the, um... In the process of, of having the idea about the chain reaction, he figured out basically how to make a bomb and also how to make a reactor, and he patented the reactor in 1934, but because of the s- situation, the great power conflict situation that he could see happening, um, he kept that a secret. And so, um, between then and the beginning of World War II, people were working, including the Germans, on how to actually create neutron sources, right? What specific fission reactions would produce neutrons of the right energy to continue the reaction, and that was demonstrated in Germany, I think, in 1938, if I remember correctly. The first, uh, nuclear weapon patent was 1939 by the French, um, so this was actually-

Lex Fridman 1:06:11 Interesting, isn't it?

Stuart Russell 1:06:12 ... uh, you know, this was actually going on, you know, well before World War II really got going. And then, you know, the British probably had the most advanced capability in this area, but for safety reasons among others and plus just sort of just resources, they moved the program from Britain to the US, and then that became Manhattan Project. Uh, so the, the, the reason why we couldn't have any kind of oversight of nuclear weapons and nuclear technology was because we were basically already in, uh, an arms race and a war, and um...

Lex Fridman 1:06:53 But you, you mentioned then in the 20s and 30s, so what are the echoes... (laughs) The way you've described this story, I mean, there's clearly echoes. Why do you think most AI researchers...... folks who are really close to the metal. They really are not concerned about AI. They don't think about it, uh, whether it's they don't want to think about it, it's... But what are the... Yeah, why do you think that is? What are the echoes of the nuclear situation to the current AI situation and, uh, what can we do about it?

Stuart Russell 1:07:27 Mm-hmm. I, I think there is a, you know, a kind emot- motivated cognition, which is a... A term in psychology means that you believe what you would like to be true, uh, rather than what is true. And, uh, you know, it's- it's unsettling to think that what you're working on might be the end of the human race, obviously. So you would rather instantly deny it and come up with some reason why it couldn't be true. And, uh, you know, I have... I collected a, a long list of reasons that extremely intelligent, competent AI scientists have come up with for why we shouldn't worry about this. You know, for example, calculators are superhuman at arithmetic and they haven't taken over the world, so there's nothing to worry about.

Lex Fridman 1:08:20 (laughs)

Stuart Russell 1:08:21 Well, okay, my five-year-old, you know, could have figured out why that was, uh, uh, an unreasonable (laughs) and, and really quite weak argument. Um, you know, another one was, uh, you know... W- while it's theoretically possible that you could have a superhuman AI destroy the world, you know, it's also theoretically possible that a black hole could materialize-

Lex Fridman 1:08:47 Mm-hmm.

Stuart Russell 1:08:47 ... right next to the earth and destroy h- uh, humanity. I mean, yes, it's theoretically possible, quantum theoretically, extremely unlikely that it would just materialize right there. Um, but that's a completely bogus analogy because, you know, if the whole physics community on Earth was working to materialize a black hole in near-Earth orbit, right? Wouldn't you ask them, "Is that a good idea?"

Lex Fridman 1:09:11 (laughs)

Stuart Russell 1:09:11 "Is that gonna be safe?" You know, "What if you succeed?"

Lex Fridman 1:09:14 Right.

Stuart Russell 1:09:14 Right? And that's the thing, right? The AI community has sort of refused to ask itself, what if you succeed?

Lex Fridman 1:09:23 Yeah.

Stuart Russell 1:09:24 And initially, I think that was because it was too hard, but, you know, Alan Turing asked himself that and he said we'd be toast.

Lex Fridman 1:09:33 (laughs)

Stuart Russell 1:09:34 Right? If we were lucky, we might be able to switch off the power, but probably we'd be toast.

Lex Fridman 1:09:38 But there's also an aspect that because we're not exactly sure what the future holds, it's not clear exactly, so technically, what to worry about, sort of how-

Stuart Russell 1:09:51 Mm-hmm.

Lex Fridman 1:09:52 ... things go wrong. And so, uh, there is something, it feels like, maybe you can correct me if I'm wrong, but th- there's something paralyzing about worrying about something that logically is inevitable, but you don't really know what that will b- look like.

Stuart Russell 1:10:10 Yeah, I think that's, that's... Uh, it's a reasonable point and, you know, the... You know, it's certainly, in terms of existential risks, it's different from, you know, asteroid collides with the Earth.

