Lex Fridman PodcastDouglas Lenat: Cyc and the Quest to Solve Common Sense Reasoning in AI | Lex Fridman Podcast #221
EVERY SPOKEN WORD
150 min read · 30,011 words- 0:00 – 1:11
Introduction
- LFLex Fridman
The following is a conversation with Doug Lenat, creator of Cyc. A system that for close to 40 years and still today, has sought to solve the core problem of artificial intelligence, the acquisition of common sense knowledge and the use of that knowledge to think, to reason, and to understand the world. To support this podcast, please check out our sponsors in the description. As a side note, let me say that in the excitement of the modern era of machine learning, it is easy to forget just how little we understand exactly how to build the kind of intelligence that matches the power of the human mind. To me, many of the core ideas behind Cyc in some form, in actuality or in spirit, will likely be part of the AI system that achieves general super intelligence. But perhaps more importantly, solving this problem of common sense knowledge will help us humans understand our own minds, the nature of truth, and finally how to be more rational and more kind to each other. This is the Lex Fridman podcast and here is my conversation with Doug Lenat.
- 1:11 – 9:17
What is Cyc?
- LFLex Fridman
Cyc is a project launched by you in 1984 and still is active today, whose goal is to assemble a knowledge base that spans the basic concepts and rules about how the world works. In other words, it hopes to capture common sense knowledge, which is a lot harder than it sounds. (laughs) Can you elaborate on this mission and maybe perhaps speak to the various sub-goals within this mission?
- DLDouglas Lenat
When I was a faculty member in the computer science department at Stanford, my colleagues and I did research in all sorts of artificial intelligence programs. So, natural language understanding programs, robots, expert systems, and so on. And we kept hitting the very same brick wall. Our systems would have impressive early successes, and so if your only goal was academic, namely to, um, get enough material to write a journal article, uh, that might actually suffice. But if you're really trying to get AI, um, then you have to somehow get past the brick wall, and the brick wall was the programs didn't have what we would call common sense. They didn't have general world knowledge. They didn't really understand what they were doing, what they were saying, what they were being asked. Um, and so very much like a, um, um, a clever dog performing tricks, we could get them to do tricks but they never really understood what they were doing. Sort of like when, uh, you get a dog to fetch your morning newspaper. Uh, the dog might do that successfully, but the dog has no idea what a newspaper is or what it says or anything like that.
- LFLex Fridman
What does it mean to understand something? Can you maybe elaborate on that a little bit? Is it ... Is understand an action of, like, combining little things together, like, through inference or is understanding the wisdom you gain over time that forms a knowledge?
- DLDouglas Lenat
I think of understanding more like a, um, uh, think of it more like the ground you stand on which, um, could be very shaky, could be very unsafe, um, but most of the time is not because underneath it is more ground, um, and eventually, you know, rock and, and other things. But, um, layer after layer after layer, that solid foundation is there, and you rarely need to think about it, you rarely need to count on it, but occasionally you do. And, um, um, I've never used this analogy before so bear with me, but, um, I think the same thing is true in, in terms of getting computers to understand things which is, uh, you ask a computer a question, for instance Alexa or, um, some robot or something, and, um, maybe it gets the right answer. Um, but if you were asking that of a human, you could also say things like, "Why?" Or, "How might you be wrong about this?" Or something like that. And the person, you know, would, would answer you, and, you know, it might be a little annoying if you have a small child and they keep asking why questions in series. Eventually, you get to the point where you throw up your hands and say, "I don't know. It's just the way the world is." Um, but for many layers, you actually have that, that layered solid foundation of support so that when you need it, you can count on it. And when do you need it? Well, when things are unexpected, when you come up against a situation which is novel. For instance, when you're driving, um, it may be fine to have a small program, a small set of rules that cover, you know, 99% of the cases, but that 1% of the time when something strange happens, um, you really need to draw on common sense. For instance, um, my wife and I were driving recently, um, and there was a trash truck in front of us, and, um, I guess they had packed it too full and the back exploded and trash bags went everywhere, uh, and we had to, you know, make a split second decision, are we going to slam on our brakes or are we gonna swerve into another lane? Um, are we gonna just run it over? Um, because there were cars all around us, um, and, you know, in front of us was a large, um, trash bag and we know what we throw away in trash bags, probably not a safe thing to run over.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
Um, over on the, the left was, um, a bunch of fast food restaurant, um, trash bags-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... and it's like, "Oh, well, those things are just, like, styrofoam and leftover food. We'll run over that." And so that was a safe thing for us to, to do. Now, that's the kind of thing that's gonna happen maybe once in your life (laughs) and, um, but the point is that there's almost no telling...... what little bits of knowledge about the world you might actually need, um, in some situations which were unforeseen.
- LFLex Fridman
But see, when you sit on that mountain or that ground that goes deep of knowledge, in order to make a split-second decision about, uh, fast food trash or random trash from the back of a, a trash truck, uh, y- you need to be able to leverage that ground you stand on in some way. It's not merely, you know, it's, it's not enough to just have a lot of ground to stand on, it's your ability to leverage it, to utilize in a split... like integrate it all together to make that split-second decision. And I, I, I suppose understanding isn't just having, uh, common sense knowledge to access, it's the act of accessing, accessing it somehow. Like cor- correctly, um, filtering out the n- the parts of the knowledge that are not useful, selecting only the useful parts, and, uh, effectively making conclusive decisions.
- DLDouglas Lenat
So, let's tease apart two different tasks really, both of which are, um, incredibly important and even necessary, um, if you're going to have this in a usable, uh, useful, usable fashion as opposed to say like library books sitting on a shelf-
- LFLex Fridman
Right.
