Yann LeCun: Deep Learning, ConvNets, and Self-Supervised Learning | Lex Fridman Podcast #36

Lex Fridman 0:00 The following is a conversation with Yann LeCun. He's considered to be one of the fathers of deep learning, which, if you've been hiding under a rock, is the recent revolution in AI that's captivated the world with the possibility of what machines can learn from data. He's a professor at New York University, a vice president and chief AI scientist at Facebook, and co-recipient of the Turing Award for his work on deep learning. He's probably best known as the founding father of convolutional neural networks, in particular, their application to optical character recognition and the famed MNIST dataset. He is also an outspoken personality, unafraid to speak his mind in a distinctive French accent and explore provocative ideas, both in the rigorous medium of academic research and the somewhat less rigorous medium of Twitter and Facebook. This is the Artificial Intelligence podcast. If you enjoy it, subscribe on YouTube, give it five stars on iTunes, support it on Patreon, or simply connect with me on Twitter, @lexfridman, spelled F-R-I-D-M-A-N. And now, here's my conversation with Yann LeCun.

Lex Fridman 1:11 You said that 2001: Space Odyssey is one of your favorite movies. HAL 9000 decides to get rid of the astronauts, for people who haven't seen the movie, spoiler alert, because he, it, she believes that the astronauts, they will interfere with the mission. Do you see HAL as flawed in some fundamental way, or even evil, or did he do the right thing?

Yann LeCun 1:38 Neither. There's no notion of evil in that- in that context, other than the fact that people die. But it was an example of what people call, uh, value misalignment, right? You give an objective to a machine and the machine str- strives to achieve this objective, and if you don't put any constraints on this objective, like, don't kill people and don't do things like this, the machine, given the power, will do stupid things just to achieve this- this objective, or damaging things to achieve this objective. It's a little bit like... I mean, we are used to this in the context of human society. We- we put in place laws to prevent people from doing bad things because

Lex Fridman 2:22 (laughs)

Yann LeCun 2:22 ... spontaneously they would do those bad things, right? So, we have to shape their cost function, their objective function, if you want, through laws to kind of correct, and education, obviously, to sort of correct for- for those.

Lex Fridman 2:36 So, maybe just pushing a little further on- on that point, HAL, you know, there's a mission. There's a- there's fuzziness around the- the ambiguity around what the actual mission is, but, you know, d- do you think that there will be a time, from a utilitarian perspective, where an AI system, where it is not misalignment, where it is alignment for the greater good of society, that an AI system will make decisions that are difficult?

Yann LeCun 3:05 Well, that's the trick. I mean, uh, eventually, we'll have to figure out how to do this. And again, th- we're not starting from scratch because we've been doing this with humans for- for millennia.

Lex Fridman 3:15 Oh, yeah, yeah.

Yann LeCun 3:16 So, designing objective functions for people is something that we know how to do, and we don't do it by, um, you know, programming things, although th- the legal code is called code, um, so (laughs) that tells you something, and it's actually the design of an objective function. That's really what legal code is, right? It tells you, "Here is what you can do. Here is what you can't do. If you do it, you pay that much." That's- that's an objective function. So, there is this idea somehow that it's a new thing for people to try to design objective functions that are aligned with the common good, but no. We've been writing laws for millennia, and that's exactly what it is. Um, so this- that's where, you know, uh, the science of, uh, lawmaking and, uh- and computer science will-

Lex Fridman 4:00 Come together.

Yann LeCun 4:01 ... will come together.

Lex Fridman 4:02 So, it's nothing- there's nothing special about HAL or AI systems. It's just a continuation of tools used to make some of these difficult ethical judgments that laws make, essentially.

Yann LeCun 4:13 Yeah, and we- and we have systems like this already that, you know, uh, make many decisions for- for ourselves in society that, you know, need to be designed in a way that they, like, you know, rules about things that someti- sometimes have bad side effects, and we have to be flexible enough about those rules so that they can be broken when it's obvious that they shouldn't be applied. So, you don't see this on the camera here, but all the decoration in this room is all pictures from 2001: A Space Odyssey. (laughs)

Lex Fridman 4:41 Wow.

Yann LeCun 4:42 Um...

Lex Fridman 4:42 Is that by accident, or is there a lot-

Yann LeCun 4:43 It's not by accident. It's by design.

Lex Fridman 4:45 (laughs) Oh, wow. So, if you were, um... If you were to build HAL 10,000, so an improvement of HAL 9000, what would you improve?

Yann LeCun 4:57 Well, first of all, I wouldn't ask it to hold secrets and tell lies, because that's really what breaks it in the end. That's the- the fact that it's asking itself questions about the purpose of the mission, and it, you know, pieces things together, that it's heard, you know, all the secrecy of the preparation of the mission and the fact that there was a discovery on- on the lunar surface that really was kept secret. And- and one part of HAL's memory knows this, and the other part is- does not know it and is supposed to not tell anyone, and that creates a internal conflict.

Lex Fridman 5:28 So, you think there never should be a set of things that an AI system should not be allowed, like a- a set of facts that should not be shared with the human operators?

Yann LeCun 5:42 Well, I think no. I think the- I think it should be a bit like, in the design of, um, autonomous AI systems, um, there should be the equivalent of, you know, the- the- the oath that, uh, Hippocratic Oath calls for-

Lex Fridman 5:58 Hippocratic Oath, yeah.

Yann LeCun 5:59 ... that, uh, doctors, uh, sign up to, right? So, there's certain things, certain rules that- that- that you have to abide by. And we can sort of hardwire this into- into our- into our machines to kind of make sure they don't go... So, I'm not, you know, an advocate of the- the three do- three laws of robotics, you know, the Asimov kind of thing, because I don't think it's practical, but...... but, you know, some, some level of, uh, of limits. But, but to be clear, uh, this is not, these are not questions that are kind of really worth asking today because we just don't have the technology to do this. We don't, we don't have autonomous intelligent machines. We have intelligent machines, semi-intelligent machines that are very specialized, but they don't, they don't really sort of satisfy an objective. They're just, you know, kind of trained to do one thing.

Lex Fridman 6:46 Yeah.

Yann LeCun 6:46 So, until we have some idea for a design of a full-fledged autonomous intelligent system, asking the question of how, how we design this objective I think is a little, a little too abstract.

Lex Fridman 6:58 It's a little too abstract. There's useful elements to it, in that it helps us understand our own ethical codes as humans. So, even just as a thought experiment, if you imagine that an AGI system is here today, how, how would we program it is a kind of nice thought experiment of constructing how should we have a law, have a system of laws for us humans. It's, it's just a nice practical tool. And I think there's echoes of that idea, too, in the A- AI systems we have today that don't have to be that intelligent.

Yann LeCun 7:34 Yeah.

Lex Fridman 7:34 Uh, like autonomous vehicles.

Yann LeCun 7:35 Right.

Lex Fridman 7:35 Is these things start creeping in that, uh, we're thinking about, but certainly they shouldn't be framed as, as HAL.

Yann LeCun 7:42 Yep.

Lex Fridman 7:43 Looking back, what is the most... I'm sorry if it's a silly question,

Lex Fridman 7:49 but what is the most beautiful or surprising idea in deep learning or AI in general that you have ever come across? Sort of personally where you sat back and, and just had this kind of, "Oh, that's pretty cool" moment, "That's nice-"

Yann LeCun 8:04 Well-

Lex Fridman 8:04 ... and surprising."

Yann LeCun 8:05 ... I don't know if it's an idea rather than a sort of empirical, uh, fact. The fact that you can build gigantic neural nets, train th- train them on, you know, relatively small amounts of data, relatively, with stochastic gradient descent, and that it actually works, uh, breaks everything you read in every textbook, right? Every pre-deep learning textbook that told you you need to have fewer parameters than you have data samples, um, you know, if you have a non-convex objective function, you have no guarantee of convergence. You know, all those things that you read in textbook, and they tell you, "Stay away from this," and they're all wrong. And, uh...

Lex Fridman 8:45 A huge number of parameters, non-convex, and somehow with just very relative to the number of, uh, parameters data, it's able to learn anything.

Yann LeCun 8:54 Right.

Lex Fridman 8:55 Does that still surprise you today?

Yann LeCun 8:57 Well, it, it was kind of obvious to me before I knew anything that th- that's, that this is a good idea, and then it became surprising that it worked because I started reading those textbooks.

Lex Fridman 9:08 (laughs)

Yann LeCun 9:09 Okay? So, like, you know-

Lex Fridman 9:10 Wait, so, okay, so can you talk through the intuition of why it was obvious to you, if you remember?

Yann LeCun 9:14 Well, okay, so the intuition was it's- it's sort of like, you know, those people in the late 19th century who proved that heavier than, than air f- uh, flight was impossible, right? And of course you have birds, right? They do fly. And so on the face of it, it- it's obviously wrong as an empirical question, right? And so we have the same kind of thing that, you know, the, we know that the brain works. We don't know how, but we know it works, and we know it's a large network of neurons and interaction, and that learning takes place by changing the connections. So, kind of getting this level of inspiration without copying the details, but sort of trying to derive basic principles, you know, that kind of gives you a, a clue as to which direction to go. There's also the idea somehow that I've, I've been convinced of since I was an undergrad, that... even before, that intelligence is inseparable from learning. So you... the idea somehow that you can create an intelligent machine by basically programming, for me was a non-starter for, you know, from the start. Ev- every intelligent entity that we know about arrives at this intelligence through learning. Um, so learning, you know, machine learning was, uh, a completely obvious path. Uh, also because I'm lazy, so, you know, kind of... (laughs)

Lex Fridman 10:33 (laughs) You just automate basically everything in learning is the automation of-

Yann LeCun 10:37 Exactly.

Lex Fridman 10:37 ... intelligence.

Yann LeCun 10:37 Right.

