Douglas Lenat: Cyc and the Quest to Solve Common Sense Reasoning in AI | Lex Fridman Podcast #221

Lex Fridman 0:00 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.

Lex Fridman 1:11 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?

Douglas Lenat 1:36 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.

Lex Fridman 2:55 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?

Douglas Lenat 3:10 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.

Lex Fridman 5:36 Mm-hmm.

Douglas Lenat 5:37 Um, over on the, the left was, um, a bunch of fast food restaurant, um, trash bags-

Lex Fridman 5:43 Mm-hmm.

Douglas Lenat 5:43 ... 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.

Lex Fridman 6:08 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.

Douglas Lenat 7:08 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-

Lex Fridman 7:23 Right.

Douglas Lenat 7:23 ... 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?

Lex Fridman 7:49 Right.

Douglas Lenat 7:49 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?

Lex Fridman 8:53 Right.

Douglas Lenat 8:54 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?

Lex Fridman 9:09 Right.

Douglas Lenat 9:10 Um, and those are both really important, um, uh, questions.

Lex Fridman 9:13 And, and like a corollary addition to the first one is

Lex Fridman 9:17 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-

Douglas Lenat 9:53 Well, I, I-

Lex Fridman 9:53 ... how many facts do you need? (laughs)

Douglas Lenat 9:56 I c- (laughs) I, I guess I'm going to disappoint you by giving you an actual answer to your question.

Lex Fridman 10:00 Oh, man. Okay.

Douglas Lenat 10:01 Um-

Lex Fridman 10:01 Well, no. Uh, this sounds exciting. (laughs)

Douglas Lenat 10:03 Yes. Okay. So, so now we have like, um, three, three things to, to talk about.

Lex Fridman 10:09 We'll keep adding more. (laughs)

Douglas Lenat 10:10 Although, although, that's okay. The first and the third are related.

Lex Fridman 10:13 Yes.

Douglas Lenat 10:13 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?

Lex Fridman 10:34 Mm-hmm.

Douglas Lenat 10:34 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.

Lex Fridman 11:15 Was Feynman there by chance? 'Cause he, he liked your s- he commented about your system Eurisko at the time.

Douglas Lenat 11:20 No. No. He wa- he wasn't part of this meeting.

Lex Fridman 11:23 Okay.

Douglas Lenat 11:23 Um, but, um-

Lex Fridman 11:23 That's a heck of a meeting anyway. (laughs)

Douglas Lenat 11:25 Well, I think Ed Feigenbaum was there. I think-

Lex Fridman 11:28 Yeah.

Douglas Lenat 11:28 ... 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.

Lex Fridman 12:20 (laughs)

Douglas Lenat 12:20 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."

Lex Fridman 12:34 Yeah.

Douglas Lenat 12:34 "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?"

Lex Fridman 13:09 Mm-hmm.

Douglas Lenat 13:09 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."

Lex Fridman 13:22 Mm-hmm.

Douglas Lenat 13:23 "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-

Lex Fridman 14:07 Mm.

Douglas Lenat 14:08 ... this written down by hand. And a marvelous, um, uh, coincidence o- opportunity existed, um, right at that point in time, the early 1980s. There was something called the Japanese Fifth Generation Computing Effort. Japan had threatened to do, in computing and AI and hardware, what they had just finished doing in consumer electronics and the automotive industry, namely wresting control away from the United States and, more generally, away from the West. Um, and so America was scared, um, and, um, Congress, uh, did something. That's how you know it was a long time ago-

Lex Fridman 14:45 (laughs)

Douglas Lenat 14:45 ... 'cause Congress did something. Congress p-

Lex Fridman 14:47 (laughs)

Douglas Lenat 14:47 Congress passed something called the National Cooperative Research Act, NCRA, and what it said was, "Hey, all you big American companies..." That's also how you know it was a long time ago, because they were American companies rather than multinational companies. "Hey, all you big American companies, um, normally it would be an antitrust violation if you colluded on R&D, but we promise for the next 10 years, uh, we won't prosecute any of you if you do that, um, to help combat this threat." And so overnight, um, the first two consortia, research consortia, in America sprang up, um, both of them coincidentally in Austin, Texas. Uh, one called Sematech, focusing on hardware chips and so on, and then one called MCC, the Microelectronics and Computer Technology Corporation, focusing on, more on software, on databases and AI and natural language understanding and things like that. Um, and, um, I got the opportunity, thanks to my friend Woody Bledsoe, um, who, uh, was one of the people, um, who founded that, to come and be its principal scientist, and he said, you know, um, and he sent, uh, Admiral Bob Inman, who was the person running MCC, uh, came and talked to me and said, "Look, Professor, you know, you're talking about doing this project. It's gonna involve, um, person centuries of effort. Uh, you've only got a handful of graduate students. You do the math, it's gonna take you, like, um, you know, uh, longer than the rest of your life to finish this project. But if you move to the wilds of Austin, Texas, uh, we'll put ten times as many people on it and, um, you know, you'll be done in, um, a few years." And so that was pretty exciting, and so, um, I did that. I, um, uh, took my leave from Stanford. I came to, to Austin. I worked for MCC, and, um, good news and bad news. The bad news is that all of us were off by an order of magnitude.