Lex Fridman 1:10:22 Right.

Stuart Russell 1:10:23 Right? Which, again, is quite possible. Uh, you know, it's happened in the past. Uh, it'll probably happen again. We don't right- we don't know right now, but if we did detect an asteroid that was going to hit the Earth in 75 years time, we'd certainly be doing something about it.

Lex Fridman 1:10:39 Well, it's clear there's a big rock and there's-

Stuart Russell 1:10:42 (laughs)

Lex Fridman 1:10:42 We'll probably have a meeting and see what do we do about the big rock. With AI...

Stuart Russell 1:10:46 Right. With AI, I mean, there are very few people who think it's not gonna happen within the next 75 years. I know Rod Brooks doesn't think it's gonna happen. Uh, maybe Andrew Ng doesn't think it's happening. But, you know, a l- a lot of the people who work day to day, uh, you know, as you say, at the rock face, they think it's gonna happen. I think the median estimate from AI researchers is somewhere in 40 to 50 years from, from now or maybe even a little... You know, I think in Asia, they think it's gonna be even faster than that. I am, I am a little bit more conservative. I think it probably take longer than that. But I think it's, you know, as happened with nuclear weapons, um-

Lex Fridman 1:11:29 It could happen overnight.

Stuart Russell 1:11:30 ... it can happen overnight, that you have these breakthroughs. And we need more than one breakthrough but, you know, the... It's on the order of half a dozen. I mean, this is a very rough scale, but sort of half a dozen breakthroughs of that nature would have to happen for us to reach the- superhuman AI. But the, you know, the AI research community is vast now. The massive investments from governments, from corporations, uh, tons of really, really smart people. You know, you just have to look at the rate of progress in different areas of AI to see that things are moving pretty fast. So, so to say, "Oh, it's just gonna be thousands of years," I mean, I- I don't see any basis for that. You know, I see, you know, for example, the, the, the Stanford hundred year AI project, right? Which, um, is supposed to be sort of, you know, the serious establishment view. Uh, their most recent report actually said it's probably not even possible.

Lex Fridman 1:12:34 Oh, wow.

Stuart Russell 1:12:35 Right? Which if, if you want a perfect example of people in denial, that's it. Because, you know, for the whole history of AI, we've been saying to philosophers who said it wasn't possible, "Well, you have no idea what you're talking about. Of course it's possible," right? "Give me an, give me an argument for why it couldn't happen." And there isn't one, right? And now because people are worried that maybe AI might get a bad name or, or I just don't want to think about this, they're saying, "Okay, well, of course it's not really possible," you know? And we imagine, right? Imagine if, you know, the, y- the leaders of the cancer biology community...... uh, got up and said, "Well, you know, of course, curing cancer, it's not really possible." (laughs)

Lex Fridman 1:13:18 (laughs)

Stuart Russell 1:13:19 (laughs) It would be a complete outrage-

Lex Fridman 1:13:22 (laughs)

Stuart Russell 1:13:22 ... a- and it would dismay, and, you know, uh, uh, I, I, I find this really a strange phenomenon. So...

Lex Fridman 1:13:31 That ƒ (inaudible) .

Stuart Russell 1:13:33 Okay, so if you accept that it's possible, and if you accept that it's probably going to happen, I, I, the point that you're making, that, you know, how does it go wrong? A valid question. Without that, without an answer to that question, then you're stuck with what I call the gorilla problem, which is, you know, the problem that the gorillas face.

Lex Fridman 1:13:55 Yeah.

Stuart Russell 1:13:55 Right? They made something more intelligent than them, namely us, uh, a few million years ago, and now, (laughs) now, they're in deep doodoo.

Lex Fridman 1:14:03 Yeah.

Stuart Russell 1:14:04 Uh, so the- there's really nothing they can do. They've lost the control. They, they failed to solve the control problem of controlling humans, and, uh, so they, they lost. Um, so we don't want to be in that situation. And if the gorilla problem is, is the only formulation you have, there's not a lot you can do.

Lex Fridman 1:14:23 Mm-hmm.