Lex Fridman 10:39 S- so

Lex Fridman 10:40 do you think... So, what is learning then? Wha- what falls under learning? Because do you think of reasoning as learning?

Yann LeCun 10:49 Well, reasoning is certainly a, a, a consequence of, of learning as well, just like other functions of, uh, of the brain. The big question about reasoning is how do you make reasoning compatible with gradient-based learning?

Lex Fridman 11:03 Do you think neural networks can be made to reason?

Yann LeCun 11:05 Yes, that there is no question about that.

Lex Fridman 11:07 (laughs)

Yann LeCun 11:07 Again, we have a good example, right? The question is, is how. So the question is, how much prior structure do you have to put into the neural net so that something like human reasoning will emerge from it, you know, from learning? Uh, another question is all of our kind of model of what reasoning is that are based on logic are discrete and, and, and are therefore incompatible with gradient-based learning.

Lex Fridman 11:32 Mm-hmm.

Yann LeCun 11:33 And I'm a s- very strong believer in this idea of gradient-based learning. I don't believe that, uh, other types of learning that don't use kind of gradient information, if you want.

Lex Fridman 11:42 So you don't like discrete mathematics! You don't like anything discrete?

Yann LeCun 11:45 Well, that's... it's not that I don't like it, it's just that it's, it's incompatible with learning and I'm, I'm a big fan of learning, right? So in fact that's perhaps one reason why, uh, deep learning has been kind of looked at with suspicion by a lot of computer scientists, because the math is very different. The math that, uh, you use for deep learning i- you know, kind of has more to do with, you know, cybernetics, uh, the kind of st- math you do in electrical engineering than the, the kind of math you do in computer science. And, and you know, nothing in, in machine learning is exact, right? Computer science is all about sort of, you know, obsessive-compulsive attention to details of like, you know, every index has to be right and-

Lex Fridman 12:24 Right.

Yann LeCun 12:24 ... you can prove that an algorithm is correct, right? Uh, machine learning is th- the science of sloppiness, really. (laughs)

Lex Fridman 12:32 (laughs) That's beautiful. So-Okay. Maybe let's feel around in the dark of what is a neural network that reasons, or a system that is, works with continuous functions, that's able to do, uh, build knowledge however we think about reasoning. Build on previous knowledge, build on extra knowledge, create new knowledge, generalize outside of any training set ever built? What does that look like if, uh... Yeah, ma- maybe do you have inklings of thoughts of what that might look like?

Yann LeCun 13:10 Uh, yeah. I mean, yes and no. If I had precise ideas about this, I think, you know, (laughs) we'd be building it right now. But... And there are people working on this, or whose main research interest is actually exactly that, right? So, what you need to have is a working memory. So, you need to have some device, if you want, uh, some subsystem that can store, uh, a relatively large number of factual episodic information for, you know, a reasonable amount of, amount of time. So, y- you, you know, in the, in the brain, for example, there are kind of three main types of memory. One is the, the sort of memory of the, the, the state of your cortex and that sort of disappears within 20 seconds. You can't remember things for more than about 20 seconds or a minute if, if, if you don't have any other form of memory. Uh, the second type of memory, which is longer term but still short term, is the hippocampus. So you can... You know, you came into this building, you remember where the, where the, the exit is, where the, uh, elevators are. Um, you have some map of that building, that's stored in your hippocampus. You might remember something about what I said, you know, a few minutes ago-

Lex Fridman 14:21 I forgot it all already, but part-

Yann LeCun 14:22 ... of course, it's been erased, but-

Lex Fridman 14:23 Yeah.

Yann LeCun 14:23 ... you know, but that, that would be in your-

Lex Fridman 14:25 Some parts, yeah.

Yann LeCun 14:25 ... in your hippocampus. Uh, and then the, the longer-term memory is in the synapse, the synapses, right? Um, so what you need if you want a system that's capable of reasoning is that you want a hippocampus-like thing, right? And that's what people have tried to do with memory networks and, you know, neural training machines and stuff like that, right? A- and now with transformers, which have sort of a memory in their kind of self-attention system. You can, you can think of it this way. So, so that's one element you need. Another thing you need is some sort of network that can access this memory, get an information back, and then kinda crunch on it, and then do this iteratively multiple times. Because a chain of reasoning is, is a process by which you, you, you can, uh, you update your knowledge about the state of the world, about, you know, what's gonna happen, et cetera. And that, that has to be this sort of recurrent operation basically.

Lex Fridman 15:27 And you think that kind of... I- if we think about a transformer, so that seems to be too small to contain the knowledge that's, that's, uh... To represent the knowledge that's contained in Wikipedia, for example.

Yann LeCun 15:39 Well, a transformer doesn't have this idea of, uh, recurrence. It's got a fixed number of layers and that's the number of steps that, you know, limits basically its representation.

Lex Fridman 15:47 So, but recurrence would build on the knowledge somehow.

Yann LeCun 15:51 Yeah.

Lex Fridman 15:51 I mean, it would, it would evolve the knowledge and ex- expand the amount of information perhaps, or useful information within that knowledge.

Yann LeCun 16:00 Yeah.

Lex Fridman 16:00 But is, is this something that just can emerge with size? Because it seems like everything we have now is too small.

Yann LeCun 16:06 Not just. No, it's not cle- it's not, it's not clear. I mean, how, how you access and write into an associative memory in an efficient way... I mean, sort of the original memory network maybe had something like the right architecture. But, uh, if you try to scale up a memory network so that the memory contains all of Wikipedia, I mean, it doesn't quite work.

Lex Fridman 16:24 Right.

Yann LeCun 16:25 So, so there, there's, there's a need for new ideas there. Okay. But it's not the only form of reasoning. So, there's another form of reasoning which is through... Which is very classical also in s- some types of AI, uh, and it's based on, uh... Let's call it energy minimization. Okay. So, you have, uh, some sort of objective, s- some energy function that represents the, the, the, um, the quality or the negative quality. Okay. Energy goes up when things get bad and they get low when things g- g- get good. So, let's say you, you want to figure out, you know, what gestures do I need to, to do to grab an object or walk out the door. If you have a good model of, uh, your own body, a good model of the environment, using this kind of energy minimization, you can make a, you can make. You can do planning. And it's, uh, a cl- in optimal control, it's called, it's called mode- model predictive control. You have a model of what's gonna happen in the world as a consequence of your actions, and that allows you to, by energy minimization, figure out the sequence of action that optimizes a particular objective function which measures, you know, minimizes the number of times you're gonna hit something and the energy you're gonna spend doing the gesture and et cetera. So, so that's one form of reasoning. Planning is a form of reasoning. And perhaps what led to the ability of humans to reason is the fact that... Or, you know, species, you know, that appeared before us had to do some sort of planning to be able to hunt and survive, and survive the winter in particular. And so, you know, it's the same capacity that, that you need to have.

Lex Fridman 18:03 So,

Lex Fridman 18:04 in your intuition, is, um... If we look at expert systems and, and encoding knowledge as logic systems, and as graphs, in this kind of way, is not a useful way to think about knowledge?

Yann LeCun 18:20 Graphs are a little brittle. Or, or logic, uh, representation, so basically, you know, variables that, that have values and then constrained between them that are represented by rules, uh, is a little too rigid and too brittle, right?

Lex Fridman 18:32 Mm-hmm. Yes.

Yann LeCun 18:32 So, one of the... You know, some of the early efforts in that respect, um...... were, were to put probabilities on them. So a rule, you know, you know, if you have this and that symptom, you know, you have this, uh, disease with that probability, and you should prescribe that antibiotic with that probability, right? That's the, the Mycin system from the, from the '70s. Um, and that, that's what that branch of AI led to, uh, you know, based on networks and graphical models and causal inference and variational pro- you know-

Lex Fridman 19:03 Yep.

Yann LeCun 19:04 ... method. So, so there, there is, I mean, certainly, uh, uh, a lot of interesting work going on in this area. The main issue with this is, is knowledge acquisition. How do you, uh, reduce a bunch of data to a graph of this type?

Lex Fridman 19:19 Yeah, it relies on the expert to ... on the human being to encode that or to add knowledge.

Yann LeCun 19:25 And that's essentially impractical. (laughs)

Lex Fridman 19:28 Yeah. So it's not scalable, right?

Yann LeCun 19:28 So that, that's, uh, that's, that's a big question. The second question is, do you want to represent knowledge as symbols, and do you want to manipulate them with logic? And again, that's incompatible with learning. So, uh, one suggestion which, you know, Geoff Hinton has been advocating for many decades is replace symbols by, uh, vectors. Think of it as pattern of activities in a bunch of neurons or units or whatever you want to call them, and replace logic by continuous functions.

Lex Fridman 19:58 Mm-hmm.

Yann LeCun 19:59 Okay? And that becomes now compatible. There's a very good set of ideas by, uh, written in a, a paper about 10 years ago by, uh, Leon Bottou, on, uh, who is here at, at Facebook. Um, um, the title of the paper is From Machine Learning to Machine Reasoning, and his idea is that, uh, a learning s- a learning system should be able to manipulate objects that are in the same sp- in a space and then put the result back in the same space. So, it's this idea of working memory, basically, and it's, uh, it's very enlightening.

Lex Fridman 20:30 And in a sense, that might learn something like the simple expert systems. I mean, it's w- you can learn basic logic operations there.

Yann LeCun 20:42 Yeah, quite possibly.

Lex Fridman 20:42 Yeah.

Yann LeCun 20:43 Yeah. There's a big debate on sort of how much prior structure you have to put in for this kind of stuff to emerge. That's the debate I have with Gary Marcus and people like that.

Lex Fridman 20:51 Yeah. Yeah, so ... And the other person,

Lex Fridman 20:55 so I, I just talked to Judea Pearl-

Yann LeCun 20:57 Mm-hmm.

Lex Fridman 20:57 ... from you mentioned causal inference world. So, his worry is that the current neural networks are not able to learn what causes what, causal inference between things.