Lex Fridman 16:42 Mm-hmm.

Douglas Lenat 16:42 Um, that it turns out what you need are tens of millions of these, um, pieces of knowledge about, um, everyday sort of like, um, if you have a coffee cup with stuff in it and you turn it upside down, the stuff in it's gonna fall out.

Lex Fridman 16:56 Mm-hmm.

Douglas Lenat 16:56 Um, so you need tens of millions of pieces of knowledge like that, even if you take trouble to make each one as general as it possibly could be.

Lex Fridman 17:05 Mm-hmm.

Douglas Lenat 17:05 Um, but, um, the good news was that thanks to, um, uh, initially the Fifth Generation effort and then later, um, US government, um, agency funding and so on, we were able to get enough funding not for, um, a couple person centuries of time, but for a couple person millennia of time-

Lex Fridman 17:26 Mm.

Douglas Lenat 17:26 ... which is what we've spent since 1984 getting CYC to contain the tens of millions of rules that it needs in order to really capture, uh, and span, uh, sort of not all of human knowledge, but the things that you assume other people kn- the things you count on other people knowing. Um, and, uh, so by now we've done that, um, and, uh, the good news is since you've waited 38 years just about to talk to me-

Lex Fridman 17:55 (laughs)

Douglas Lenat 17:56 ... uh, we're, uh, we're about at the, um, at the end of that process. So most of what we're doing now is not putting in even what you would consider common sense, but more putting in, um, domain-specific, application-specific knowledge about, um, uh, healthcare, um, in a certain hospital or about, um, um-... oil, um, pipes getting, um, clogged up or whatever the applications happen to be. So, we've almost come full circle and we're doing things very much like the expert systems-

Lex Fridman 18:26 Mm-hmm.

Douglas Lenat 18:27 ... of the 1970s and the 1980s, except instead of resting on nothing-

Lex Fridman 18:32 Mm-hmm.

Douglas Lenat 18:32 ... and being brittle, they're now resting on this massive pyramid if you will, this massive lattice of commonsense knowledge so that when things go wrong, when something unexpected happens, they can fall back on more and more and more general principles, um, eventually bottoming out in things like, um, for instance, if we have a problem with the microphone, one of the things you'll do is unplug it, plug it in again, and hope for the best, right?

Lex Fridman 18:58 Mm-hmm.

Douglas Lenat 18:58 Because that's one of the general pieces of knowledge you have in dealing with electronic equipment or software systems, or, um, things like that.

Lex Fridman 19:05 Is there a basic principle like that? Like, is there, is it possible to encode something that generally captures this idea of turn it off and turn it back on and see if it fixes?

Douglas Lenat 19:15 Oh, absolutely. That's one of the things that, um, that Cyc knows and-

Lex Fridman 19:19 It's actually one of the fundamental, um, laws of nature-

Douglas Lenat 19:23 (laughs) .

Lex Fridman 19:23 ... I believe.

Douglas Lenat 19:25 I wouldn't, I wouldn't call it a law. It's, it's more like a, um...

Lex Fridman 19:29 Seems to work every time so it sure, sure looks like a law. I don't know.

Douglas Lenat 19:34 So that, um, that basically, um, um, covered the, um, the resources needed.

Lex Fridman 19:39 Yeah.

Douglas Lenat 19:39 And then we had to devise a method to actually figure out,

Douglas Lenat 19:43 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-

Lex Fridman 22:01 Mm-hmm.

Douglas Lenat 22:01 ... 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.

Lex Fridman 23:04 Mm-hmm.

Douglas Lenat 23:04 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-"

Lex Fridman 23:32 Mm-hmm.

Douglas Lenat 23:32 "... 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-

Lex Fridman 23:38 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-

Douglas Lenat 23:46 Well, right.

Lex Fridman 23:46 ... all that kind of stuff.

Douglas Lenat 23:47 In general, things are not both alive and dead at the same time.

Lex Fridman 23:50 Yeah.

Douglas Lenat 23:50 And-

Lex Fridman 23:51 Unless sp- special cats in, in theoretical physics examples.

Douglas Lenat 23:55 Well, that raises, um, a couple important points.

Lex Fridman 23:58 Well, that's the Onion headline situation type of thing. Okay, sorry.

Douglas Lenat 24:01 But, no, no, so, so what you bring up is this really important

Douglas Lenat 24:04 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-

Lex Fridman 24:20 (laughs) .

Douglas Lenat 24:20 ... 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.

Lex Fridman 24:30 Mm-hmm.

Douglas Lenat 24:30 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.

Lex Fridman 24:45 Mm-hmm.

Douglas Lenat 24:45 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.

Lex Fridman 25:19 Mm-hmm.

Douglas Lenat 25:20 Globally. Um, but you live your life every day as though the surface of the Earth were flat.

Lex Fridman 25:27 Mm-hmm.

Douglas Lenat 25:27 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-

Lex Fridman 25:46 Mm-hmm.

Douglas Lenat 25:46 ... because the Earth is locally flat.

Lex Fridman 25:49 Mm-hmm.