Stuart Russell 1:14:23 Right? Other than to say, "Okay, we should try to stop. W- you know, we should just not make the humans or, or, or in this case-

Lex Fridman 1:14:30 Right, which-

Stuart Russell 1:14:30 ...not make the AI." And I think that's really hard to do, to, uh... And I'm not actually proposing that that's a feasible course of action. Um, and I w- I also think that, you know, if properly controlled, AI could be incredibly beneficial. So the... But it seems to me that there's a, there's a consensus that one of the major failure modes is this loss of control, that we create AI systems that are pursuing incorrect objectives, and because the AI system believes it knows what the objective is, it has no incentive to listen to us anymore, so to speak, right? It, it's just carrying out the, the strategy that it, it has computed as being the optimal solution. And, uh, you know, it may be that, in the process, it needs to acquire more resources to increase the possibility of success or, you know, prevent various failure modes by defending itself against interference. And so that collection of problems, I think, is something we can address.

Lex Fridman 1:15:43 Yes.

Stuart Russell 1:15:45 Uh, the other problems, uh, are, roughly speaking, you know, misuse, right? So even if we solve the control problem, we make perfectly safe, controllable AI systems, well, why? You know, why does Dr. Evil going to use those, right? He wants to just take over the world, and he'll make unsafe AI systems that, that then get out of control. So that's one problem which is sort of a, you know, uh, partly a policing problem, partly a, a sort of a cultural problem for the profession of how we teach people, uh, what kinds of AI systems are safe.

Lex Fridman 1:16:23 You talk about autonomous weapons system and how pretty much everybody agrees that there's too many ways that that can go horribly wrong, and this great, uh, Slaughterbots movie that kinda illustrates that beautifully.

Stuart Russell 1:16:35 (laughs)

Lex Fridman 1:16:35 And it-

Stuart Russell 1:16:35 Well, I want to talk a- that, that's another, there's another topic I, I'm happy to talk about. The... Just wanna mention the th- what I see as the third major failure mode, which is overuse. Not so much misuse, but overuse of AI, that we become overly dependent. So I call this the WALL-E problem. So if you've seen WALL-E-

Lex Fridman 1:16:55 Yeah.

Stuart Russell 1:16:55 ... the movie, all right, all the humans are on the spaceship, and the machines look after everything for them, and they just watch TV and drink Big Gulps. And, uh, they're all sort of obese and stupid and, and they've sort of totally lost any notion of human autonomy. And, um, you know, so i- i- in effect, right, this would happen like the slow-boiling frog, right? We would gradually turn over more and more of the management of our civilization to machines, as we are already doing.

Lex Fridman 1:17:28 Mm-hmm.

Stuart Russell 1:17:29 And this, you know, this, if this process continues, you know, we, we sort of gradually switch from sort of being the masters of technology to just being the guests. Right? So, so we become guests on a cruise ship, you know, which is fine for a week, but not f- not for the rest of eternity.

Lex Fridman 1:17:48 Right.

Stuart Russell 1:17:49 You know, and it's almost irreversible, right? Once you, once you lose the incentive to, for example, you know, learn to be an engineer or a doctor or a sanitation, uh, operative or, or any other of the, uh, uh, the infinitely many ways that we maintain and propagate our civilization. You know, if you, if you don't have the incentive to do any of that, you won't, and then it's really hard to recover.

Lex Fridman 1:18:19 And of course, AI is just one of the technologies that could... That third failure mode result in that. There's probably other... Technology in general detaches us from, um...

Stuart Russell 1:18:28 It does a bit, but the, the, the-

Lex Fridman 1:18:30 ...

Stuart Russell 1:18:30 (...) independence. ...difference is that in terms of the knowledge to, to run our civilization, you know-

Lex Fridman 1:18:36 Ah.

Stuart Russell 1:18:36 ...up to now, we've had no alternative but to put it into people's heads.

Lex Fridman 1:18:40 Right.

Stuart Russell 1:18:41 Right? And if you, if you-

Lex Fridman 1:18:41 But with software, with Google, I mean, so software in general, so AI broadly defined

Stuart Russell 1:18:45 So, so compute- computers in general-

Lex Fridman 1:18:47 Yeah.