Yann LeCun 21:12 So, I think, I think he's right and wrong about this. Uh, if he's talking about the sort of classic, uh, type of neural nets, people sort of didn't worry too much about this. But there's a lot of people now working on causal inference, and there's a paper that just came out last week by Leon Bottou, among others, David Lopez-Paz and a bunch of other people, exactly on that problem of, um, how do you kind of, you know, get a neural net to sort of pay attention to real causal relationships, which may also solve issues of bias in data and things like this, so ...

Lex Fridman 21:48 I'd, I'd like to read that paper, because, uh, that ultimately the challenge there is also seems to fall back on the human expert to ultimately decide causality between things.

Yann LeCun 22:01 Uh, people are not very good at establishing causality, first of all. So, first of all, you talk to physicists, and physicists actually don't believe in causality-

Lex Fridman 22:08 Oh, I see.

Yann LeCun 22:08 ... because look at the, all the basic laws of microphysics are time reversible.

Lex Fridman 22:14 Yeah, there's no-

Yann LeCun 22:14 So, there is no causality.

Lex Fridman 22:15 The arrow of time is not real.

Yann LeCun 22:16 Right.

Lex Fridman 22:17 Yeah.

Yann LeCun 22:17 It's, it's as soon as you start looking at macroscopic systems where there is unpredictable randomness, where there is clearly an arrow of time, but it's a big mystery in physics actually, uh, how that emerges. Um, so-

Lex Fridman 22:29 Is it emergent, or is it, uh, part of the fundamental fabric of reality, yeah, of nature?

Yann LeCun 22:34 Or is it a bias of intelligent systems that-

Lex Fridman 22:37 Yeah.

Yann LeCun 22:37 ... you know, because of the second law of thermodynamics, we perceive a particular arrow of time, but in fact, um, it's kind of arbitrary, right?

Lex Fridman 22:44 Right. So yeah, physicists, mathematicians, they don't care about ... I mean, the math doesn't care about the flow of time.

Yann LeCun 22:51 Well, certainly, certainly microphysics doesn't. People themselves are not very good at establishing causal, um, causal relationships. If you ask this, um, I think it was in one of, uh, Seymour Papert's book on, on, um, like children learning, you know, he studied with Jean Piaget, you know. He's the guy who co-authored the book Perceptron with Marvin Minsky that kind of killed the first wave of neural nets.

Lex Fridman 23:14 (laughs)

Yann LeCun 23:14 But, but he was actually a, a, a learning person. He, uh, in the sense of studying learning in humansand, and machines. That's why he got interested in perceptron. And, uh, and he, he, he, he wrote that if you ask a little kid, uh, about what is the cause of, of the wind, a lot of kids will say ... they will think for a while and they'll say, "Oh, it's the, the branches in the trees. They move and that creates wind," right?

Lex Fridman 23:39 Mm-hmm.

Yann LeCun 23:40 So, they get the causal relationship backwards, and it's because their understanding of the world and intuitive physics is not that great, right? I mean, these are like, you know, four or five-year-old kids. You know, it gets better, and then you understand that this ... it can't be, right?

Lex Fridman 23:54 But there are many things which we can ... because of our common sense understanding of things, what people call common sense-

Yann LeCun 24:02 Yeah.

Lex Fridman 24:03 ... and with our understanding of physics, we can ... There's a lot of stuff that we can figure out causality. Even with diseases, we can figure out what's not causing what, uh, often.

Yann LeCun 24:14 Yeah.

Lex Fridman 24:14 There's a lot of mystery, of course, but the idea is that you should be able to encode that into systems, 'cause it seems unlikely they'd be able to figure that out themselves.

Yann LeCun 24:22 Well, whenever we can do intervention, but, you know, all of humanity has been completely deluded for millennia, probably since exi- existence, about a very, very wrong causal relationship where whatever you can explain, you attribute it to, you know, some deity, some divinity, right?

Lex Fridman 24:38 Yeah. Yeah.

Yann LeCun 24:38 Um, and that's a cop-out. That's a way of saying like, "I don't know the cause, so, you know, God did it."

Yann LeCun 24:43 Right?

Lex Fridman 24:43 Yeah.So, you mentioned Marvin Minsky and, uh, the irony of, um, uh, you know, maybe causing the first AI winter. You were there in the '90s, you were there in the '80s, of course. Uh, in the '90s, why do you think people lost faith in deep learning-

Yann LeCun 25:01 (laughs)

Lex Fridman 25:01 ... in the '90s and found it again a decade later, over a decade later?

Yann LeCun 25:06 Yeah, it wasn't called deep learning yet. It was just called neural nets.

Lex Fridman 25:09 Neural networks.

Yann LeCun 25:09 Um, yeah, they lost interest... I mean, I think I would put that around 1995, at least the machine learning community. There was always a neural net community, but it, but it became kinda disconnected from sort of mainstream machine learning, if you want. There were... it was basically electrical engineering that ke- kept at it.

Lex Fridman 25:32 Right.

Yann LeCun 25:33 Uh, and computer science.

Lex Fridman 25:34 Just gave up.

Yann LeCun 25:35 Gave up on, uh, on neural nets. I don't, I don't know. You know, I was too close to it to really sort of analyze it with sort of a, a unbiased eye, if you want. But I, I would, I would, I would, I would make a few guesses. So, the first one is, at the time, neural nets were v- it was very hard to make them work in the sense that you would, you know, implement back prop in your favorite language, um, and that favorite language was not Python, it was not MATLAB, it was not any of those things 'cause they didn't exist, right? You had to write it in Fortran, or C, or something like this, right? Um, so you would, uh, experiment with it. You would probably make some very basic mistakes, like, you know, barely initialize your weights, make the network too small because you read in a textbook, you know, you don't want too many parameters, right? And of course, you know, and you would train on XOR because you didn't have any other dataset to train on. And of course, you know, it works half the time, so you'd say, "Hm, I give up." Also, you were training with batch gradient, which, you know, isn't really sufficient. So, there was a lot of... there was a bag of tricks that you had to know to make those things work or you, you had to reinvent, and a lot of people just didn't, and they just couldn't make it work. Um, so that's one thing. The investment in software platform to be able to kind of, you know, display things, figure out why things don't work, kind of get a good intuition for how to get them to work, have enough flexibility so you can create, you know, uh, network architectures like convolutional nets and stuff like that. It was hard. I mean, you had to write everything from scratch. And again, you didn't have any Python or MATLAB or anything, right? So what-

Lex Fridman 27:14 I, I read that... Sorry to interrupt, but I read that you wrote in, in Lisp the, your first, uh, versions of, uh, LeNet with the convolutional networks, which-

Yann LeCun 27:23 Definitely.

Lex Fridman 27:23 ... by the way, one of my favorite languages.

Yann LeCun 27:25 It's... yeah.

Lex Fridman 27:25 Uh, that's how I knew you were legit.

Yann LeCun 27:27 (laughs)

Lex Fridman 27:27 Ah, Turing Award whatever. This, what, uh-

Yann LeCun 27:29 (laughs)

Lex Fridman 27:29 ... you programmed in Lisp. That's-

Yann LeCun 27:30 It's still my favorite language. But it's not that we programmed in Lisp, it's that we had to write our Lisp interpreter, okay?

Lex Fridman 27:38 Yeah.

Yann LeCun 27:38 'Cause it's not like-

Lex Fridman 27:38 That's right. (laughs)

Yann LeCun 27:39 ... we could use one that existed.

Lex Fridman 27:40 Yeah.

Yann LeCun 27:41 So, we wrote our Lisp interpreter that we hooked up to, uh, you know, a backend library that we wrote also for sort of neural net computation. And then, uh, after a few years, around 1991, we invented this idea of, uh, basically having modules that know how to forward propagate and back propagate gradients and then interconnecting those modules in a graph. 01:02 had made proposals on this, about this in the late '80s, and we were able to implement this using our Lisp system. Eventually, we wanted to use that system to make... build production code for character recognition at Bell Labs, so we actually wrote a compiler for that Lisp interpreter, so that... Patricia Simard, who is now at Microsoft, kinda did, did the bulk of it with Leon and, and me. And, uh, and so we could write our system in Lisp and then compile to C, and then we'll have a self-contained complete system that could kinda do the entire thing. Neither PyTorch nor TensorFlow can do this today, yet.

Lex Fridman 28:36 (laughs)

Yann LeCun 28:36 Okay. It's coming.

Lex Fridman 28:38 Yeah. (laughs)

Yann LeCun 28:40 Uh, (laughs) I mean, there's something like that in PyTorch called, you know, uh, TorchScript.

Lex Fridman 28:44 Mm-hmm.

Yann LeCun 28:44 And so, you know, we had to write our Lisp interpreter, we had to write our Lisp compiler, we had to invest a huge amount of effort to do this. And not everybody... if, if you don't completely believe in the concept, you're not going to invest the time to do this.

Lex Fridman 28:57 Right.

Yann LeCun 28:57 Now, at the time also, you know, or today, this would turn into Torch or PyTorch, or TensorFlow, or whatever. We'd put it in open source so everybody would use it and, you know, realize it's good. Back before 1995, working at AT&T, there's no way the lawyers would let you release anything in open source of this nature, and so we could not distribute our code really.

Lex Fridman 29:20 And, uh, on that point, and sorry to go on a million tangents, but on that point, I also read that there was some almost pat- like, a patent on convolutional networks-

Yann LeCun 29:30 Yeah, there was.

Lex Fridman 29:30 ... at Bell Labs. So that... (laughs)

Yann LeCun 29:33 There was.

Lex Fridman 29:33 First of all, I mean, just-

Yann LeCun 29:35 There's two, actually. (laughs)

Lex Fridman 29:37 (laughs) Uh, that ran out, uh, the- that-

Yann LeCun 29:40 Thankfully.

Lex Fridman 29:40 (laughs) Yeah.