Douglas Lenat 25:49 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.

Lex Fridman 26:08 Mm-hmm.

Douglas Lenat 26:08 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.

Lex Fridman 27:16 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?

Douglas Lenat 27:39 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.

Lex Fridman 29:09 To ask a slightly technical question, is this inheritin- uh, inheritance a tree or a graph?

Douglas Lenat 29:15 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.

Lex Fridman 30:45 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.

Douglas Lenat 30:53 Yes. By, by the way, I should mention-

Lex Fridman 30:54 Mm-hmm.

Douglas Lenat 30:55 ... 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-

Lex Fridman 31:04 (laughs)

Douglas Lenat 31:04 ... to come up with his number.

Lex Fridman 31:05 (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.

Douglas Lenat 31:58 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-

Lex Fridman 32:16 Mm-hmm.

Douglas Lenat 32:16 ... or stated in a Wikipedia article or something like that. So, what we're representing and the things that we need a small number of tens of millions of are more like rules of thumb, rules of good guessing, things which are usually true and which help you to make sense of the facts that are, um, sort of sitting off in some database or some other, uh, more static storage.

Lex Fridman 32:39 So, they're, they're, they're almost like platonic forms. So, like, when you read stuff on Wikipedia, that's going to be like projections of those ideas. You read an article about the fact that Elvis died, that's a projection of the idea that, uh, humans are mortal and, like, you know ... Uh, very few Wikipedia articles will write, "Humans are immortal." (laughs)

Douglas Lenat 33:00 E- exactly. That's what I meant about ferreting out the unstated things in texts.

Lex Fridman 33:05 Yes.

Douglas Lenat 33:05 What are all the things that were assumed? And so those are things like, um, if you have a problem with something, turning it off and on, um, often fixes it for reasons we don't really understand and we're not happy about, or, um, people can't be both alive and dead at the same time and ... Or water flows, uh, downhill.

Lex Fridman 33:23 Mm.

Douglas Lenat 33:23 If you search online for water flowing uphill and water flowing downhill, you'll find more references for water flowing uphill because it's used as a kind of a, um, a metaphorical-

Lex Fridman 33:33 Mm-hmm.

Douglas Lenat 33:33 ... reference for some unlikely thing because, of course, everyone already knows that water flows downhill.

Lex Fridman 33:39 Exactly.

Douglas Lenat 33:39 So, why would anyone bother saying that?

Lex Fridman 33:41 Do you have a word you prefer? Because we said fact isn't the right word. Is there a word like concepts or-

Douglas Lenat 33:48 I, I would say assertions.

Lex Fridman 33:49 Assertions.

Douglas Lenat 33:49 U- uh, assertions or rules, because I'm not talking about rigid rules, but rules of thumb. But assertions, um, is a nice one that covers, um, all of these things.

Lex Fridman 33:59 Yeah. As a programmer, to me, assert has a very, uh, dogmatic, authoritarian feel to them, like-

Douglas Lenat 34:06 Oh, I'm sorry. (laughs)

Lex Fridman 34:08 (laughs) I'm so sorry. Okay, but as- assertions works. Okay. So, if we go back to that room with Marvin Minsky, with you, the, uh, all these seminal figures, uh, uh, Ed Feigennbaum, um, uh, thinking about the, this very philosophical but also engineering question. We can also go back a couple of decades before then in thinking about artificial intelligence broadly when people were thinking about, you know, how do you create superintelligent systems, general intelligence, and I think people's intuition was off at the time, and I mean, this continues to be the case that we're not ... When we're grappling with these, these exceptionally difficult ideas, we're not always ... It's very difficult to truly understand ourselves when we're, when we're thinking about the human mind to, to introspect how difficult it is to engineer intelligence, to solve intelligence. We're not very good at estimating that. And you are somebody who has really stayed with this question for decades. Do you ... What's your sense from the 1984 to today, have you gotten a stronger sense of just how much knowledge is required? So, you've kind of said with some level of certainty that it's still on the order of magnitude of tens of millions.

Douglas Lenat 35:30 Right.

Lex Fridman 35:30 But-

Douglas Lenat 35:30 For, for the first several years, I would have said that it was on the order of one or two million.

Lex Fridman 35:36 Yeah.

Douglas Lenat 35:36 And so, um, it took, it took us about five or six years to realize that we were off, um, by a, by a factor of 10.

Lex Fridman 35:44 But I guess what I'm asking, you know, Marvin Minsky was very confident in the '60s when he was saying- (laughs)

Douglas Lenat 35:49 Yes.

Lex Fridman 35:49 Right? What's your sense if you, you know, 200 years from now, you're still, you know, you're, you're, you're not going to be any longer in this in particular biological body, but your brain will still be, uh, (laughs) in a digital form and you'll be looking back, would you think you were smart today, like your intuition was right or do you think you may be really off?

Douglas Lenat 36:19 So, I, I think I'm, I'm right enough, and let me explain what I mean by that, which is, um, sometimes, like, if you have an old-fashioned pump, you have to prime the pump-

Lex Fridman 36:30 (laughs) Yeah, right.