- DLDouglas Lenat
... um, and so on. Um, where the knowledge might be there, but, you know, if, uh, a fire comes, the books are gonna burn because they don't know what's in them and they're just gonna, um, sit there while they burn. Um, so there are two, there are two aspects of using the knowledge. One is a kind of a theoretical, um, how is it possible at all, and then the second aspect of what you said is how can you do it quickly enough?
- LFLex Fridman
Right.
- DLDouglas Lenat
Um, so how can you do it at all is something that philosophers have grappled with. And fortunately, philosophers 100 years ago and, um, even earlier, um, developed a kind of, um, formal language, um, like English, um, it's called, um, uh, predicate logic or first order logic or, um, something like predicate calculus and so on. So there's a way of representing things in this formal language, um, which enables a mechanical procedure to sort of grind through and algorithmically produce all of the same logical entailments, all the same logical conclusions that you or I would from that same set of pieces of information that are represented that way. Um, so, um, that, that sort of raises, um, a couple questions. One is how do you get all this information from say observations and English and so on into this logical form?
- LFLex Fridman
Right.
- DLDouglas Lenat
And secondly, how can you then efficiently, um, run th- these algorithms to actually get the information you need in, in the case I mentioned, in a tenth of a second rather than say, um, in, you know, 10 hours or 10,000 years of computation?
- LFLex Fridman
Right.
- DLDouglas Lenat
Um, and those are both really important, um, uh, questions.
- LFLex Fridman
And, and like a corollary addition to the first one is
- 9:17 – 19:43
How to form a knowledge base of the universe
- LFLex Fridman
how many such things do you need to gather for it to be useful in certain contexts? So like what... in order... y- you mentioned philosophers. In order to capture this world and represent it in a logical way and a f- and with a form of logic, like how many statements are required? Is it five? Is it 10? Is it 10 trillion? Is it like... tha- that's, as far as I understand is, uh, probably still an open question and may forever be an open question, uh, to s- to say like definitively about. To describe the universe perfectly-
- DLDouglas Lenat
Well, I, I-
- LFLex Fridman
... how many facts do you need? (laughs)
- DLDouglas Lenat
I c- (laughs) I, I guess I'm going to disappoint you by giving you an actual answer to your question.
- LFLex Fridman
Oh, man. Okay.
- DLDouglas Lenat
Um-
- LFLex Fridman
Well, no. Uh, this sounds exciting. (laughs)
- DLDouglas Lenat
Yes. Okay. So, so now we have like, um, three, three things to, to talk about.
- LFLex Fridman
We'll keep adding more. (laughs)
- DLDouglas Lenat
Although, although, that's okay. The first and the third are related.
- LFLex Fridman
Yes.
- DLDouglas Lenat
Um, so let's leave the efficiency question aside for now. So, um, how, how does all this information get represented in logical form so that these algorithms, um, resolution theorem proving and other algorithms can actually grind through all the logical consequences of what you said?
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
Um, and that ties into your question about, well, how many of these things do you need? Um, because if the answer is small enough, um, then by hand you could write them out one at a time. So, um, in, um, the, um, 19... early, um, 1984, um, I held a meeting at Stanford where I was a, a faculty member there, um, where we assembled, um, about half a dozen of the smartest people I know, um, people like, um, Allen Newell and Marvin Minsky and Alan Kay and, um, um, a few others.
- LFLex Fridman
Was Feynman there by chance? 'Cause he, he liked your s- he commented about your system Eurisko at the time.
- DLDouglas Lenat
No. No. He wa- he wasn't part of this meeting.
- LFLex Fridman
Okay.
- DLDouglas Lenat
Um, but, um-
- LFLex Fridman
That's a heck of a meeting anyway. (laughs)
- DLDouglas Lenat
Well, I think Ed Feigenbaum was there. I think-
- LFLex Fridman
Yeah.
- DLDouglas Lenat
... um, um, Josh Lederberg was there. So we, we have, um, um, all these different, um, smart people, and we were... um, we came together to a- address the question that, that you raised which is if it's important to represent common sense knowledge and world knowledge in order for AIs to not be brittle, in order for AIs not to just have the veneer of intelligence, well, how many pieces of common sense, how many if/then rules, for instance, would we have to actually write in order to essentially cover what, what people expect perfect strangers to already know about the world?And, um, I expected there would be an enormous divergence of opinion, um, and computation, but amazingly everyone got an answer which was around a million.
- LFLex Fridman
(laughs)
- DLDouglas Lenat
Um, and one person, one person got the answer, um, by saying, "Well, um, look you can only burn into human long-term memory a certain number of things per unit time, like maybe one every 30 seconds or something."
- LFLex Fridman
Yeah.
- DLDouglas Lenat
"And other than that, it's just short-term memory and it flows away like water and so on. So, by the time you're, say, 10 years old or so, um, how many things could you possibly have burned into your long-term memory? And it's like, about a million." Um, another person went in a completely different direction and said, "Well, if you look at the number of words, um, in a, um, a dictionary, not a whole dictionary, but for someone to, um, essentially, um, be considered to be fluent in a language, how many words would they need to know? And then about how many things about each word would you have to tell it?"
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
And so they got to a million that way. Um, another, another person said, "Well, let's actually look at one single, um, short, one-volume desk encyclopedia article."