Stuart Russell 1:18:47 ...but, but the, you know, the knowledge o- of how, you know, how a sanitation system works, you know, that's an... The AI has to understand that. It's no good putting it into Google. So, I mean, we've, we've always put knowledge in- on paper, but paper doesn't run our civilization. It only runs when it goes from the paper into people's heads again. Right? So we've always propagated civilization-... through human minds.

Lex Fridman 1:19:11 Mm-hmm.

Stuart Russell 1:19:12 And we've spent about a trillion person years doing that.

Lex Fridman 1:19:15 (laughs)

Stuart Russell 1:19:16 I, I... Literally, right?

Lex Fridman 1:19:17 Trillion.

Stuart Russell 1:19:17 You c- you can work it out.

Lex Fridman 1:19:18 Yeah, yeah.

Stuart Russell 1:19:18 It's about, right? It's about just over 100 billion people who've ever lived and, uh, each of them has spent about 10 years learning stuff-

Lex Fridman 1:19:26 Yeah.

Stuart Russell 1:19:26 ... to keep their civilization going. And, uh, so that's a trillion person years we put into this effort.

Lex Fridman 1:19:31 A beautiful way to describe-

Stuart Russell 1:19:33 And, uh-

Lex Fridman 1:19:33 ... all of civilization. (laughs)

Stuart Russell 1:19:34 And now, we're, you know, we're in danger of, of throwing that away. So this is a problem that AI can't solve. It's not a technical problem.

Lex Fridman 1:19:41 Yeah.

Stuart Russell 1:19:41 It's a... You know, and i- if we do our job right, the AI systems will say, you know, "The human race doesn't, in the long run, want to be passengers in a cruise ship. The human race wants autonomy. This is part of human preferences. So we, the AI systems, are not going to do this stuff for you."

Lex Fridman 1:20:00 Mm-hmm.

Stuart Russell 1:20:01 "You've got to do it for yourself." Right? "I'm not gonna carry you to the top of Everest in an autonomous helicopter. You have to climb it if you want to get the benefit," um, and so on, so... But I'm afraid that because we are shortsighted and lazy, w- we're gonna override the AI systems and... And there's an amazing short story that I recommend to everyone (laughs) that I talk to about this called, um, The Machine Stops, written in 1909 by E.M. Forster, who, you know, wrote novels about the British Empire and sort of things that became costume dramas on the BBC, but he wrote this one science fiction story-

Lex Fridman 1:20:42 Mm-hmm.

Stuart Russell 1:20:43 ... which is an amazing vision of the future. It has, it has basically iPads, uh, it has video conferencing, it has MOOCs, uh, it has com- computer-

Lex Fridman 1:20:55 (laughs) MOOCs, nice.

Stuart Russell 1:20:55 ... com- computer-induced obesity. I mean, literally, the, the, what-

Lex Fridman 1:20:58 The whole thing.

Stuart Russell 1:20:59 It's what people spend their time doing, i- is giving online courses or listening to online courses-

Lex Fridman 1:21:04 Yeah.

Stuart Russell 1:21:04 ... and, and, and talking about ideas, but they never get out there in the real world. They're, uh, they don't really have a lot of face-to-face contact. Uh, everything is done online, you know. So all the things we're worrying about now, uh, were described in this story and, and then the human race becomes more and more dependent on the machine, loses knowledge of how things really run-

Lex Fridman 1:21:27 Yeah.

Stuart Russell 1:21:27 ... uh, and then becomes vulnerable to collapse. And, uh, so it's a, it's a pretty unbelievably amazing story for someone writing in 1909 to, to imagine-

Lex Fridman 1:21:38 A hundred years ago plus.

Stuart Russell 1:21:39 ... all this. Yeah.

Lex Fridman 1:21:40 So there's very few people that represent artificial intelligence more than you, Stuart Russell. (laughs)

Stuart Russell 1:21:47 If you say so, okay. That's very kind. (laughs)

Lex Fridman 1:21:49 So-

Stuart Russell 1:21:50 So it's all my fault. Is that what you're saying?

Lex Fridman 1:21:51 It's all your fault. (laughs)

Stuart Russell 1:21:51 Okay, good.

Lex Fridman 1:21:53 No, right. Um, i- y- you're often brought up as the person, "Well, Stuart Russell, like, the AI person is worried about this, that's why you should be worried about it." Do you feel the burden of that? I don't know if you feel that at all. But when I talk to people, like from... Uh, y- you talk about s- people outside of computer science-

Stuart Russell 1:22:15 Mm-hmm.