Douglas Lenat 36:31 ... and then eventually it starts. So, I think I'm, I'm right enough in the sense that-

Lex Fridman 36:36 To prime the pump. (laughs)

Douglas Lenat 36:37 ... what we've, what we've built-

Lex Fridman 36:38 Yeah.

Douglas Lenat 36:38 ... even if it isn't, so to speak, everything you need, um, it's primed the knowledge pump enough that Cyc can now itself help to learn more and more automatically on its own by reading things and understanding and occasionally asking questions like a, a student would or something, and by doing experiments and discovering things on its own and so on. So, through a combination of, um, Cyc-powered discovery and Cyc-powered reading, um, it will be able to bootstrap itself. Maybe it's the final 2%. Maybe it's the final 99%. So, even if I'm, uh, if I'm wrong, um, all I really need to, to build is a system which has primed the pump enough that it can begin.... um, that cascade upward, that self-reinforcing, uh, sort of quadratically or maybe even exponentially increasing, um, um, path upward, um, that we get from, for instance, talking with each other. That's why, um, um, humans today know so much more than humans 100,000 years ago. We're not really that much smarter (laughs) than people were 100,000 years ago, but there's so much more knowledge, and we have language, and we can communicate, um, we can check things on Google, and so on. So effectively, we have this enormous power at our fingertips, um, and there's almost no limit to how much you could learn if you wanted to because you've already gotten to a certain level of understanding of the world that enables you to read all these articles and understand them, that enables you to go out and, if necessary, do experiments, although that's slower, as a way of gathering data, um, and so on. And, and I think this is really, um, an important point which is, if we have artificial intelligence, real general artificial intelligence, human-level artificial intelligence, then people will become smarter. Um, it's not so much that it'll be us versus the AIs. It's more like us and the AIs together-

Lex Fridman 38:39 Mm-hmm.

Douglas Lenat 38:40 ... will be able to do things that require more creativity, um, that would take too long right now, but we'll be able to do lots of things in parallel. Uh, we'll be able to misunderstand each other less.

Lex Fridman 38:50 Mm-hmm.

Douglas Lenat 38:51 Um, uh, there's all sorts of, um, value that effectively, for an individual, would mean that individual will, for all intents and purposes, be smarter.

Lex Fridman 39:01 Mm-hmm.

Douglas Lenat 39:01 And that means that humanity, as a species, will be smarter. And when was the last time that any invention qualitatively im- made a huge difference in human intelligence? Um, you have to go back a long ways. It wasn't like the internet, or the computer, or s- mathematics or something. It was all the way back to the development of language. We sort of look back on pre-linguistic cavemen as, well, you know, they, they weren't really intelligent, were they? (laughs) They weren't really human, were they? And I think that, um, as you said, 50, 100, 200 years from now, um, people will look back on people today, um, right before the advent of these sort of lifelong general AI uses-

Lex Fridman 39:50 Mm-hmm.

Douglas Lenat 39:51 ... um, and say, "You know, those poor, those poor people, they weren't really human, were they?"

Lex Fridman 39:55 Mm-hmm. They, exactly. So, you said a lot of really interesting things. By the way, I would maybe, uh, try to argue that the internet is on, is on the order of, um, the kind of big leap in improvement that n- the invention of language was 'cause there's-

Douglas Lenat 40:14 Well, it's certainly a big leap in one direction. We're not sure whether it's upward or downward.

Lex Fridman 40:18 Well, I, I mean s- very specific parts of the internet, which is access to information like a website like Wikipedia, like, ability for human beings from across the world to access information s- very quickly.

Douglas Lenat 40:29 So, I, I could take either side of this argument, and since you just took one side-

Lex Fridman 40:33 Okay, great.

Douglas Lenat 40:34 ... I'll give you the other side which is that, um, almost nothing has done more harm, um, than, uh, something like the internet and access to that information, um, in two ways. One is it's made people more, um, uh, globally ignorant-

Lex Fridman 40:50 Mm-hmm.

Douglas Lenat 40:51 ... um, in the same way that calculators made us, um, more or less enumerate. So, when I was growing up, we had to use slide rules, we had to be able to estimate-

Lex Fridman 41:01 Yeah.

Douglas Lenat 41:02 ... um, and so on. Today, um, people don't really understand numbers. They don't really understand math. They don't really estimate very well at all, um, and so on. Uh, they don't really understand the difference between trillions and billions and millions and so on very well, um, uh, because calculators do that all for us. Um, and, um, thanks to, uh, things like the internet and search engines, um, uh, that same kind of juvenileism, um, is reinforced and making people, um, essentially be able to live their whole lives not just without being able to do arithmetic and estimate, but now without actually having to really know almost anything-

Lex Fridman 41:41 Right.

Douglas Lenat 41:41 ... because any time they need to know something, they'll just go and look it up.

Lex Fridman 41:45 You're right.

Douglas Lenat 41:45 Um, and so-

Lex Fridman 41:45 And I could tell you could play both sides of this, and it is a double-edged sword. You can, of course, say the same thing about language. Probably people, when they invented language, they were criticized. You know, it used to be we would just, if we were angry, we would just kill a person, and if we were in love, we would just have sex with them.