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
"And so we'll look at, you know, what was, like, a, um, a four-paragraph article or something, I think, about grebes." Grebes are a type of waterfowl. Um, and, "If we were going to sit there and represent every single thing that was there, um, how many assertions, or rules, or statements would we have to write in this logical language and so on? And then multiply that by all of the number of articles that there were and so on." So all of these estimates came out with a million, um, and so, um, if you do the math, it turns out that, like, oh, well then maybe in, um, something like, um, 100, um, person years, um, in one or two person centuries, we could actually get-
- 19:43 – 24:04
How to train an AI knowledge base
- DLDouglas Lenat
well, what are the tens of millions of things that we need to tell the system? And for that, we found, um, a few, um, techniques which worked really well. One is to take, um, any piece of text almost, could be an advertisement, it could be a transcript, it could be a novel, it could be an article, um, and don't pay attention to the actual type that's there. The, the black space on the white page. Pay attention to the compliment of that, the white space if you will. So, what did the writer of this sentence assume that the reader already knew about the world? For instance, if they used a pronoun, how did they figure out that y- why did they think that you would be able to understand what the intended referent of that pronoun was? If they used an ambiguous word, how did they think that you would be able to figure out what they meant by that word? Um, the other thing we look at is the gap between one sentence and the next one. What are all the things that the writer expected you to fill in and infer occurred between the end of one sentence and the beginning of the other? So like, if the sentence says, um, uh, "Fred Smith robbed the Third National Bank," period, um, uh, "He was sentenced to 20 years in prison." Period. Well, between the first sentence and the second, you're expected to infer things like Fred got caught, Fred, um, got arrested, Fred went to jail, Fred had, um, a trial, Fred was found guilty, um, and so on. If my next sentence starts out with something like, "The judge..." um, then you assume it's the judge at his trial. If my next sentence starts out something like, "The arresting officer...", you assume that it was the police officer who arrested him after he committed the crime and so on. So, um, that's a, those are two techniques for getting, um, that knowledge. The, the other thing we sometimes look at is, uh, sort of like fake news or, uh, sort of humorous, um, Onion headlines or, um, headlines in, um, the Weekly World News, if you know what that is, or the National Enquirer where it's like, "Oh, we don't believe this." Then we introspect on why don't we believe it. So, there are things like, um, uh, "B-17 lands on the moon." You know, it's like why don't we, what do we know about the world that causes us to believe that that's just silly-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... or, uh, something like that? Or, um, another thing we look for are contradictions, um, where, which, things which can't both be true, um, and we say to, like, "What is it that we know that causes us to know that both of these can't be true at the same, at the same time?" For instance, in one of the Weekly World News, um, um, editions, in one article it talked about how Elvis was sighted, um, you know, even though he was, uh, you know, getting on in years and so on. And another article in the same one talked about people seeing Elvis' ghost, okay? So it's like why, why do we believe that at least one of these articles, you know, must be wrong, um, and so on. So, um, so we have a series of techniques like that that enable our people, um, and by now, uh, we have about 50 people working full-time on this and have for, for decades. So, we've put in the thousands of person years of effort. We've built up these tens of millions of rules. We constantly police the system to make sure that we're saying things as generally as we possibly can.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
Um, so, um, you don't want to say things like, um, "No mouse is also a moose." Uh, because if you said things like that, um, then you'd have to add another one or two or three zeros onto the number of assertions you'd actually have to have. So, um, at some point, we generalize things more and more and we get to a point where we say, "Oh, yeah, for any two biological taxons, if we don't know explicitly that one is a generalization of another-"
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
"... then almost certainly they're disjoint." Um, a member of one is not going to be a member of the other and so on. So-
- LFLex Fridman
And the same thing with Elvis and the ghost. It, it has nothing to do with Elvis. It's more about h- human nature and the, the, the mortality and-
- DLDouglas Lenat
Well, right.
- LFLex Fridman
... all that kind of stuff.
- DLDouglas Lenat
In general, things are not both alive and dead at the same time.
- LFLex Fridman
Yeah.
- DLDouglas Lenat
And-
- LFLex Fridman
Unless sp- special cats in, in theoretical physics examples.
- DLDouglas Lenat
Well, that raises, um, a couple important points.
- LFLex Fridman
Well, that's the Onion headline situation type of thing. Okay, sorry.
- DLDouglas Lenat
But, no, no, so, so what you bring up is this really important
- 24:04 – 48:25
Global consistency versus local consistency
- DLDouglas Lenat
point of like, well, how do you handle exceptions and inconsistencies and so on? Um, and one of the hardest lessons for us to learn, it took us about five years to, to really grit our teeth and, um, learn to love it-
- LFLex Fridman
(laughs) .
- DLDouglas Lenat
... um, is we had to give up global consistency. So the knowledge base can no longer be...... consistent. So this is a kind of scary thought. I grew up watching Star Trek.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
And any time a computer was inconsistent, it would either freeze up or explode or take over the world, or something bad would happen. Um, or if you come from a mathematics background, uh, once you can prove false, you can prove anything, so that's not good.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
Um, and so on. So, um, that's why, um, the, um, old knowledge-based systems were all very, very consistent, but the trouble is that, um, by and large, our models of the world, the way we talk about the world, and so on, there are all sorts of inconsistencies that creep in here and there, um, that will sort of kill some attempt to build some enormous globally consistent knowledge base. And so, what we had to move to was a system of local consistency. So a good analogy is you know that the surface of the Earth is more or less spherical.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
Globally. Um, but you live your life every day as though the surface of the Earth were flat.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
You know, when you're talking to someone in Australia, you don't think of them as being oriented upside down to you. When you're planning a trip, you know, even if it's 1,000 miles away, um, you may think a little bit about time zones, but you rarely think about the curvature of the Earth and so on. And for most purposes, you can live your whole life without really worrying about that-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... because the Earth is locally flat.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
In much the same way, the Cyc knowledge base is divided up into almost, like, tectonic plates, which are individual contexts, and each context is more or less consistent. But there can be small inconsistencies at the boundary beh- between one context and the next one and so on.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
And so by the time you move, say, 20 contexts over, there could be glaring inconsistencies. So eventually, you get from th- uh, the normal modern real world context that we're in right now to something, you know, like, um, Roadrunner cartoon context where physics is very different and, in fact, life and death are very different because no matter how many times he's killed, you know, the coyote comes back in the next scene and, and so on. So, um, that, that was a hard lesson to learn, and we had to make sure that our representation language, the way we r- the way that we actually encode the knowledge and represent it, was expressive enough that we could talk about things being true in one context and false in another, things that are true at one time and false in another, things that are true, let's say, in one region like one country but false in another, things that are true in one person's belief system but false in another person's belief system, uh, things that are true at one level of abstraction and false at another. For instance, at one level of abstraction, you think of this table as a solid object, but at, you know, down at the atomic level, it's mostly empty space and so on.