Yann LeCun 29:42 In 2007.

Lex Fridman 29:42 In 2007, uh, I... what... look, can we, can we just talk about that for a s- I know you're at Facebook, but you're also at NYU, and, and what does it mean to patent ideas like these, software ideas essentially, or what... or mathematical ideas, or what are they?

Yann LeCun 30:03 Okay. So, they're not mathematical ideas. So, they are, you know, algorithms. And there was a period-

Lex Fridman 30:08 Got it.

Yann LeCun 30:08 ... where the US, uh, patent office would allow the patent of software as long as it was embodied. Uh, the Europeans are very different. They don't, they don't quite accept that. They have a different concept, but, you know. I don't, I no longer bel- I mean, I never actually strongly believed in this, but I, I don't believe in this kind of patent. Facebook basically doesn't believe in this kind of patent. Google files patents because they've been burned with Apple, and so now they do this for defensive purpose. But usually they say, "We're not gonna sue you if you infringe."

Lex Fridman 30:45 Right.

Yann LeCun 30:45 Facebook has a, has a similar policy. They say, you know, "We file patent on certain things for defensive purpose. We're not gonna sue you if you infringe unless you sue us." Um, so the...... the industry does not believe in, in patents. They are there because of, you know, the legal landscape and, and, and various things but, but I don't really believe in patents for this kind of stuff.

Lex Fridman 31:07 Okay, so, so that's... that's a great thing. So I- I-

Yann LeCun 31:10 I'll tell you a war story, actually.

Lex Fridman 31:11 Yeah.

Yann LeCun 31:11 So what happens was the- the first, uh, the first patent about convolutional net was about, kind of the early version of convolutional net that didn't have separate pooling layers. It had, uh, uh, convolutional layers which tried more than one, if you want, right?

Lex Fridman 31:24 Mm-hmm.

Yann LeCun 31:25 And then there was a second one on convolutional nets with separate pooling layers, uh, trained with back prop. And they were filed as- filed in '89 and 1990 or something like this. At the time, the life, life of a patent was 17 years. So here's what ha- what happened over the next few years, is that we started developing, um, character recognition technology around, uh, convolutional nets. And, uh, in 1994, a check reading system was deployed in ATM machines. In 1995, it was for large, uh, check reading machines in back offices, et cetera. And those systems were developed by an engineering group that we were collaborating with, uh, at AT&T, and they were commercialized by NCR, which at the time was a subsidiary of AT&T.

Lex Fridman 32:11 Mm-hmm.

Yann LeCun 32:11 Now, AT&T split up in 1996, eh, 199- early 1996, and the lawyers just looked at all the patents and they distributed the patents among the various companies. They gave the neuro- the convolutional net patent to NCR because they were actually selling products that used it. But nobody at NCR had any idea what a convolutional net was.

Lex Fridman 32:32 (laughs) Yeah.

Yann LeCun 32:33 Okay? So between 1996 and 2007... So there's a whole period until 2002 where I didn't actually work on machine learning or convolutional net. I resumed working on this around 2002. And between 2002 and 2007, I was working on them, crossing my finger that nobody at NCR would notice.

Lex Fridman 32:51 (laughs)

Yann LeCun 32:51 And nobody noticed.

Lex Fridman 32:52 Yeah, and I- and I hope that th- this kind of somewhat, as you said, lawyers aside, relative openness of the community now will continue.

Yann LeCun 33:03 It accelerates the entire progress of, of the industry. And, you know, the- the problems that Facebook and Google and others are facing today is not whether Facebook or Google or Microsoft or IBM or whoever is ahead of the other. It's that we don't have the technology to build those things we want to build. We want to build intelligent virtual assistants that have common sense. We don't have a monopoly on good ideas for this. We don't believe we do. Maybe others do- believe they do, but we don't, okay? If a startup tells you they have the secret to, you know, human level intelligence and common sense, don't believe them. They don't. And it's gonna take the entire, uh, work of the world research community for a while to get to the point where you can go off and- and each of those companies can start to build things on this. We're not there yet.

Lex Fridman 33:51 Absolutely, and this- this calls to the- the gap between the space of ideas and the rigorous testing of those ideas, of practical application that you often speak to.

Lex Fridman 34:03 You've, uh, written advice saying, "Don't get fooled by people who claim to have a solution to artificial general intelligence, who claim to have an AI system that work just like the human brain, or who claim to have figured out how the brain works. Ask them what the error rate they get on MNIST or ImageNet." So...

Yann LeCun 34:23 Yeah, this is a little dated, by the way. (laughs)

Lex Fridman 34:25 Two thous- I mean, five years?

Yann LeCun 34:27 Yes.

Lex Fridman 34:27 Who's counting? Okay. But I think your opinion is still MNIST and ImageNet, uh, yes, may be dated. There may be new benchmarks, right? But I think that philosophy is one you still and- and- and somewhat hold, that benchmarks and the practical testing, the practical application is where you really get to test the ideas.

Yann LeCun 34:48 Well, it may not be completely practical. Like for example, you know, it could be a toy dataset or it could... but it- but it has to be some sort of task that the community as a whole has accepted as some sort of standard, you know, kind of benchmark, if you want. It doesn't need to be real. So for example, uh, many years ago here at FAIR, um, people, you know, Jason West and Antoine Board and a few others proposed the- the Baby tasks, which were kind of a, a toy problem to test the ability of machines to reason actually, to access working memory and things like this. And, uh, it was very useful even though it wasn't a real task. MNIST is kind of halfway a real task. Uh, so, you know, toy problems can be very useful. It's just that... I was really struck by the fact that a- a lot of people, particularly a lot of people with money to invest, would be fooled by people telling them, "Oh, we have, you know, the algorithm of the cortex and you should give us 50 million."

Lex Fridman 35:39 Yes, absolutely. So there's a lot of people who... who try to take advantage of the hype for business reasons and so on. But let me sort of talk to this idea that new ideas, the- the ideas that push the field forward may not yet have a benchmark, or it may be very difficult to establish a benchmark.

Yann LeCun 36:00 I agree. That's part of the process. Establishing benchmarks is part of the process.

Lex Fridman 36:04 So what are your thoughts about... So we have these benchmarks on- around stuff we can do with images, from classification, to captioning-

Yann LeCun 36:14 Mm-hmm.

Lex Fridman 36:15 ... to just every kind of information you can pull off from images in the surface level. There's audio datasets, there's some video. What can we start... Natural language. What kind of stuff... What kind of benchmarks do you see that start creeping onto more something like intelligence, like reasoning, like... maybe you don't like the term, but AGI, echoes of that kind of, uh-

Yann LeCun 36:40 Yeah, so-

Lex Fridman 36:40 ... simulation?

Yann LeCun 36:41 ... a lot of people are working on interactive environments in which you can- you can train and test, uh, intelligent systems. So, so there are- for example, uh, you know, it's... th- the classical paradigm of, uh, supervised learning is that you- you have a dataset, you partition it into a training set, validation set, test set, and there's a clear protocol, right?... but what if the, that assumes that the samples are static- statistically independent. You can exchange them, the order in which you see them doesn't, shouldn't matter, you know, things like that. But what if the answer you give determines the next sample you see? Which is the case, for example, in robotics, right? Your robot does something and then it gets exposed to a new room and depending on where it goes, the room would be different. So, that's the, that creates the exploration problem.

Lex Fridman 37:28 Mm-hmm.

Yann LeCun 37:30 The, what if the samples, so that creates also a dependency between samples, right? You, you, if you move, if you can only move in, in space, the next sample you're gonna see is gonna be probably in the same building most likely, right? So, so there's, so the, all the assumptions about the validity of this training set, test set, uh, hypothesis break-

Lex Fridman 37:51 Mm-hmm.

Yann LeCun 37:51 ... whenever a machine can take an action that has an influence in the, in the world, and it's what is gonna see. So, people are setting up artificial, uh, environments where, where that takes place, right? The robot runs around a, a 3D model of a house and can interact with objects and things like this. So, you do robotics by simulation, you have those, you know, OpenAI gym type thing, or MuJoCo kinda simulated, uh, robots, and you have games, you know, things like that. Um, so that, that's where the field is going really, um, this kind of environment. Now, back to the question of AGI. Like, I don't like the term AGI because it implies that human intelligence is general.

Lex Fridman 38:34 Mm-hmm.

Yann LeCun 38:35 And human intelligence is nothing like general. It's very, very specialized. We think it's general, we like to think of ourselves as having general intelligence, we don't. We're very specialized. We're only slightly more general than-

Lex Fridman 38:47 Why does it feel general? So, you kind of, uh, the, the term general... I think what's impressive about humans is the ability to learn, as we were talking about learning, to learn in just so many different domains. It's perhaps not arbitrarily general, but just you can learn in many domains and integrate that knowledge somehow.

Yann LeCun 39:08 Okay.

Lex Fridman 39:08 And pers- and knowledge persists.

Yann LeCun 39:09 So, let me take a very specific example.

Lex Fridman 39:11 Yes.

Yann LeCun 39:12 It's not an example, it's more like a, a quasi-mathematical demonstration. So, you have about 1 million fibers coming out of one of your eyes, okay? 2 million total, but let's, let's talk about just one of them.

Lex Fridman 39:23 Mm-hmm.

Yann LeCun 39:23 It's 1 million nerve fibers, your optical nerve. Let's imagine that they are binary, so they can be active or inactive, right? So, the input to your visual cortex is 1 million bits.

Lex Fridman 39:34 Mm-hmm.

Yann LeCun 39:36 Now, they're connected to your brain in a particular way, and your brain has connections that are kind of a little bit like a conventional net, they're kind of local, uh, you know, in space and things like this.

Lex Fridman 39:47 Mm-hmm.

Yann LeCun 39:47 Now, imagine I play a trick on you. It's a pretty nasty trick, I admit. I, uh, I cut your optical nerve-

Lex Fridman 39:55 Mm-hmm.