Douglas Lenat 42:00 (laughs)

Lex Fridman 42:00 And now everybody's writing poetry and bullshit.

Douglas Lenat 42:03 (laughs)

Lex Fridman 42:03 They, you know, you k- should just be direct. You should, like, have physical contact. Enough of this words and books, and it's, you're, you're not actually experiencing... Like, if you read a book, you're not experiencing the thing. This is nonsense.

Douglas Lenat 42:16 That, that, that's right.

Lex Fridman 42:16 (laughs)

Douglas Lenat 42:16 If you read a book about how to make butter, that's not the same as-

Lex Fridman 42:19 Yeah.

Douglas Lenat 42:19 ... if you had to, like, learn it and do it yourself-

Lex Fridman 42:21 Exactly, do it yourself.

Douglas Lenat 42:22 ... and, and so on. So, so-

Lex Fridman 42:23 (laughs) Right.

Douglas Lenat 42:23 ... let's just say that something is gained but something is lost every time you have, um, these, these sorts of dependencies, um, on technology. Um, and overall, I think that the, um, having smarter individuals and having smarter AI-augmented, um, human species, um, will be, um, one of the few ways that we'll actually be able to overcome some of the global problems we have involving, um, poverty and starvation, and, uh, global warming and, um, overcrowding, all the other, uh, problems that, um, that are besetting, um, the planet. We really need to be smarter, and there're really only two routes to being smarter. One is through, uh, biochemistry and, um, uh, genetics, um, genetic engineering. The other route is through having general AIs that, um, augment our intelligence, um, and, um, you know, hopefully one of those two, uh, ways of, uh, paths to salvation will, will come through before it's too late.

Lex Fridman 43:29 Yeah, absolutely. I agree with you, and obviously as an engineer, I have, um...... I have a better sense and an optimism about the technology side of things because you can control things there more. Biology is just such a giant mess. We're living through a pandemic now. There's so many ways that nature can just be just destructive and destructive in a way where it doesn't even notice you. You're n- it's not like a battle of humans versus virus. It's just like, "Huh, okay," and then you could just wipe out (laughs) an entire species.

Douglas Lenat 43:59 The, the other problem with, um, the internet is that it has enabled us to surround ourselves with an echo chamber, with a, a bubble of, um, like-minded people, which means that you can have, uh, truly bizarre, um, theories, conspiracy theories, fake news, and so on, um, promulgate, um, and surround yourself with people who essentially reinforce, um, what you want to believe or what you already believe about the world. Um, and in the, in the old days, that was much harder to do when you had, say, only three TV networks, or even before when you had no TV networks and you had to actually, like, look at the world and make your own, um, reasoned decisions.

Lex Fridman 44:44 I like the push and pull of our dance that we're doing because then I'll just say in the old world, having come from the Soviet Union, because you had one or a couple of networks, then propaganda could be much more effective, and then the government can overpower its people by telling you the truth, and then, uh, starving millions and, uh, torturing millions, and putting millions into camps, and starting wars with a propaganda machine allowing you to believe that you're actually good- doing good in the world. With the internet, because of all the, quote-unquote, "conspiracy theories", some of them are actually challenging the power centers, the very kind of power centers that a century ago would have led to the death of millions. So, there's a... it's again this double-edged sword. And I, I very much agree with you on the AI side. It's, it's often an intuition that people have that somehow AI will be used to, uh, maybe overpower people by certain select groups, and to me it's not at all obvious that that's the likely scenario. To me the likely scenario, uh, especially just having observed the trajectory of technology, is it'll be used to empower people. It'll be used to extend the capabilities of, uh, individuals across the world, because there's a lot of money to be made that way. (laughs) Like, improving people's lives you can make a lot of money.

Douglas Lenat 46:03 I agree. I think that the, the main, the main thing that AI prostheses, AI amplifiers will do for people is make it easier, maybe even unavoidable, for them to do good critical thinking. Um, so pointing out logical fallacies, logical contradictions, and so on in, um, things that they otherwise would just blithely believe. Um, pointing out, um, uh, essentially data which, um, they should take into consideration, um, if they really want to, um, um, learn the truth about something and so on. So, I think, um, doing not just educating in the sense of, um, pouring facts into people's heads, but educating in the sense of arming people with the ability to do good critical thinking-

Lex Fridman 46:55 Mm-hmm.

Douglas Lenat 46:56 ... um, is, um, enormously powerful. Um, the education system that we have in the US and worldwide generally don't do a good job of that, um, but, um, I believe that the AI-

Lex Fridman 47:09 The AIs will. (laughs)

Douglas Lenat 47:11 The AI, the AIs will.

Lex Fridman 47:11 Yeah.

Douglas Lenat 47:11 The AIs can and will. In the same way that everyone can have their own, um, um, Alexa or Siri or, um, Google Assistant or whatever, um, um, everyone will have this sort of cradle to grave, um, assistant, um, which will get to know you, which you'll get to trust. It'll model you, you'll model it, um, and, um, it'll call to your attention things which will, in some sense, make your life better, easier, um, less, um, mistake-ridden, and so on, less regret-ridden, um, if you listen to, listen to it.