- LFLex Fridman
Mm-hmm, mm-hmm. So, then that's fascinating and, but it puts a lot of pressure on context to do a lot of work. So you say tectonic plates. Is it possible to formulate contexts that are general and big that, uh, do this kind of capture of knowledge bases or do you then get turtles on top of turtles again where there's just a huge number of contexts?
- DLDouglas Lenat
So, it's good you asked that question 'cause you're, you're pointed in the right direction, which is you want context to be first class objects in your system's knowledge base, in particular, in Cyc's knowledge base. Um, and by first class object, I mean that it should, we should be able to have Cyc think about and talk about and reason about one context or another context the same way it reasons about coffee cups and tables and people and fishing and so on. Um, and so contexts are just terms in its language, just like the ones I mentioned, and so Cyc can reason about, um, contexts. Contexts can arrange hierarchically and so on. Um, and so you can say things about, let's say, things that are true in the modern era. Things that are true in a particular year would then be a sub-context of the, the things that are true in, um, a bro- let's say, a century or a millennium or something like that. Things that are true in Austin, Texas, um, are generally gonna be a specialization of, um, things that are true in Texas which is gonna be a specialization of things that are true in the United States and so on. Um, and so you don't have to say things over and over, over again, um, at all these levels. You just say things at the most general level that it applies to and you only have to say it once, and then it essentially inherits to all these more specific, um, contexts.
- LFLex Fridman
To ask a slightly technical question, is this inheritin- uh, inheritance a tree or a graph?
- DLDouglas Lenat
Oh, you definitely have to think of it as a graph. Um, so we could talk about, for instance, why the Japanese fifth generation computing effort failed. There were about half a dozen different reasons. One of the reasons they failed was because they tried to represent knowledge as a tree rather than as a graph. Um, and so each node in their representation could only have one parent node. So if you had a table that was a wooden object, a black object, a flat object and so on, you had to choose one and that's the only parent it could have. Uh, when of course, you know, depending on what it is you need to reason about it, sometimes it's important to know that it's made out of wood, like if we're talking about a fire. Sometimes it's important to know that it's flat if we're talking about resting something on it and so on. So, um, uh, one of the, um, one of the problems was that they wanted a kind of Dewey Decimal numbering system for all of their concepts which meant that each node could only have, at most, um, 10 children and each node could only have one parent. Um, and, uh, while that does enable the Dewey Decimal type, um, numbering of concepts, labeling of concepts, um, it prevents you from representing all the things you need to about, um, objects in our, in our world. And that was one of the things which, um, they never were able to overcome and I think that was one of the main reasons that that project failed.
- LFLex Fridman
So w- we'll return to some of the doors you've opened but if we can go back to that room in 1984, around there, with Marvin Minsky in Stanford.
- DLDouglas Lenat
Yes. By, by the way, I should mention-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... that Marvin, um, wouldn't do his estimate until someone brought him an envelope so that he could literally do a back of the envelope calculation-
- LFLex Fridman
(laughs)
- DLDouglas Lenat
... to come up with his number.
- LFLex Fridman
(laughs) Well, because I, I feel like the conversation in that room is an important one. You know? That's, that's how, um ... Sometimes science is done in this way. A few people get together and plant the seed of ideas, and they reverberate throughout history, and some, some kind of, uh, dissipate and disappear, and some, you know, Drake equation, and, you know, they ... You know, it seems like a meaningless equation, somewhat meaningless, but I think it drives and motivates a lot of scientists. And when the aliens finally show up, that equation will get even more, uh, valuable because then we'll get, be able to as- in the long arc of history, the Drake equation will pre-, um, will prove to be quite useful (laughs) , I think. And in that same way, a conversation of just how many facts are required to capture the basic common sense knowledge of the world, that's a fascinating question.
- DLDouglas Lenat
I want to distinguish between what you think of as facts and the kind of things that we represent. So, um, we, we map to and essentially make sure that Cyc has the ability to, as it were, read and access the kind of facts you might find, say, um, in, uh, Wikidata-
- LFLex Fridman
Mm-hmm.
- 48:25 – 54:05
Automated reasoning
- LFLex Fridman
things. Let me jump back to the idea of automated reasoning. So, the acquisition of new knowledge has been done in this very interesting way, um, but primarily by humans doing this, um-
- DLDouglas Lenat
Yes, you can think of, uh, monks in their cells in, uh, medieval Europe, um, you know, carefully illuminating manuscripts-
- LFLex Fridman
Yeah.
- DLDouglas Lenat
... and so on, and-
- LFLex Fridman
It's a very difficult and amazing process actually, because it allows you to truly ask the question about the w- in the white space, what is assumed. I think this exercise is, um... like, very few people do this, right? They just do it subconsciously. They-
- DLDouglas Lenat
Well, well, by, by definition-
- LFLex Fridman
... perform this. By definition.
- DLDouglas Lenat
... right? Because, because those pieces of elided, of omitted information, of those missing steps as it were, um, are pieces of common sense. If you actually included all of them, it would, it would almost be offensive or confusing to the reader. It's like, "Why are they telling me all these things? Of course I know that, you know, all these things." Um, and so, um, uh, so it's-
- LFLex Fridman
Ju-
- DLDouglas Lenat
It's one of these things which almost by its very nature, um, has, has almost never been explicitly written down anywhere, um, because, uh, by the time you're old enough to talk to other people and so on, um, you know, if you survived to that age-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... presumably you already got pieces of common sense, like, um, you know, if something causes you pain whenever you do it, probably not a good idea to keep doing it.