Yann LeCun 39:55 ... and I put a device that makes a random perturbation of, a permutation-

Lex Fridman 40:00 Mm-hmm.

Yann LeCun 40:00 ... of all the nerve fibers. Well, now what comes to your, to your brain is a fixed but random permutation of all the pixels.

Lex Fridman 40:07 Mm-hmm.

Yann LeCun 40:09 There's no way in hell that your visual cortex, even if I do this to you in infancy, will actually learn vision to the same level of quality that you can.

Lex Fridman 40:19 G- got it. And you're saying there's no way you've relearned that?

Yann LeCun 40:22 No, because now two pixels that are nearby in the world will end up in very different places in your visual cortex, and your neurons there have no connections with each other because they're only connected locally.

Lex Fridman 40:33 So, this whole, our entire, the hardware is built in many ways to support-

Yann LeCun 40:38 The locality of the real world.

Lex Fridman 40:39 Yeah.

Yann LeCun 40:40 Yes. That's specialization. Okay, now-

Lex Fridman 40:42 Yeah, but it's still pretty damn impressive. So, it's not perfect generalization. It's, it's not even close.

Yann LeCun 40:47 No, no, it's, it's, it's-

Lex Fridman 40:48 Yeah.

Yann LeCun 40:48 ... it's not that it's not even close, it's not at all, okay?

Lex Fridman 40:51 Yeah, it's not. It's specialized, yeah.

Yann LeCun 40:52 So, so how many Boolean functions... So, le- let's imagine you want to train your visual system to, uh, you know, recognize particular patterns of those 1 million bits, okay? So, that's a Boolean function, right? Either the pattern is here or not here, there's a two, two-way classification with 1 million binary inputs. How many such Boolean functions are there? Okay. You have, you have 2 to the 1 million combinations of inputs. For each of those, you have an output bit, and so you have 2 to the, 2 to the 1 million Boolean functions of this type.

Lex Fridman 41:29 Yeah.

Yann LeCun 41:29 Okay? Which is an unimaginably large number. How many of those functions can actually be computed by your visual cortex? And the answer is a tiny, tiny, tiny, tiny, tiny, tiny sliver.

Lex Fridman 41:41 Okay.

Yann LeCun 41:41 Like an enormously tiny sliver.

Lex Fridman 41:43 Yeah. Yeah.

Yann LeCun 41:45 So, we are ridiculously specialized. You know, okay?

Lex Fridman 41:49 But (laughs) , okay, that's an argument against the word general. I think there-

Yann LeCun 41:54 Yes.

Lex Fridman 41:54 ... there's a w- I, I, uh, there's, I agree with your intuition, but I'm not sure it's, it seems the bra- uh, the, the brain is im- impressively, uh, capable of adjusting to things. So-

Yann LeCun 42:09 It's because we can't imagine tasks that are outside of our comprehension.

Lex Fridman 42:15 Outs- so, yeah (laughs) .

Yann LeCun 42:15 Right? So, we think, we think we are general because we're general of all the things that we can apprehend.

Lex Fridman 42:20 Oh, so, yeah.

Yann LeCun 42:20 But there is a huge world out there of things that we have no idea. We call that heat, by the way.

Lex Fridman 42:26 Heat?

Yann LeCun 42:27 Heat. So, uh (laughs) , at least physicists call that heat, or they call it entropy, which is-

Lex Fridman 42:32 Entropy.

Yann LeCun 42:33 ... kind of, you know, you have a thing full of gas, right?

Lex Fridman 42:38 Mm-hmm. Closed system full of gas.

Yann LeCun 42:40 Right? Closed or not closed. It has, uh, you know, pressure, it has, uh, temperature, it has, you know, and you can write equations, PV equal nRT, you know, things like that, right? When you reduce the volume, the temperature goes up, the pressure goes up, you know, things like that, right? Uh, for a perfect gas at least. Those are the things you can know about that system.And it's a tiny, tiny number of bits compared to the complete information of the state of the entire system.

Lex Fridman 43:10 Mm-hmm.

Yann LeCun 43:10 Because state of the entire system will give you the position and momentum of every, every molecule of the gas. And what you don't know about it is the entropy, and you interpret it as heat. The energy contained in that thing is, is what we call heat. Now, it's very possible that, in fact, there is some very strong structure in how those molecules are moving. It's just that they are in a way that we are just not wired to perceive.

Lex Fridman 43:39 Yeah, we're ignorant to it. And there's a n- near infinite amount of things we're not wired to perceive.

Yann LeCun 43:45 Yeah.

Lex Fridman 43:45 And you're right, that's a, that's a nice way to put it. We're general to all the things we can imagine-

Yann LeCun 43:50 (laughs)

Lex Fridman 43:51 ... which is a very tiny-

Yann LeCun 43:53 Yeah.

Lex Fridman 43:53 ... subset of all things that are possible.

Yann LeCun 43:54 So it's like Kolmogorov complexity or the Kolmogorov Chaitin summing of complexity.

Lex Fridman 43:59 Yeah.

Yann LeCun 44:00 You know, every bit string or every integer is random, except for all the ones that you can actually write down.

Lex Fridman 44:07 (laughs)

Yann LeCun 44:08 (laughs)

Lex Fridman 44:11 Yeah, okay, so beautifully put. But, you know, so we can just call it artificial intelligence. We don't need to have a general. St-

Yann LeCun 44:18 Or human-level, human-level intelligence is good.

Lex Fridman 44:20 T- w- l- You know, y- you'll start... Anytime you touch human, it gets t- it gets interesting because, um, uh, you know, (sighs) it's d- it's just because th- we attach ourselves to human and it's difficult to define what human intelligence is.

Yann LeCun 44:36 Yep.

Lex Fridman 44:37 Nevertheless, my definition is, uh, maybe, uh, damn impressive intelligence, okay?

Yann LeCun 44:43 (laughs)

Lex Fridman 44:44 Damn impressive demonstration of intelligence, whatever.

Yann LeCun 44:46 (laughs)

Lex Fridman 44:46 And so on that topic,

Lex Fridman 44:49 most successes in deep learning have been in supervised learning.

Yann LeCun 44:53 Mm-hmm.

Lex Fridman 44:53 What is your view on unsupervised learning? Is there hope to reduce involvement of human input and still have, uh, successful systems that are have practical use?

Yann LeCun 45:08 Yeah, I mean, there's definitely a hope. It's, it's more than a hope actually. It's, uh, it's, you know, mounting evidence for it, and that's basically all I do. Like the only thing I'm interested in at the moment is, I call it self-supervised learning, not unsupervised, 'cause unsupervised learning is, is a loaded term. People who know something about machine learning, you know, tell you, "So you're doing clustering or PCA?"

Lex Fridman 45:30 Right, right.

Yann LeCun 45:30 Which is not the case. And the wide public, you know, when you say unsupervised learning, "Oh my God, you know, machines are gonna learn by themselves and without supervision?"

Lex Fridman 45:37 (laughs)

Yann LeCun 45:37 You know, they see this as-

Lex Fridman 45:39 Where's, where's the parents? (laughs)

Yann LeCun 45:40 Yeah. So, so I call it self-supervised learning because, in fact, the underlying algorithms that are used are the same algorithms as the supervised learning algorithms, um, except that what we train them to do is not predict a particular set of variables, like the category of a, of an image, and, and not to predict a set of variables that have been provided by human labelers. But what you train the machine to do is basically reconstruct a piece of its input that it's being, is being masked, masked out, essentially. You can think of it this way, right? So show a piece of video to a machine and ask it to predict what's gonna happen next. And of course, after a while, you can show what, what happens and the machine will kinda train itself to do better at that task. You can do, like all the latest, most successful models in natural language processing use self-supervised learning, you know, sort of BERT-style systems, for example, right? You show it a window of a dozen words on a test corpus, you take out 15% of the words, and then you train the machine to predict the words that are missing.

Lex Fridman 46:50 Mm-hmm.

Yann LeCun 46:51 That's self-supervised learning. It's not predicting the future, it's just, you know, predicting things in the middle. But you could have it predict the future. That's what language models do. Um-

Lex Fridman 46:59 So you construct... So in an unsupervised way, you construct a model of language. Do you think-

Yann LeCun 47:05 Or video, or the physical world, or whatever, right?

Lex Fridman 47:09 Reality. How far do you think that can take us? Do you think-

Yann LeCun 47:13 Very far, I think.

Lex Fridman 47:14 ... BERT understands anything?

Yann LeCun 47:18 To some level. It has, you know, a, a shallow understanding of, uh, of text, but it needs to... I mean, to have kind of true human-level intelligence, I think you need to ground language in reality, so some people are attempting to do this, right?

Lex Fridman 47:32 Attempted to do this.

Yann LeCun 47:32 Having systems that can have some visual representation of what, what is being talked about, which is one reason you need those interactive environments actually. But there's like a, a huge technical problem that is not solved, and that explains why self-supervised learning works in the context of natural language, but does not work in the context, or at least not well, in the context of image recognition and, uh, video, although it's making progress quickly. And the reason, uh, that reason is the fact that it's much easier to represent uncertainty in the prediction in the context of natural language than it is in the context of things like video and images. So for example, if I ask you to predict what words are missing, you know, 15% of the words that are-

Lex Fridman 48:15 Right.

Yann LeCun 48:15 ... have taken out.

Lex Fridman 48:17 The possibility is just small. I mean that-

Yann LeCun 48:19 It's small, right? There is 100,000 words in the, in the lexicon, and you, what the machine spits out is a, a big probability vector, right? It's a bunch of numbers between zero and one that sum to one. And we know how to do, how to do this with computers.

Lex Fridman 48:33 Right.