Lex Fridman 47:46 Yeah, I'm, uh, in full agreement with you about th- this, like, space of technology and I think it's super exciting. And from my perspective, integrating emotional intelligence, so even things like friendship and companionship and love into those kinds of systems, uh, as opposed to helping you just grow intellectually as a human being, allow you to grow emotionally, which is ultimately what makes life amazing, is to, to sort of, you know, the, the old pursuit of happiness. So, it's not just the pursuit of reason, it's the pursuit of happiness too.

Douglas Lenat 48:19 Yes.

Lex Fridman 48:20 The f- the full spectrum. Well, l- let me, um, sort of... 'cause you mentioned if... so many fascinating

Lex Fridman 48:25 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-

Douglas Lenat 48:39 Yes, you can think of, uh, monks in their cells in, uh, medieval Europe, um, you know, carefully illuminating manuscripts-

Lex Fridman 48:47 Yeah.

Douglas Lenat 48:47 ... and so on, and-

Lex Fridman 48:47 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-

Douglas Lenat 49:07 Well, well, by, by definition-

Lex Fridman 49:07 ... perform this. By definition.

Douglas Lenat 49:08 ... 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-

Lex Fridman 49:31 Ju-

Douglas Lenat 49:31 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-

Lex Fridman 49:47 Mm-hmm.

Douglas Lenat 49:47 ... 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.

Lex Fridman 49:56 (laughs)

Douglas Lenat 49:57 (laughs) .

Lex Fridman 49:58 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?

Douglas Lenat 50:18 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."

Lex Fridman 52:12 Mm-hmm.

Douglas Lenat 52:12 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."

Lex Fridman 52:56 Mm-hmm.

Douglas Lenat 52:57 "Is one of these seven things true?" And then, you, the expert, will look at those, um, seven things-

Lex Fridman 53:02 Got it.

Douglas Lenat 53:02 ... 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.

Lex Fridman 53:22 So, the- that significantly reduces the mental effort or significantly increases the efficiency of the teacher, the human teacher.

Douglas Lenat 53:30 E- exactly, and m- it makes more or less anyone able to-

Lex Fridman 53:33 Right.

Douglas Lenat 53:33 ... 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.

Lex Fridman 53:54 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

Lex Fridman 54:05 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?

Douglas Lenat 54:21 I, I think of the machine learning work, um, as more or less what our right brain hemispheres do.

Lex Fridman 54:29 (laughs)

Douglas Lenat 54:29 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.

Lex Fridman 54:47 Mm.

Douglas Lenat 54:47 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-

Lex Fridman 54:56 Mm-hmm.

Douglas Lenat 54:56 ... 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.

Lex Fridman 55:12 Mm-hmm.

Douglas Lenat 55:12 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-

Lex Fridman 56:45 Mm-hmm.

Douglas Lenat 56:45 ... 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.

Lex Fridman 56:56 Mm-hmm.

Douglas Lenat 56:56 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-

Lex Fridman 57:26 Mm-hmm.

Douglas Lenat 57:26 ... 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-

Lex Fridman 57:53 So, that's human interpretable, or at least doctor human interpretable.

Douglas Lenat 57:55 Exa- exactly.

Lex Fridman 57:56 Yeah.

Douglas Lenat 57:56 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.

Lex Fridman 58:31 Mm-hmm.

Douglas Lenat 58:32 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-

Lex Fridman 59:41 Mm-hmm.

Douglas Lenat 59:43 ... 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.

Lex Fridman 59:55 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.

Douglas Lenat 1:00:15 Yes.

Lex Fridman 1:00:16 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-

Douglas Lenat 1:00:30 Yes.

Lex Fridman 1:00:30 ... 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?

Douglas Lenat 1:00:48 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.

Lex Fridman 1:01:04 Mm.

Douglas Lenat 1:01:04 That, um, you know, if you, you sort of learn things at the fringe of what you know already.

Lex Fridman 1:01:11 Right.

Douglas Lenat 1:01:11 Uh, you learn this new thing is similar to what you know already and here are the differences and the new things you had to learn about it and so on. So, the more you know, the more and more easily you can learn new things. But unfortunately, inversely, if you don't really know anything (laughs) , then it's really hard to learn anything. (laughs) And so, um, if you're not careful, if you start out with too small, um, uh, sort of a core to start this process, um, it never really takes off. And so that's why I view this as a pump priming exercise to get a big enough manually produced, even though that's kind of ugly duckling technique-

Lex Fridman 1:01:47 Mm-hmm.

Douglas Lenat 1:01:47 ... put in the elbow grease to produce a large enough core, um, that you will be able to do all the kinds of things you're imagining, um, without, uh, without sort of, um, ending up with the kind of, um, wacky brittlenesses that we see, for example, in, um, GPT-3, um, where, um, it, uh, uh, you know, you'll tell it a story about, um, um, you know, someone, uh, uh, putting poison, um-... you know, plotting to poison someone and so on. And then the, um, you know, then, you know, GPT-3 says, um, what's, you say, "What's the very next sentence?" And the next sentence is, oh, yeah, that person then drank the poison they just put together. It's like, that's probably not what happened-

Lex Fridman 1:02:28 Mm-hmm. (laughs)

Douglas Lenat 1:02:29 ... for someone or, um, if you go to Siri and, um, um, you know, I think I have, uh, where, where can I go for, um, help with my, um, um, alcohol problem or something? It'll come back and say, "I found seven liquor stores near you."