- LFLex Fridman
(laughs)
- DLDouglas Lenat
(laughs) .
- LFLex Fridman
What ideas do you have, given how difficult this stuff is, what ideas are there for how to do it automatically without using humans or at least not, um, you know, doing like a large percentage of the work for humans, and then humans only do the very high-level supervisory work?
- DLDouglas Lenat
So, we have, um, uh, in fact two directions we're pushing on very, very heavily, uh, currently at PsychCorp, and one involves natural language understanding and the ability to read what people have explicitly written down and, and to, to pull knowledge in that way. Um, but the other is to build a series of knowledge editing tools, knowledge entry tools, knowledge, um, capture tools, knowledge, um, testing tools, and so on. Think of them as like user interface, um, suite of software tools if you want, something that will help people to more or less automatically expand and extend the system, um, in areas where, for instance, they want to build some app, have it do some application or something, uh, like that. So, I'll give you an example of one, um, which is something called, um, abduction. So, you've probably heard of like deduction, uh, uh, and, um, induction and so on, but abduction is unlike those. Abduction is not sound, um, it's just useful. (laughs) So, uh, for instance, um, deductively, if someone is out in the rain and i- they're gonna get all wet, and, um, when they enter a room, they might be all wet and so on. So, that's deduction. But if someone were to walk into the room right now and they were dripping wet, um, we would immediately look outside to say, "Oh, did it start to rain?" Or something like that. Now, um, why did we say maybe it started to rain? That's not a sound logical inference, but it's certainly a reasonable, um, abductive, um, leap to say, "Well, one of the most common ways that a person would have gotten dripping wet is if they had gotten caught out in the rain or something like that."
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
Um, so, um, what, what does that have to do with what we were talking about? So, suppose you're building, uh, one of these applications and the system gets some answer wrong, and you say, "Oh, yeah. The answer to this question is, um, this one, not the one you came up with." Then what the system can do is it can use everything it already knows about common sense, general knowledge, the domain you've already been telling it about, um, and context, like we talked about, and so on, and say, "Well, here are, um, seven alternatives, each of which I believe is plausible given everything I already know, and if any of these seven things were true, I would have come up with the answer you just gave me instead of the wrong answer I came up with."
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
"Is one of these seven things true?" And then, you, the expert, will look at those, um, seven things-
- LFLex Fridman
Got it.
- DLDouglas Lenat
... and say, "Oh, yeah. Number five is actually true." And so without actually having to tinker down at the level of logical assertions and so on, um, you'll be able to educate, um, the system the same way that you would help educate another person who you were trying to apprentice or something like that.
- LFLex Fridman
So, the- that significantly reduces the mental effort or significantly increases the efficiency of the teacher, the human teacher.
- DLDouglas Lenat
E- exactly, and m- it makes more or less anyone able to-
- LFLex Fridman
Right.
- DLDouglas Lenat
... to be a teacher, um, in that, um, in that way. So, that's, that's part of the, the answer, and then the other is that, uh, the system on its own will be able to, um, through reading, through, um, conversations with other people and so on, um, learn the same way that, um, you or I or, um, other humans do.
- LFLex Fridman
First of all, that's, that's a beautiful vision. Um, I- I'll have to ask you about semantic web in, in a second here, but first, um, are there, when we talk about specific
- 54:05 – 1:06:43
Direct uses of AI and machine learning
- LFLex Fridman
techniques, do you find something inspiring or directly useful from the whole space of machine learning, deep learning, these kinds of spaces of techniques that have, uh, been shown effective for certain kinds of problems in the recent, um, now decade and a half?
- DLDouglas Lenat
I, I think of the machine learning work, um, as more or less what our right brain hemispheres do.
- LFLex Fridman
(laughs)
- DLDouglas Lenat
So, um, being able to, um, take a bunch of data and recognize patterns, being able to statistically infer things and so on. Um, and, um, you know, I, I certainly wouldn't want to not have a right brain hemisphere.
- LFLex Fridman
Mm.
- DLDouglas Lenat
But I'm also glad that I have a left brain hemisphere as well, something that can metaphorically sit back and puff on its pipe and think about-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... um, this thing over here. It's like, why might this have been true, um, and, um, uh, what are the implications of it, and how should I feel about that and why and so on. So, um, thinking more deeply and slowly, um, um, what Kahneman called thinking slowly versus thinking quickly.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
Whereas you want machine learning to think quickly, but you want the ability to think deeply even if it's a little, um, slower. Um, so I'll give you an example of a project we did recently with, um, NIH involving the Cleveland Clinic and, um, a couple other, um, institutions that we ran a project for. Um, and what it did was it took, um, GWAS, genome-wide association studies, um, uh, those are, uh, sort of big databases of patients that came into a hospital. Um, they got their DNA sequenced because the cost of doing that has gone from, um, infinity to billions of dollars to hundred of dollars or so, um, and so now patients routinely get their DNA sequenced. So, you have these big databases of...... the SNPs, the single nucleotide polymorphisms, the point mutations in a patient's DNA, and the disease that happened to bring them into the hospital. So, now you can do correlation studies, machine learning studies of which mutations, um, uh, are associated with and led to which physiological problems and diseases and so on, like getting arthritis and, um, and so on. And the problem is that those correlations turn out to be very spurious. They turn out to be very noisy. Um, very many of them, um, have led doctors onto wild goose chases and so on, and so they wanted a way of eliminating or, uh, the bad ones or focusing on the good ones. And so-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... uh, this is where Cyc comes in, which is Cyc takes those sort of A to Z correlations between point mutations and, um, the medical condition that needs treatment.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
Um, and we say, okay, let's use all this public knowledge and common sense knowledge, um, about what reactions occur where in the human body, um, what polymerizes what, what catalyzes what reactions and so on, and let's try to put together a 10 or 20 or 30-step causal explanation of why that mutation might have caused that medical condition. And so Cyc would put together, in some sense, some Rube Goldberg-like-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... um, chain that would say, oh yeah, that, um, mutation, um, if it got expressed would be this, um, um, altered protein, which because of that, if it got to this part of the body would catalyze this reaction, and by the way, that would cause more bioactive vitamin D in the person's blood. And anyway, 10 steps later, that screws, screws up bone resorption and that's why this person got osteoporosis early in life and so on. N-
- LFLex Fridman
So, that's human interpretable, or at least doctor human interpretable.