Yann LeCun 48:34 So there, representing uncertainty in the prediction is relatively easy, and that's, in my opinion, why those techniques work for NLP. For images, if you ask... if you block a piece of an image and you ask the system reconstruct that piece of the image, there are many possible answers. Uh, they are all perfectly legit, right? And how do you represent that th- this set of possible answers? You, you can't train a system to make one prediction. You can't train a neural net to say, "Here it is. That's the image," because it's, it, there's a whole set of things that are compatible with it. So how do you get the machine to represent not a single output, but a whole set of outputs?And, y- you know, similarly with, uh, video prediction. There's a lot of things that can happen in the future of video. You're looking at me right now, I'm not-

Lex Fridman 49:21 Yeah.

Yann LeCun 49:21 ... moving my head very much. But, you know, I might, you know, turn my, my head to the left or to the right.

Lex Fridman 49:26 Right.

Yann LeCun 49:27 If you don't have a system that can predict this, and you train it with least square to kind of minimize the error with the prediction and what I'm doing, what you get is a blurry image of myself-

Lex Fridman 49:36 Right.

Yann LeCun 49:37 ... in all possible future positions that I might be in, which is not a good prediction.

Lex Fridman 49:41 But, so there might be other ways to do the self-supervision, right? For visual scenes.

Yann LeCun 49:48 Like what? (laughs)

Lex Fridman 49:49 If I t- I mean, if I knew, I wouldn't tell you.

Yann LeCun 49:52 (laughs)

Lex Fridman 49:52 I would publish it first. I don't know. Uh, uh, uh-

Yann LeCun 49:55 No, there might be.

Lex Fridman 49:56 (laughs) So, I mean, and these are kind of, uh, there might be artificial ways of like self-play in games, the way you can simulate part of the environment. You can-

Yann LeCun 50:05 Oh, that doesn't solve the problem. It's just a way of generating data.

Lex Fridman 50:10 But because you have more of a cont- like, I mean, you can control... Yeah, it's a way to generate data.

Yann LeCun 50:16 Mm-hmm.

Lex Fridman 50:16 That's right. And because you can do huge amounts of data generation, that doesn't... You're right. That's... Well, it's-

Yann LeCun 50:22 I-

Lex Fridman 50:22 It's a creeps up on the problem from the side of data. (laughs) And you don't think that's the right way to creep up on the problem?

Yann LeCun 50:27 It doesn't solve this problem of handling uncertainty in the world.

Lex Fridman 50:30 Yeah.

Yann LeCun 50:30 Right? So, if you, if you have a machine, uh, learn a predictive model of the world in a game that is deterministic or quasi-deterministic, uh, it's easy, right? Just, you know, give a few frames of the game to a ConvNet, put a bunch of layers, and then have... The game generates the next few frames.

Lex Fridman 50:49 Mm-hmm.

Yann LeCun 50:49 And, and if the game is deterministic, it works fine. Um, and that includes, you know, feeding th- th- the system with the action that your little character is, is gonna take. The problem comes from the fact that the real world and c- most games are not entirely predictable. And so there, you get those blurry predictions, and you can't do planning with blurry predictions. Right? So if you have a perfect model of the world, you can, in your head, run this model with a hypothesis for a sequence of actions, and you're going to predict the outcome of that sequence of actions.

Lex Fridman 51:26 Mm-hmm.

Yann LeCun 51:28 But if your model is imperfect, how can you plan?

Lex Fridman 51:32 Yeah, it quickly explodes.

Lex Fridman 51:34 What are your thoughts on the extension of this, which topic I'm super excited about, it's connected to something you were talking about in terms of robotics, is active learning. So, as opposed to sort of unsup- c- completely unsupervised or self-supervised learning, you ask the system for human help for, for selecting parts you want to annotate next.

Yann LeCun 51:58 Uh-huh.

Lex Fridman 51:58 So, if you think about a robot exploring a space, or a baby exploring a space, or a system exploring a dataset, every once in a while, asking for human input.

Yann LeCun 52:07 Mm-hmm.

Lex Fridman 52:08 Do, do you see value in that kind of work? Do, do-

Yann LeCun 52:12 I don't see transformative value. Uh, it- it's gonna make things that we can already do more efficient, or they will learn slightly more efficiently.

Lex Fridman 52:20 Yeah.

Yann LeCun 52:20 But it's not gonna make machines sort of significantly more intelligent. I think... And I... And, and by the way, there is no opposition, there is no, um, uh, conflict between self-supervised learning, reinforcement learning, and supervised learning, or imitation learning, or active learning.

Lex Fridman 52:37 Right.

Yann LeCun 52:39 I see self-supervised learning as a, as a preliminary to all of the above.

Lex Fridman 52:43 Yes.

Yann LeCun 52:44 So, the example I use very often is, how is it that... So if you use classical reinforcement learning, d- deep reinforcement learning, if you want, the best methods today, s- so-called model-free reinforcement learning, to learn to play Atari games, uh, take about 80 hours of training to reach the level that any human can reach in about 15 minutes.

Lex Fridman 53:09 Mm-hmm.

Yann LeCun 53:11 Um, they get better than humans, but it takes them a long time. Uh, AlphaStar, okay, the, you know, Ariel Vinial's and, and his team's, uh, system to play, um, to, um, to play StarCraft, uh, plays, you know, a single map, a single type of, uh, player, and can reach better than human level with about the equivalent of 200 years of training, playing against itself. It's 200 years, right? It's not some- something that no, no human can-

Lex Fridman 53:48 (laughs)

Yann LeCun 53:49 ... could ever do. Now-

Lex Fridman 53:50 I mean, I'm not sure what else to take away from that.

Yann LeCun 53:52 Okay. Now, take those algorithms, the best RL algorithms we have today, to train a, a car to drive itself.

Lex Fridman 54:01 Hmm.

Yann LeCun 54:01 It would probably have to drive millions of hours. It will have to kill thousands of pedestrians. It will have to run into thousands of trees. It will have to run off cliffs.

Lex Fridman 54:09 Yeah.

Yann LeCun 54:09 And it had to run off cliff multiple times before it figures out that it's a bad idea, first of all.

Lex Fridman 54:14 Yeah.

Yann LeCun 54:15 And second of all, before it figures out how not to do it. And so, I mean, this type of learning obviously does not reflect the kind of learning that animals and humans do. There is something missing that's really, really important there. And my hypothesis, which I've been advocating for like five years now, is that we have predictive models of the world that include the ability to predict under uncertainty. And what allows us to not run off a cliff when we learn to drive, most of us can learn to drive in about 20 or 30 hours of training without ever crashing, causing any accident.

Lex Fridman 54:52 Mm-hmm.

Yann LeCun 54:52 And if we drive next to a cliff, we know that if we turn the wheel to the right, the car is gonna run off the cliff and nothing good is gonna come out of this. Because we have a pretty good model of intuitive physics that tells us, you know, the car is gonna fall. We know, we know about gravity. Babies learn this around the age of eight or nine months, that objects don't float, they fall.

Lex Fridman 55:11 Mm-hmm.

Yann LeCun 55:12 Um, and, you know, we have a pretty good idea of the effect of turning the wheel on the car, and you know, we know we need to stay on the road. So there's a lot of things that we bring to the table, which is basically our predictive model of the world...And that model allows us to not do stupid things and to basically stay within the context of, uh, things we need to do. We still face, you know, unpredictable situations and that's how we learn, but that allows us to learn really, really, really quickly. So that's called model-based reinforcement learning. There's some imitation and supervised learning because we have a driving instructor that tells us occasionally what to do. Uh, but most of the learning-

Lex Fridman 55:50 Is model-based.

Yann LeCun 55:51 ... is learning the model-

Lex Fridman 55:53 Yeah.

Yann LeCun 55:53 ... learning physics that we've done since we were babies. That's where all, almost all the learning takes place.

Lex Fridman 55:58 And the physics is somewhat transferrable from, it's transferrable from scene to scene? Stupid things are the same everywhere.

Yann LeCun 56:05 Yeah. I mean, if you, you know, y- you have experience of the world, you don't need to be, uh, particularly, from a particularly intelligent species to know that if you spill water from a container, the, you know, the, the rest is gonna get wet, and you might get wet.

Lex Fridman 56:21 (laughs)

Yann LeCun 56:21 Um, so, you know, cats know this, right?

Lex Fridman 56:24 Yeah.

Yann LeCun 56:24 Um, so the main p- problem we need to solve is how do we learn models of the world? That's, and that's what I'm interested in. That's what self-supervised learning is all about.

Lex Fridman 56:34 If you were to try to construct a benchmark for, uh, let's, let's look at MNIST. I love that dataset.

Yann LeCun 56:42 (laughs)

Lex Fridman 56:42 But, uh, uh, if y- do you think it's useful, interesting/possible to, uh, perform well on MNIST with just one example of each digit? And how would we solve that problem?

Yann LeCun 56:58 The answer is probably yes. The question is what, uh, other type of learning are you allowed to do?

Lex Fridman 57:03 Yes.

Yann LeCun 57:03 So if what you're allowed to do is train on some gigantic dataset of labeled digit, that's called transfer learning. And we know that works, okay?

Lex Fridman 57:10 Yeah.

Yann LeCun 57:11 Uh, we do this at Facebook, like, in production, right? We, we train large convolutional nets to predict hashtags that people type on Instagram, and we train on billions of images, literally billions. And, uh, and then we chop off the last layer and fine-tune on whatever task we want. That works really well. You can beat, you know, the ImageNet record with this. We actually, uh, open-sourced the whole thing, like, a few weeks ago.

Lex Fridman 57:31 Yeah-

Yann LeCun 57:32 Um-

Lex Fridman 57:32 ... that's, that's still pretty cool. But yeah, so what, and yet, what would be-

Yann LeCun 57:35 Mm-hmm.

Lex Fridman 57:35 ... impressive, and, and what's useful and impressive? What kind of transfer learning would be useful and impressive? Is it Wikipedia, that kind of thing?

Yann LeCun 57:42 No, no.

Lex Fridman 57:42 Natural language?