Lex Fridman 1:02:44 Right.

Douglas Lenat 1:02:44 You know? (laughs)

Lex Fridman 1:02:45 (laughs)

Douglas Lenat 1:02:45 And you know, so on. So, you know, it's one of these things where, um, yes, it may be helpful, um, most of the time, it may even be correct-

Lex Fridman 1:02:55 Yeah.

Douglas Lenat 1:02:55 ... most of the time, but if it doesn't really understand what it's saying and if it doesn't really understand why things are true and doesn't really understand how the world, world works, then some fraction of the time it's gonna be wrong.

Lex Fridman 1:03:07 Mm-hmm.

Douglas Lenat 1:03:07 Now, if your only goal is to sort of find relevant information like search engines do, um, then being right 90% of the time is fantastic. That's unbelievably great.

Lex Fridman 1:03:19 (laughs)

Douglas Lenat 1:03:19 Okay. However, if your goal is to like, um, you know, save the, the life of your child who has some medical problem or your goal is to, uh, be able to drive, you know, for the next 10,000 hours of driving without getting into a fatal accident and so on, then, you know, um, error rates down at the 10% level or even the 1% level are not really acceptable.

Lex Fridman 1:03:41 Mm-hmm. I like the model of with that learning happens at the edge and then you kind of think of knowledge as this sphere. So, uh, if you want a large sphere because the, uh, the learning is happening on the surface.

Douglas Lenat 1:03:56 E- e- exactly. So you have the, the, what you can learn next increases quadratically as the diameter of that sphere, um, goes up.

Lex Fridman 1:04:05 But it, it's nice because you think when you know nothing, it's like you can learn anything, but the reality not really. (laughs)

Douglas Lenat 1:04:12 Right. If you know, if you know nothing, you can really learn nothing.

Lex Fridman 1:04:15 (laughs) Exactly.

Douglas Lenat 1:04:15 You can appear to learn. So I, I'll, I'll also, um, one, one of the, um, anecdotes, uh, I could go back and, um, give you about why, uh, why I feel so strongly about this personally.

Lex Fridman 1:04:27 Mm-hmm.

Douglas Lenat 1:04:28 Um, was, um, in, um, 1980, '81, um, my daughter Nicole was born, and she's actually doing fine now, but when she was a baby, um, she was diagnosed as having meningitis and, um, doctors wanted to do all these scary things. Um, and, uh, my wife and I were very, um, worried and we could not get a meaningful answer from her doctors about exactly why they believed this, what the alternatives were and so on. And fortunately, a friend of mine, Ted Shortliffe, was another, um, assistant professor in computer science, um, at Stanford at the time, and he'd been building a program called MYCIN, which was a medical diagnosis program that happened to specialize in, um, uh, blood, um, infections like meningitis. And so he had privileges at Stanford Hospital because he was also an MD. Um, and so, uh, we got hold of her chart, and we put in her case, and it came up with exactly the same diagnoses and exactly the same therapy recommendations. But the difference was, because it was a knowledge-based system, a rule-based system, it was able to tell us step by step by step, um, why, um, this was the diagnosis and step by step why this was the best, um, um, therapy, the best, um, procedure to, um, um, to, to do for her and so on. And there was a real epiphany because that made all the difference in the world. Instead of blindly having to trust an authority, we were able to understand what was actually going on. And, um, so at that, at that time, I realized that that really is what was missing in computer programs was that even if they got things right, because they didn't really understand, um, the way the world works and why things are the way they are, uh, they weren't able to give explanations of their answer. Um, you know, and, you know, it's one thing to, to use a machine learning system that says, "This is what you should..." you know, "I, I think you should get this operation." And you say, "Why?" And it says, you know, "0.83." And you say, "No, in more detail, why?" And it says, "0.831."

Lex Fridman 1:06:38 (laughs)

Douglas Lenat 1:06:38 You know? (laughs) Okay. That's not really very compelling-

Lex Fridman 1:06:41 Yeah.

Douglas Lenat 1:06:41 ... and that's not really very helpful.

Lex Fridman 1:06:43 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.

Douglas Lenat 1:06:52 Sure.

Lex Fridman 1:06:53 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?

Douglas Lenat 1:07:08 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.

Lex Fridman 1:09:49 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.

Douglas Lenat 1:10:07 Oh, of course, at that- at that level.

Lex Fridman 1:10:09 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.

Douglas Lenat 1:11:17 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.

Lex Fridman 1:12:11 (laughs) Mm-hmm.

Douglas Lenat 1:12:11 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."

Lex Fridman 1:12:27 (laughs)

Douglas Lenat 1:12:27 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-

Lex Fridman 1:12:33 And of course, the search engines are now using that kind of idea.

Douglas Lenat 1:12:37 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-

Lex Fridman 1:13:47 Mm-hmm.