- DLDouglas Lenat
Exa- exactly.
- LFLex Fridman
Yeah.
- DLDouglas Lenat
And, um, the important thing even more than that is, um, uh, you shouldn't really trust that 20-step, um, uh, Rube Goldberg chain any more than you trust that initial A to Z correlation, except two things. One, if you can't even think of one causal chain to explain this, um, then that correlation probably was just noise to begin with. And secondly, and even more powerfully, along the way, that causal chain will make predictions like the one about having more bioactive vitamin D in your blood.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
So, you can now go back to the data about these patients and say, by the way, did they have slightly elevated levels of bioactive vitamin D in their blood and so on? And if the answer is no, that strongly disconfirms your whole causal chain, and if the answer is yes, that somewhat confirms that causal chain. And so using that, we were able to take this, um, these correlations from this GWAS database, and we were able to, um, um, essentially focus the, the doctors', focus the researchers' attention on the very small percentage of correlations that had, um, some explanation, and even better, some explanation that also made some independent prediction that they could confirm or disconfirm by looking at the data. So, think of it like this kind of synergy where you want the right brain machine learning to quickly come up with possible answers. You want the left brain Cyc-like AI to, um, you know, think about that and like, think about why that might have been the case and what else would be the case if that were true and so on, and then suggest things back to the right brain to quickly check out again-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... um, to, um... So it's that kind of synergy back and forth which I think is really what's gonna lead to general AI, not, um, narrow, brittle machine learning systems and not just something like Cyc.
- LFLex Fridman
Okay, so that, that's a brilliant synergy, but I w- I was also thinking in terms of the automated expansion of the knowledge base, you mentioned N- NLU. This is very early days in the machine learning space of this, but self-supervised learning methods, you know, you have these language models, GPT-3 and so on, that just read the internet.
- DLDouglas Lenat
Yes.
- LFLex Fridman
And they form representations that can then be mapped to something useful. The question is, what is the useful thing? Uh, like they're now playing with a pretty cool thing called OpenAI Codex which is generating programs from documentation. Okay-
- DLDouglas Lenat
Yes.
- LFLex Fridman
... that's kind of useful, it's cool, but my question is can it be used to generate, um, in part maybe with some human supervision, uh, Cyc-like assertions, help feed Cyc more assertions from this giant body of internet data?
- DLDouglas Lenat
Yes. That i- that is in fact one of our goals is how can we harness machine learning, how can we harness natural language processing, um, to increasingly automate the knowledge acquisition process, the growth of Cyc? And that's what I meant by priming the pump.
- 1:06:43 – 1:17:16
The semantic web
- DLDouglas Lenat
- LFLex Fridman
There's this idea of the semantic web, and when I first heard about, I just fell in love with the idea. It was the obvious next step for the internet.
- DLDouglas Lenat
Sure.
- LFLex Fridman
And, uh, maybe you can speak about what is the semantic web, what are your thoughts about it, how your vision and mission and goals with Cyc are connected, integrated. Like, are they dance partners? Are they aligned? What are your thoughts there?
- DLDouglas Lenat
So, think of the semantic web as a kind of knowledge graph, and Google already has, uh, something they call knowledge graph, for example, um, which is sort of like a node and link diagram. So you have these, um, nodes that represent concepts or words or terms, um, and then there are some arcs, um, that connect them that might be labeled. Um, and so you might have a node, um, uh, with like one person that represents one person and, um, uh, let's say a, um, a husband link that then points to that person's husband. And so there'd be then another link that went from that person, labeled wife, that went back to the, um, first node and so on. So having, having this kind of representation is really good if you want to represent, um, binary relations, um, um, essentially relations between two things. And if you, so if you have, um, um, the equivalent of like three-word sentences, um, you know, like, uh-... Fred's wife is Wilma, or something like that. You can represent that very nicely using, uh, these kinds of, uh, graph structures or using something like the semantic web and, um, and so on. But the, uh, the problem is that, um, very often what you want to be able to express takes a lot more than three words and a lot more than simple, um, graph structures like that to represent. So, for instance, um, uh, if you've, um, read r- or seen, uh, Romeo and Juliet, you know, I could say to you something like, uh, "Remember when Juliet drank the potion that put her into a kind of suspended animo- animation? When Juliet drank that potion, what did she think that Romeo would think when he heard from someone that she was dead?" Um, and you could basically understand what I'm saying. You could understand the question. You could probably remember the answer was, well, she thought that, um, this, uh, friar would have gotten the message to Romeo saying that she was gonna do this, but the friar didn't, and so... So, um, you're able to represent and reason with these much, much, much more complicated expressions, um, that go way, way beyond what simple, um, three, as it were, three-word or four-word English sentences are, which is really what the semantic web can represent and really what knowledge graphs can represent.
- LFLex Fridman
If you could step back for a second because it's- it's funny you went to- into specifics, and maybe you can elaborate. But I was also referring to semantic web as the vision of converting data on the internet into something that's interpretable, understandable by machines.
- DLDouglas Lenat
Oh, of course, at that- at that level.