Yann LeCun 57:43 So I don't think transfer learning is really where we should focus. We should try to do, you know, have a kinda scenario for a benchmark where you have unlabeled data, and you can ... And it's very large number of unlabeled data. Uh, it could be video clips. It could be where you do, y- you know, frame prediction. It could be images where you could choose to, you know, mask a piece of it.

Lex Fridman 58:09 Mm-hmm.

Yann LeCun 58:10 It could be whatever, but they're unlabeled, and you're not allowed to label them. So you do some training on this, and then you train on a particular supervised task, ImageNet or MNIST, and you measure how your test error decrease- or variation error decreases as you increase the number of labeled training samples.

Lex Fridman 58:34 Mm-hmm.

Yann LeCun 58:36 Okay? And, and what, what you'd like to see is, is that y- you know, your, your error decreases much faster than if you train from scratch, from random weights. So that to reach the same level of performance in a completely supervised, uh, purely supervised system would reach, you would need way fewer samples. So, that's the crucial question because it will answer the question to, like, you know, people are interested in medical image analysis. Okay, you know, if I wanna get to a particular, uh, level of, uh, of error rate for this task, uh, I know I need, um, a million samples, can I do, you know, self-supervised pre-training to reduce this to about 100 or something?

Lex Fridman 59:17 And you think the answer there is self-supervised pre-training?

Yann LeCun 59:20 Yeah. Some form. Some form, some form of it.

Lex Fridman 59:25 Telling you, active learning, but you disagree. You, you are-

Yann LeCun 59:28 No, it's u- it's not useless. It's just not gonna lead to a quantum leap. It's just gonna make things that we already do-

Lex Fridman 59:34 So you're way smarter than me.

Yann LeCun 59:35 ... better and efficient.

Lex Fridman 59:35 I just disagree with you.

Yann LeCun 59:37 (laughs)

Lex Fridman 59:37 And, but I don't have anything to back that. It's just intuition. So I worked with a lot of large-scale datasets, and there's something that might be magic in active learning, but okay. At least I said it publicly. (laughs)

Yann LeCun 59:50 (laughs)

Lex Fridman 59:50 At least I'm being an idiot publicly. Okay.

Yann LeCun 59:53 It's not being an idiot. It's, you know, working with the data you have. I mean, I mean, certainly people are doing things like, okay, I have 3,000 hours of, you know, uh, imitation learning for self-driving car, but most of those are incredibly boring. What I like is select, you know, 10% of them that are kind of the most informative. And with just that, I would probably s- reach the same. So it's a weak form of, uh, of, of active learning if you want.

Lex Fridman 1:00:16 Yes.

Yann LeCun 1:00:16 Right?

Lex Fridman 1:00:16 But there might be a r- a, a, a much stronger version.

Yann LeCun 1:00:20 Yeah, that's right.

Lex Fridman 1:00:20 And that's what-

Yann LeCun 1:00:21 When the machine asks- Yeah.

Lex Fridman 1:00:22 ... if it exists.

Yann LeCun 1:00:23 Yeah.

Lex Fridman 1:00:23 Uh, uh, the g- the question is how much stronger can you get?

Lex Fridman 1:00:26 Elon Musk is confident, I talked to him recently, uh, he's confident that large-scale data and deep learning can solve the autonomous driving problem. What are your thoughts on the limitless possibilities of deep learning in this space?

Yann LeCun 1:00:40 Well, it's, it's obviously part of the solution. I mean, I don't think we'll ever have a self-driving system, or at least not in the foreseeable future, that does not use deep learning. Let me put it this way. Now, how much of it? So in the history of sort of engineering, particularly sort of, sort of AI-like systems, there's, uh, generally a first phase where everything is built by hand, then there is a second phase, and that was the case for autonomous driving, you know, 20, 30 years ago. There's a phase where there's a little bit of learning is used, but there's a lot of engineering that's involved in kind of, you know, taking care of corner cases and, and putting limits, uh, et cetera because the learning system is not perfect. And then as technology progresses-... we end up relying more and more on learning. That's the history of character recognitions, the history of speech recognition, now computer vision, natural language processing. And I think the same is going to happen with, uh, with autonomous driving, that currently the, the, uh, the methods that are closest to providing some level of autonomy, some, you know, decent level of autonomy, where you don't expect a driver to kind of do anything, is where you constrain the world. So, you only run within, you know, 100 square kilometers or square miles in Phoenix, where the weather is nice and the roads are wide. Which is what Waymo is doing. You, uh, completely over-engineer the car with tons of LiDARs and sophisticated sensors that are too expensive for consumer cars, but they're fine if you just run a fleet. And you engineer the thing, uh, the hell out of the, everything else. You, you map the entire world so you have complete 3D model of everything, so the only thing that the perception system has to take care of is moving objects and, and, and construction and sort of, you know, things that, that weren't in your map. And you can engineer a good, you know, SLAM system and all that stuff, right?

Lex Fridman 1:02:33 Mm-hmm.

Yann LeCun 1:02:34 So, so that's kind of the current approach that's closest to some level of autonomy, but I think eventually, the long-term solution is gonna rely more and more on learning and possibly using a combination of self-supervised learning and model-based reinforcement or something like that.

Lex Fridman 1:02:50 But ultimately, learning will be at, not just at the core, but really the fundamental part of the system?

Yann LeCun 1:02:57 Yeah. It already is, but it'll become more and more.

Lex Fridman 1:03:00 What do you think it takes to build a system with human-level intelligence? You talked about the AI system in the movie Her being way out of reach, our current reach.

Yann LeCun 1:03:09 Mm-hmm.

Lex Fridman 1:03:10 This might be outdated as well. But, uh-

Yann LeCun 1:03:12 It's still way out of reach.

Lex Fridman 1:03:13 ... it's still way out of reach. Uh, what would it take to build Her, do you think?

Yann LeCun 1:03:19 Um, so I can tell you the first two obstacles that we have to clear, but I don't know how many obstacles there are after this. So, the image I usually use is that there is a bunch of mountains that we have to climb. And we can see the first one, but we don't know if there are 50 mountains behind it or not. And this might be a good sort of metaphor for why AI researchers in the past have been overly optimistic about the result of, uh, of AI. You know, for example, um, uh, Newell and Simon, right, wrote the General Problem Solver, and they called it the General Problem Solver.

Lex Fridman 1:03:51 (laughs) General Problem Solver.

Yann LeCun 1:03:52 Okay? And, of course, the first thing you realize is that all the problems you want to solve are exponential, and so you can't actually use it for anything useful. But, you know-

Lex Fridman 1:04:00 Yeah. So, yeah, all you see is the first peak. So, in general, what are the first couple of peaks for Her?

Yann LeCun 1:04:05 Um, so the first peak, which is precisely what I'm working on, is, uh, self-supervised learning. How, how do we get machines to learn models of the world by observation, kind of like babies and like young animals? So, I- we've been working with, um, you know, cognitive scientists. So, this Emmanual Dupoux, who's at FAIR in Paris is, uh... Half time is also a researcher in French university. And he, he, um, he has this chart that shows that which- how many months of, of life baby humans kind of learn different concepts. And you can mea- you can measure this in sort of various ways. So, things like, um, uh, distinguishing animate objects from inanimate, inanimate object, you can, you can tell the difference at age two, three months. Whether an object is gonna stay stable or it's gonna fall, you know, about four months you can tell.

Lex Fridman 1:05:03 Mm-hmm.

Yann LeCun 1:05:03 You know, there, there are various things like this. And then things like gravity, the fact that objects are not supposed to float in the air, but are supposed to fall, you learn this around the age of eight or nine months. If you look at a lot of, you know, eight-month-old babies, you give them a bunch of toys on their high chair, first thing they do is to throw them on the ground and they look at them. It's because, you know, they're learning about- actively learning-

Lex Fridman 1:05:24 Yeah.

Yann LeCun 1:05:25 ... about gravity.

Lex Fridman 1:05:26 Gravity, yeah. (laughs)

Yann LeCun 1:05:26 Okay? So, they're not trying to annoy you, but they, you know, they need to do the experiment, right?

Lex Fridman 1:05:32 Yeah.

Yann LeCun 1:05:33 Uh, so, you know, how do we get machines to learn like babies? Mostly by observation with a little bit of interaction, and learning those, those, those models of the world, because I think that's really a crucial piece of an intelligent, autonomous system. So, if you think about the architecture of an intelligent, autonomous system, it needs to have a predictive model of the world. So, something that says, "Here is the world at time T. Here is the state of the world at time T plus one if I take this action." And it's not a single answer. It can be a-

Lex Fridman 1:05:59 Yeah, it can be a distribution, yeah.

Yann LeCun 1:06:01 ... 'cause it's functions. Yeah. Well, but we don't know how to represent distributions in high-dimensional continuous spaces.

Lex Fridman 1:06:04 Vectors in 3D. That's fine.

Yann LeCun 1:06:04 So, it's gotta be something weaker than that, okay?

Lex Fridman 1:06:06 Right, yeah.

Yann LeCun 1:06:07 But with some, some representation of uncertainty. If you have that, then you can do what optimal control theorists call model predictive control, which means that you can run your model with a hypothesis for a sequence of action and then see the result. Now, what you need, the other thing you need is some sort of objective that you want to optimize. "Am I reaching the goal of grabbing this object? Am I minimizing energy? Am I..." Whatever, right? So, there is some sort of objective that you have to minimize. And so in your head, if you have this model, you can figure out the sequence of action that will optimize your objective. That objective is something that ultimately is rooted in your basal ganglia, at least in the human brain. That's, that's what it is. Basal ganglia computes your level of contentment or, or miscontentment. I don't know if that's a word.

Lex Fridman 1:06:54 (laughs) Unhappiness.

Yann LeCun 1:06:54 Unhappiness, okay?

Lex Fridman 1:06:55 Unhappiness, yeah. Dis- yeah.