Douglas Lenat 1:13:48 ... 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."

Lex Fridman 1:15:56 Mm-hmm.

Douglas Lenat 1:15:56 So, being able to talk about the problem-solving process as you're going through the problem-solving process-

Lex Fridman 1:16:01 Mm-hmm.

Douglas Lenat 1:16:02 ... um, is called reflection. And so, um, that's another, so, um-

Lex Fridman 1:16:05 It's important to re- be able to represent that as well.

Douglas Lenat 1:16:07 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-

Lex Fridman 1:16:26 Mm-hmm.

Douglas Lenat 1:16:27 ... 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-

Lex Fridman 1:16:45 (laughs) Yeah.

Douglas Lenat 1:16:45 ... um, little tiny baby steps-

Lex Fridman 1:16:47 Yeah.

Douglas Lenat 1:16:47 ... in the right direction.

Lex Fridman 1:16:48 You should just dive in. (laughs)

Douglas Lenat 1:16:51 And, uh, and, you know, if, if no one else is diving in, then yes, taking a baby step in the right direction is better than nothing. Uh, but it's not gonna be sufficient to actually get you the, um, the realization of the semantic web dream, which is what we all want.

Lex Fridman 1:17:06 From a flip side of that, I always wondered, you know, I've built a bunch of websites just for fun, whatever. Or say I'm a Wikipedia contributor.

Lex Fridman 1:17:16 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?

Douglas Lenat 1:17:57 I- I'd love to see exactly that kind of semi-automated understanding of what people write and what people say.

Lex Fridman 1:18:08 Mm-hmm.

Douglas Lenat 1:18:10 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-

Lex Fridman 1:18:23 Mm-hmm.

Douglas Lenat 1:18:23 ... 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-

Lex Fridman 1:18:51 Mm-hmm.

Douglas Lenat 1:18:52 ... what you've said.

Lex Fridman 1:18:53 And then you help 'em out.

Douglas Lenat 1:18:54 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.

Lex Fridman 1:19:49 Mm-hmm.

Douglas Lenat 1:19:50 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."

Lex Fridman 1:21:05 Mm-hmm.

Douglas Lenat 1:21:06 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.

Lex Fridman 1:21:15 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.

Douglas Lenat 1:21:38 Yes.

Lex Fridman 1:21:38 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-

Douglas Lenat 1:22:01 People love doing that.

Lex Fridman 1:22:02 We- we- we- there's joy to teaching.

Douglas Lenat 1:22:04 Absolutely.

Lex Fridman 1:22:04 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.

Douglas Lenat 1:22:31 What- what you're suggesting is, uh, much better even than, um, you thought it was.

Lex Fridman 1:22:37 (laughs)

Douglas Lenat 1:22:37 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.

Lex Fridman 1:23:25 Mm-hmm.

Douglas Lenat 1:23:26 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.

Lex Fridman 1:24:00 Mm-hmm.

Douglas Lenat 1:24:00 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.

Lex Fridman 1:24:23 Mm-hmm.

Douglas Lenat 1:24:23 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-

Lex Fridman 1:24:38 Mm-hmm.

Douglas Lenat 1:24:38 ... because there's no pedagogical usefulness to it. So, from your point of view as the player, you feel like you've taught it something-

Lex Fridman 1:24:46 (laughs)

Douglas Lenat 1:24:46 ... 'cause it used to make this mistake and now it doesn't-

Lex Fridman 1:24:48 Mm-hmm.

Douglas Lenat 1:24:49 ... and so on. So, there's tremendous, um, reinforcement and engagement, um, because of that and so on. So, having a system that plays the role of, um, a student and having the player play the role of the mentor is enormously powerful, um, type of, um, metaphor, just i- important, uh, way of having this sort of interface designed in a way which will facilitate exactly the kind of learning by teaching, um, that um, uh, that goes on all the time in our lives and yet which is not reflected anywhere almost in a modern education system. It was reflected in the education system that existed in Europe in the 17 and 1800s, monitorial and Lancastrian, um, education systems. It occurred in the one-room schoolhouse in the American West-

Lex Fridman 1:25:48 (laughs)

Douglas Lenat 1:25:48 ... in the 1800s and so on, where you had one school room with one teacher and it was basically, you know, five-year-olds to 18-year-olds, who were students.

Lex Fridman 1:25:59 Oh wow.

Douglas Lenat 1:25:59 And so, while the teacher was doing something, half the- half of the students would have to be mentoring the younger kids-

Lex Fridman 1:26:07 Wow.

Douglas Lenat 1:26:07 ... um, um, and so on. And that turned out to, of course, um, with scaling up of education, um, that all went away and that incredibly powerful experience just went away from the whole education, uh, institution as we know it today.

Lex Fridman 1:26:26 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?

Douglas Lenat 1:27:04 This may sound contradictory, but I-

Lex Fridman 1:27:06 (laughs)

Douglas Lenat 1:27:08 ... I think it's the feeling that this will be the only time in history that anyone ever has to teach a computer-

Lex Fridman 1:27:20 Mm-hmm.

Douglas Lenat 1:27:20 ... 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