- LFLex Fridman
So- so, uh, we should say that what is the semantic web? I mean, you could say a lot of things, but, uh, it- it might not be obvious to a lot of people when they do a Google search that, just like you said, while there might be something that's called a knowledge graph, it's- really boils down to keyword search ranked by the quality estimate of the website, integrating previous human-based Google searches and what they thought was useful. It's like some weird combination of, um, like surface-level hacks that work exceptionally well, but they don't understand the conte- uh, the full contents of the websites that they're searching. So, Google does not u- understand to the degree we've been talking about, the word understand, the contents of the Wikipedia pages as part of the search process. And the semantic web says let's try to get- come up with a way for the computer to be able to truly understand the contents of those pages. That's the dream.
- DLDouglas Lenat
Yes. So, let- let me- let me first give you a- an anecdote, uh, and then I'll answer your question. So, there's a search engine you've probably never heard of called Northern Light, and, um, um, it went out of business. But the way it worked, it was a kind of vampiric search engine and what it did was, um, it didn't index the internet at all. All it did was it, uh, negotiated and got access to data from the big search engine companies about what query was typed in and where the user ended up being happy and actually, um, then, you know, they'd type in a completely different query, unrelated query, and so on. So, it just went from query to the web page that seemed to satisfy them, um, eventually.
- LFLex Fridman
(laughs) Mm-hmm.
- DLDouglas Lenat
Um, and that's all. So, it had actual- no understanding of what was being typed in. It had no statistical data other than what I just mentioned, and it did a fantastic job. It did such a good job that the big search engine companies said, "Oh, we're not gonna sell you this data anymore."
- LFLex Fridman
(laughs)
- DLDouglas Lenat
So, then it went out of business because it had no other way of, um, taking users to where they would want to go and so on. Um-
- LFLex Fridman
And of course, the search engines are now using that kind of idea.
- DLDouglas Lenat
Yes. So, um, let's go back to what you said about the semantic web. So, the dream Tim Berners-Lee and others, um, um, dream about the semantic web at a general level, um, uh, is of course, um, um, exciting and powerful and, in a sense, the right dream to have, which is to, uh, replace the, um, the kind of, um, uh, statistically, um, statistically mapped, um, uh, linkages on the internet, um, into something that's more meaningful and semantic and actually gets at the understanding of the content and so on. Um, and, um, e- eventually, if you say, "Well, how can we do that?" Um, there's, um, uh, sort of a- a low road which is what the knowledge graphs are doing and, um, and so on, which is to say, well, if we just use the simple binary relations, we can actually get some fraction of the way toward understanding, um, and do something where, you know, in the- in the land of the- the blind, the one-eyed man is king-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... uh, kind of thing. And so, being able to even just have a toe in the water in the right direction is fantastically powerful. Um, and so that's where a lot of people stop. Um, but then you could say, well, what if we really wanted to represent, um, and reason with, um, full meaning of what's there? For instance, um, about, um, Romeo and Juliet, um, with reasoning about what Juliet believes that Romeo will believe that Juliet believed, you know, and so on. Or if you look at, um, the news, what, um, you know, President Biden believed that, um, the leaders of the Taliban would believe about, uh, the leaders of Afghanistan if they, you know, blah, blah, blah. So, um...In order to represent, um, complicated, um, sentences like that, um, and let alone reason with them, you need something which is logically, um, much more expressive than these simple, um, triples, than these simple, um, uh, knowledge graph type structures and so on. And that's why, kicking and screaming, we were led from something like, uh, the semantic web representation, which is where we started in, um, 1984, um, with frames and slots, with those kinds of triples, triple store representation. We were led kicking and screaming to this more and more general logical language, this higher order logic. So first, we were led to first order logic, and then second order, and then eventually higher order so you can represent things like modals, like believes, desires, intents, expects and so on, uh, nested ones. You can represent, um, uh, complicated kinds of negation. Um, you can represent, um, the process you're going through in trying to answer the question, so you can say things like, um, "Oh, yeah, if you're trying to do this problem by integration by parts, um, and, um, you recursively get a problem that's solved by integration by parts, that's actually okay. But if that happens a third time, you're probably off on a wild goose chase or something like that."
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
So, being able to talk about the problem-solving process as you're going through the problem-solving process-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... um, is called reflection. And so, um, that's another, so, um-
- LFLex Fridman
It's important to re- be able to represent that as well.
- DLDouglas Lenat
Exactly, you need to be able to represent all of these things, um, because in fact, people do represent them. They do talk about them. They do try and teach them to other people. You do have rules of thumb that key off of them and so on. If you can't represent it, um, then it's sort of like someone with a limited vocabulary who can't understand as easily-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... um, what you're trying to, to tell them. And so, that's, that's really why I think that the, the general dream, the original dream of semantic web is exactly right on, um, but the implementations that we've seen, um, are sort of these toe in a wa- in the water-
- LFLex Fridman
(laughs) Yeah.
- DLDouglas Lenat
... um, little tiny baby steps-
- LFLex Fridman
Yeah.
- DLDouglas Lenat
... in the right direction.
- LFLex Fridman
You should just dive in. (laughs)
- 1:17:16 – 1:26:26
Tools to help Cyc interpret data
- LFLex Fridman
Do you think there's a set of tools that I can help Cyc, uh, interpret the website I create? You know, like, this again, pushing onto the semantic web dream, is there something from the creator perspective that, um, could be done? And one of the things you said, uh, with Cyc or within Cyc that you're doing is the tooling side, making humans more powerful. But is there on the, the other humans in the other side that create the knowledge? Like, for example, you and I are having a two, three, whatever hour conversation now. Is there a way that I could convert this more, make it more accessible to Cyc, to machines? Do you think about that side of it?
- DLDouglas Lenat
I- I'd love to see exactly that kind of semi-automated understanding of what people write and what people say.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
I think of it as a kind of footnoting almost, almost like the way that when you run something in say Microsoft Word or some other document preparation system-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... Google Docs or something, you'll get underlining of questionable things that you might wanna rethink. Either you spelled this wrong or there's a strange grammatical error you might be making here or something. So, I'd like to think in terms of Cyc-powered tools that read through what it is you said or have typed in, uh, and, and try to partially understand-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... what you've said.