Yann LeCun 1:06:56 Discontentment.

Lex Fridman 1:06:57 Discontentment, right.

Yann LeCun 1:06:58 Um, and so your entire behavior is driven towards kind of minimizing that objective, which is maximizing your contentment computed by your, your basal ganglia.

Lex Fridman 1:07:10 Yeah.

Yann LeCun 1:07:10 And what you have is an objective function, which is basically a predictor of what your basal ganglia is gonna tell you.

Lex Fridman 1:07:16 Mm-hmm.

Yann LeCun 1:07:17 So, you're not gonna put your hand on fire because you know it's gonna, you know it's gonna burn and you're gonna get hurt. And you're predicting this because of your model of the world and your, your sort of predictor of this objective, right?So, you, if you have those, you have those three components, you have the, you have four components. You have the, the hardwired, uh, contentment objective, um, comp- computer, if you want, calculator, and then you have those three components. One is the objective predictor, which basically predicts your level of contentment. One is the model of the world, and there's a third module I didn't mention, which is a module that will figure out the best course of action to optimize an objective given your model.

Lex Fridman 1:08:02 Hmm.

Yann LeCun 1:08:02 Okay?

Lex Fridman 1:08:03 Yeah.

Yann LeCun 1:08:04 Call this a policy, policy network or something like that, right? Now, you need those three components to act autonomously, intelligently. And you can be stupid in three different ways. You can be stupid because your model of your, of the world is wrong.

Lex Fridman 1:08:18 Mm-hmm.

Yann LeCun 1:08:19 You can be stupid because your objective is not aligned with what you actually want to achieve, okay? Uh...

Lex Fridman 1:08:27 Then possibly-

Yann LeCun 1:08:27 In humans, that would be a psychopath.

Lex Fridman 1:08:29 Right.

Yann LeCun 1:08:30 And then the, the third thing you be- the third way you can be stupid is that you have the right model, you have the right objective, but you're unable to figure out a course of action to optimize your objective given your model.

Lex Fridman 1:08:40 Right.

Yann LeCun 1:08:41 Okay. Um...

Lex Fridman 1:08:43 (laughs)

Yann LeCun 1:08:44 Some people who are in charge of big countries actually have all three that are wrong.

Lex Fridman 1:08:47 All right. (laughs) Uh, which countries? I don't know. Okay. So, uh,

Lex Fridman 1:08:53 if, if we think about this, this agent, if you think about the movie Her, you've criticized the art project that is Sophia the robot, and what that project essentially does is uses our natural inclination to anthropomorphize things that look like human and give them more. Do you think that could be used by AI systems, like in the movie Her? So, do you think that body is needed to create a feeling of intelligence?

Yann LeCun 1:09:26 Uh, well, if Sophia was just an art piece, I would have no problem with it. But it's presented as something else.

Lex Fridman 1:09:33 Let me add that comment real quick. If creators of Sophia could change something about their marketing or behavior in general, what would it be? What, what's...

Yann LeCun 1:09:42 Oh, just about everything.

Lex Fridman 1:09:44 (laughs) I mean, don't you think... Here's a tough question. Let me... So, I agree with you. So, Sophia is not, uh, the general public feels that Sophia can do way more than she actually can.

Yann LeCun 1:09:59 That's right.

Lex Fridman 1:10:00 And the people who created Sophia are not honestly, publicly communicating, uh, trying to teach the public.

Yann LeCun 1:10:09 Right.

Lex Fridman 1:10:09 But here's a tough question. Don't you think, uh, the s- the same thing is, uh, scientists in industry and research are taking advantage of the same mis- misunderstanding in the public when they create AI companies or publish stuff, um...

Yann LeCun 1:10:30 Some companies, yes. I mean, there is no sense of... There's no desire to delude. There's no desire to s- to kinda overclaim when something is done.

Lex Fridman 1:10:38 Right.

Yann LeCun 1:10:38 Right? You publish a paper on AI that, you know, has this result on ImageNet, you know, it's pretty clear. I mean, it's not ev- not even interesting anymore. But, you know, I, I don't think there is that... I mean, the reviewers are generally not very forgiving of (laughs) -

Lex Fridman 1:10:53 Right.

Yann LeCun 1:10:53 ... of, you know, unsupported claims of, of this type. And... But there are certainly quite a few startups that have had a huge amount of hype around this that, uh, I find extremely damaging, and I've been calling it out when I've seen it. Um, so yeah. But to go back to your original question-

Lex Fridman 1:11:10 Sure.

Yann LeCun 1:11:10 ... like, the necessity of embodiment. I think, I don't think embodiment is necessary. I think grounding is necessary. So, I don't think we're gonna get machines that really understand language without some level of grounding in the real world. And it's now clear to me that language is a high enough bandwidth medium to communicate how the real world works. Um, I think for this-

Lex Fridman 1:11:30 Can you talk to ground- uh, what grounding means to you?

Yann LeCun 1:11:32 So, grounding means that... So, there is this classic problem of, uh, common sense, uh, reasoning, you know, the, the Winograd's, Winograd schema, right? And so, I tell you the, the trophy doesn't fit in the suitcase because it's too, it's too big, or the trophy doesn't fit in the suitcase because it's too small, and the "it" in the first case refers to the trophy, in the second case, to the suitcase. And the reason you can figure this out is because you know where the trophy and the suitcase are, you know one is supposed to fit in the other one, and you know the notion of size, and a big object doesn't fit in a small object, and this is a TARDIS, you know, things like that, right? So, you have this co- this knowledge of how the world works, of geometry, and things like that. Um, I don't believe you can learn everything about the world by just being told in language how the world works. I think you need some, uh, low level perception of the world, you know, be it visual, touch, you know, whatever, but some higher bandwidth perception of the world.

Lex Fridman 1:12:26 So, by reading all the world's text, you still may not have enough information.

Yann LeCun 1:12:31 That's right. There's a lot of things that just will never appear in text.

Lex Fridman 1:12:35 Wow.

Yann LeCun 1:12:35 And that you can't really in- infer. So, I think-

Lex Fridman 1:12:38 Yeah.

Yann LeCun 1:12:38 ... uh, common sense would emerge from, you know, certainly a lot of language interaction, but also with watching videos or perhaps even interacting in vu- in virtual environments, and possibly, you know, robot interacting in the real world. But I don't actually believe necessarily that this last one is absolutely necessary, but I think there, there's a, a need for some grounding.

Lex Fridman 1:13:00 But the final product doesn't necessarily need to be embodied-

Yann LeCun 1:13:04 No.

Lex Fridman 1:13:04 ... you're saying.

Yann LeCun 1:13:04 No.

Lex Fridman 1:13:05 And it just needs to have an awareness, uh, a grounding to

Yann LeCun 1:13:07 Right. But it needs to know how the world works to have, you know, to not be frustrated, frustrating to talk to.

Lex Fridman 1:13:14 Mm-hmm. And, uh, you talked about emotions being important. Uh, that's, that's a whole nother topic. But-

Yann LeCun 1:13:21 Well, so, you know, I talked about this, uh, the, the basal ganglia, uh, ganglia as the... You, your, your, you know, the thing that calcu- you know, calculates your level of miscontentment, contentment, and then there is this, uh, other module that sort of tries to do a prediction of whether you're gonna be content or not.

Lex Fridman 1:13:38 Mm-hmm.

Yann LeCun 1:13:38 That's the source of some emotion. So, fear, for example, is an anticipation of, uh, bad things that can happen to you, right?You have this inkling that there is some chance that something really bad is going to happen to you, and that creates fear. When you know for sure that something bad is going to happen to you, you kind of give up, right? It's not here anymore. It's uncertainty that creates fear. So the punchline is-

Lex Fridman 1:14:00 Yes.

Yann LeCun 1:14:01 -we're not going to have autonomous intelligence without emotions. Okay.

Lex Fridman 1:14:05 (laughs) Whatever the heck emotions are. So you mentioned very practical things of fear, but there's a lot of other mess around it.

Yann LeCun 1:14:13 But they are kind of the results of, you know, drives.

Lex Fridman 1:14:16 Yeah. There's deeper biological stuff going on. And I've talked to a few folks on this. There's fascinating stuff that ultimately connects to our

Lex Fridman 1:14:26 brain. If we create an AGI system... Sorry.

Yann LeCun 1:14:31 Human-level intelligence.

Lex Fridman 1:14:31 Human-level intelligence system, and you get to ask her one question, what would that question be?

Yann LeCun 1:14:38 (laughs) You know, I think the first one we'll create will probably not be that smart.

Lex Fridman 1:14:45 (laughs)

Yann LeCun 1:14:45 They'll be like a four-year-old. Okay.

Lex Fridman 1:14:47 So you would have to ask her a question to know she's not that smart?

Yann LeCun 1:14:52 Yeah.

Lex Fridman 1:14:54 Well, what's a good question to ask, you know, to be impressive?

Yann LeCun 1:14:57 What is the cause of wind? And if she answers, "Oh, it's because the leaves of the tree are moving and that creates wind," she's onto something.

Lex Fridman 1:15:08 And if she says that's a stupid question, she's really onto something.

Yann LeCun 1:15:12 No. And then you tell her, actually, you know, here is the real thing. And she says, "Oh, yeah, that makes sense."

Lex Fridman 1:15:20 So questions that reveal the ability to do common sense reasoning about the physical world.

Yann LeCun 1:15:27 Yeah. And, you know, something that we'll call toning inference.

Lex Fridman 1:15:30 Causal inference. Well, it was a huge honor. Congratulations-

Yann LeCun 1:15:34 It was a pleasure.

Lex Fridman 1:15:34 -on the Turing Award.

Yann LeCun 1:15:35 Thank you.

Lex Fridman 1:15:35 Yann, thank you so much for talking today.

Yann LeCun 1:15:37 Thank you.

Lex Fridman 1:15:37 Appreciate it.