Lex Fridman 1:22:15 ... when they think about this, "Stuart Russell, uh, is worried about AI safety, you should be worried too," do you feel the burden of that?

Stuart Russell 1:22:23 I mean, in a practical sense, uh, yeah, because, uh, I get, uh, you know, a dozen sometimes 25 invitations a day to talk about it, to give interviews, to write press articles, and so on. So, um, in that very practical sense, I'm seeing that people are concerned and really interested about this, um.

Lex Fridman 1:22:48 Are you worried that you could be wrong, as all good scientists are?

Stuart Russell 1:22:52 Of course. I worry about that all the time. I mean, that's, that's always been the way that I've, I've worked, you know, is like I, I have an argument in my head with myself, right?

Lex Fridman 1:23:03 (laughs)

Stuart Russell 1:23:03 So I have, I have some idea and then I think, "Okay, how could that be wrong? Or did someone else already have that idea?" So I'll go and, you know, search in as much literature as I can to f- to see whether someone else already thought of that or, or even refuted it. So, you know, I... Right now, I'm, I'm reading a lot of philosophy because, you know, in, in the form of the debates over, over utilitarianism and, and-

Lex Fridman 1:23:33 Mm-hmm.

Stuart Russell 1:23:33 ... other kinds of moral, uh, moral formulas, shall, shall we say, people have already thought through some of these issues. But, you know, one, one of the things I'm, I'm not seeing in a, in a lot of these debates is, is this specific idea about, uh, the importance of uncertainty in the objective.

Lex Fridman 1:23:55 Mm-hmm.

Stuart Russell 1:23:55 Um, that this is the way we should think about machines that are beneficial to humans, so this idea of, of provably beneficial machines based on, uh, explicit uncertainty in the objective. Um, you know, it seems to be... Y- you know, my, my gut feeling is this is the core of it. It's gonna have to be elaborated in a lot of different directions and there are a lot of-

Lex Fridman 1:24:22 Provably beneficial.

Stuart Russell 1:24:23 Yeah. But there, there... I, I mean, it has to be, right?

Lex Fridman 1:24:26 Right.

Stuart Russell 1:24:26 We, we can't afford, you know, hand-wavy beneficial-

Lex Fridman 1:24:29 Yeah.

Stuart Russell 1:24:30 ... uh, because there are-

Lex Fridman 1:24:31 Booms.

Stuart Russell 1:24:31 You know, whenever we do hand-wavy stuff, there are loopholes. And the thing about super intelligent machines is they find the loopholes, you know, just like, you know, tax evaders. Uh, if you don't write your tax law properly, (laughs) the people will find the loopholes and, and end up paying no tax, and, and, uh, so you should think of it this way and, and getting those definitions right, you know, it is really a long process, you know. So you can, you can define mathematical frameworks and within that framework you can prove mathematical theorems that, yes, this will... You know, this, this theoretical entity will be provably beneficial to that theoretical entity-

Lex Fridman 1:25:13 Mm-hmm.

Stuart Russell 1:25:13 ... but that framework may not match the real world in some crucial way. Um-

Lex Fridman 1:25:19 So it's a long process, thinking through it to iterating and so on.

Stuart Russell 1:25:22 Yeah.

Lex Fridman 1:25:22 Last question.

Stuart Russell 1:25:23 Yep.

Lex Fridman 1:25:24 Uh, you have 10 seconds to answer it. What is your favorite sci-fi movie about AI?

Stuart Russell 1:25:31 I would say Interstellar has my favorite robots-

Lex Fridman 1:25:34 Oh, beats Space o- uh, Space Odyssey?

Stuart Russell 1:25:37 TARS. Yeah, yeah, yeah. So, so TARS, the robots, uh, one of the robots in Interstellar is the way robots should behave. And, uh, I would say Ex Machina is in some ways the one-

Lex Fridman 1:25:51 Second place.

Stuart Russell 1:25:51 ... the one that makes you think, uh, in a nervous kinda way about, about where we're going.

Lex Fridman 1:25:58 Well, Stuart, thank you so much for talking today.

Stuart Russell 1:26:00 Pleasure.