- LFLex Fridman
And then you help 'em out.
- DLDouglas Lenat
Exactly. And then they put in little footnotes that will help other readers, and they put in certain footnotes of the form, "I'm not sure what you meant here. You either meant this or this or this, I bet." Uh, if you take a few seconds to disambiguate this for me, then I'll know and I'll have it correct for the next 100 people or the next 100,000 people who come here. Uh, and if it doesn't take too much effort and you want people to understand your web- your website content, not just be able to read it, but actually be able to have systems that reason with it, then yes, it will be worth your small amount of time to go back and make sure that the AI trying to understand it really did correctly understand it.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
Uh, and, you know, let's say you run a, um, a travel web- website or something like that, and people are going to be coming to it because of searches they did, uh, looking for, looking for vacations that or trips that had certain properties and might have been interesting to them for various reasons, thing- things like that. And if you've explained what's going to happen on your trip, then a system will be able to mechanically reason and connect what this person is looking for with what it is you're actually offering. And so if it understands that there's a free day in Geneva...... Switzerland, uh, then if the person coming in happens to, let's say, um, be a nurse or something like that, then even though you didn't mention it, if it can look up the fact that that's where the International Red Cross Museum is and so on, what the means and so on, then it can basically say, "Hey, you might be interested in this trip because while you have a free day in Geneva, you might want to visit that Red Cross Museum."
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
And now, even though it's not very deep reasoning, little tiny factors like that might very well cause you to sign up for that trip rather than some competitor trip.
- LFLex Fridman
Yeah. And so, there's a lot of benefit with SEO, and I actually kind of think... I think there's about a lot of things, which is the actual interface, the design of the interface makes a huge difference. How efficient it is to be productive, and also how, um, full of joy the experience is.
- DLDouglas Lenat
Yes.
- LFLex Fridman
Like I- I- l- I mean, I would love to help a machine, and not from an AI perspective, just as a human. One of the reasons I really enjoy how Tesla, um, have implemented their autopilot system is there's a sense that you're helping this machine learn. And I think humans, I mean, having children, pets-
- DLDouglas Lenat
People love doing that.
- LFLex Fridman
We- we- we- there's joy to teaching.
- DLDouglas Lenat
Absolutely.
- LFLex Fridman
For s- for some people, but I think for a lot of people, and that, if you create the interface where it feels like you're teaching as opposed to like, uh, like annoying, like correcting an annoying system, more like teaching a childlike, innocent, curious system, I think, I think you can literally just like several orders of magnitude scale the amount of good quality data being, uh, added to something like Cyc.
- DLDouglas Lenat
What- what you're suggesting is, uh, much better even than, um, you thought it was.
- LFLex Fridman
(laughs)
- DLDouglas Lenat
Uh, one of the, one of the experiences that we've all had, uh, in our lives is that we thought we understood something, but then we found we really only understood it when we had to teach it or explain it to someone or help our child do homework based on it or something like that. Despite the universality of that kind of experience, if you look at educational software today, almost all of it has the computer playing the role of the teacher, and the student plays the role of the student. But as I just mentioned, you can get a lot of learning to happen better, and as you said, more enjoyably if you are the mentor or the teacher and so on.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
So, we developed a program called Mathcraft to help sixth graders better understand math, and it doesn't actually try to teach you, the player, anything. What it does is it casts you in the role of a student essentially who has classmates who are having trouble, and your job is to watch them as they struggle with some math problem, watch what they're doing and try to give them good advice to get them to understand what they're doing wrong and so on.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
And, uh, the trick from the point of view of Cyc is it has to make mistakes, it has to play the role of the student who makes mistakes, but it has to pick mistakes which are just at the fringe of what you actually understand and don't understand and so on. So, it pulls you into a deeper and deeper level of understanding of the subject.
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
And so, if you give it good advice about what it should have done instead of what it did and so on, then, uh, Cyc knows that you now understand that mistake, you- you won't make that kind of mistake yourself as much anymore. So, Cyc stops making that mistake-
- 1:26:26 – 1:32:25
The most beautiful idea about Cyc
- DLDouglas Lenat
- LFLex Fridman
Sorry for the romantic question, but what is the most beautiful idea you learned about artificial intelligence, knowledge, reasoning from, uh, working on Cyc for 37 years? Or maybe what is the most beautiful idea, surprising idea about Cyc to you?Wha- when I look up at the stars, I kinda want... Like, that- that amazement you feel, that, wow, and you were a part of creating one of the greatest, one of the most fascinating efforts in artificial intelligence history. So, which element brings you, personally, joy?
- DLDouglas Lenat
This may sound contradictory, but I-
- LFLex Fridman
(laughs)
- DLDouglas Lenat
... I think it's the feeling that this will be the only time in history that anyone ever has to teach a computer-
- LFLex Fridman
Mm-hmm.
- DLDouglas Lenat
... this particular thing that we're now teaching it. It's- it's like painting, uh, Starry Night, you- you only have to do that once. Or creating the Pieta, you only have to do that once. You know, it's not, it's not like a, uh, it's not like a singer who has to keep... You know, it's not like Bruce Springsteen having to- to sing his greatest hits over and over again (laughs) at different concerts. It's more like a painter creating a work of art once, and then that's enough, it doesn't have to be created again, and so I really get the sense of we're telling the system things that it's useful for it to know, it's use for a computer to know, for an AI to know, and if we do our jobs right, when we do our jobs right, no one will ever have to do this again for this particular piece of knowledge. It's very, very exciting.
Episode duration: 2:52:55
Install uListen for AI-powered chat & search across the full episode — Get Full Transcript
Transcript of episode 3wMKoSRbGVs
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome