The following is a conversation with Doug Leonard creator of psych. A system, therefore close the forty 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 the reason and to understand the world support this by gas preacher got our sponsors in the description of the side. No, let me say, that, in the excitement of the modern era, 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 quality is behind psych in some form in actuality or in spirit, will likely be part of the aid 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, more rational and more kind to each other as you do, or do a few minutes of as now no ads in the middle. I try to make this interesting. So hopefully you don't skip, but if you do please still check out the sponsor links in the description is the best way to support this podcast. I use the stuff I enjoy. It may be you will too, they show is brought to you by squarespace. They make it easy for you to make your own website. I use squarespace to create god

It's challenged dot com back in the spring for the four by four by forty eight challenge. Will you run four miles every year for hours for forty eight hours so ends up being forty eight miles, it literally took a few minutes to create the website where there was a forum where people can submit their story leading up and after the challenge, and also I was able to provide basic information about when the challenges and so on. David goggins asked me if I could do real quick as that no problem got it done. Squarespace is the reason I could do it so fast and the final result looks pretty good. I think we are put the website on hold for awhile until the the next day of august challenge, like I said, I think it's great for one off events.

And major personal sites, is good for both quick websites. Big websites doesn't matter, go to lex Friedman, dot com, slash squarespace for a free trial and when you're ready to launch use their offer code legs to save ten percent off your first purchase of a website or domain. The next sponsor is by optimizers. They have a new magnesium supplement when I fast on doing keto carnivore diets. Sodium potassium magnesium are essential. Magnesium, I think, is the tricky one to get right. That's why use magnesium breakthrough from by optimizers most supplements contained only one or two force, magnesium like glass and eight or citrate, when in reality there are at least seven that your body needs and benefits from? andrew human, as somebody who talked a lot about this stuff, especially magnesium, supplementing general but off line or online. I forget who did a is to grow like this now deleted the talked about why

What are the different magnesium supplementation 's that are good and for me, magnesium breakthrough from by optimizers covers all of that: go to mag breakthrough, dot com, slash lex, for a special discount, that's mag, breakthrough, dot com, slash flex! This show is brought to you by stamps dot, com, mail and ship. Any time anywhere right from your computer without needing to take a trip to the post office, sent letters, ship packages and pay less with discount a raise from you as ps, and now, u p s in the past? I've used it to ship shirts, but in general, just doing this ad read is a little bit surreal, because so many of my favorite podcasts in the past have done ad reads for stamps: dot com so feel that come covering an old gray

It's song in doing this I'd read so I honestly feel like I've made it. I love both. Stems that come the service and stands that gum the ad read anyway save time and money were stamps dot com there's no risk all my promo code lex. You get a special offer that includes the four week trial, plus free postage and digital scale. No long term commitments or contracts just go to stamps dot, com click on the microphone at the top of the home page and type in lex, that stamps dot com, promo code, legs stamps, dot, com, never go to the post office again, this show is also brought to you by element, spelled spout element t it's an electrolyte drink mix like I've said made as before. To do look

dies correctly. The number one thing you have to get right as the electric lights: that's the differences between feeling, great and killing. Crappy, specifically, the electrolyte makes you should be concerned about is sodium potassium magnesium. That's exactly what elements takes care of four you plus the thing taste delicious. I just have a thirty two ounce bottle of water and mix in some element in there one packet mix it all up, it tastes delicious. I make sure I drink a few of those in a day and I got all the electrolytes covered. It's super easy delicious. Now, there's no excuse not to drink a lot of water and get the electrolytes both of the things that are very important. I think to feel great on a low carb died or when you doing interment fasting r, R, O mad one meal a day

it also say they have a new flavor. It's actually my favorite called watermelon salt olympians use it tech people use it. I swear by this stuff: try it at drink element, element ti dot, com, slash, lex, that's drink, l, m, n t dot com, slash lex: this show is awesome. but you by express vp, am I have used them for many years. I've loved you in there for many years. I use them today to protect my privacy on the internet. You may. When you use incognito mode and your browser that your data is super secret and protected from the outside world. In fact, I espy's can still collect that data so to protect yourself, the very base layer should be a great vpn and express vpn. Is my favorite vpn Another feature of it is that you can change location to watch, shows from a diff

parts of the world, you can basically be in the united states and make it appear to the internet as if you're in a totally different location in the world. It's a way to travel in cyberspace in terms of usability, I think is probably the biggest reason I love expressing. Is it's fast. It works out a device super easy to use. One button to get the job done. Operating systems. I can think of our supported. Linux included my favorite operating system and you go to express, appeared the cost less like spot to get an extra three months free, that's express gps, dot, com slush lux pod. This is the legs. Freedom in pakistan in here is my conversation

I would doug Leonard. The psych is a project launched by you in nineteen, eighty four and still is active today, whose goal is to assemble a knowledge base that spans the basic concepts and rules, but how the world works. In other words, it hopes to capture common

says knowledge, which is a lot harder than sounds. Can you elaborate on this mission and maybe perhaps speak to the various sub goals? Within this mission, 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 programmes. So natural language, understanding programmes, robots, expert, sis and so on, and we kept hitting the very same brick wall. Our systems would have impressive early successes, and so, if you're only goal was academic, namely to get enough material to write a journal article that might actually suffice? But if you're really trying to get a, I then you have to somehow get past the brick wall on the brick wall. Was the programmes didn't have

what we would call common sense. They didn't have general world knowledge, they didn't we understand what they were doing, what they were saying, what they were being asked and so very much like a on 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 you get a dog to fetch your morning newspaper. The dog might do that successfully, but the dog has no idea what it newspaper is or what it says or anything like that. What does it mean to understand something came may be elaborated, not a little bit. Is it his understand and action of like Mining little things together like through inference, oars, understand the wisdom. You gain overtime that forms a knowledge I think of it. standing more like a think of it more like the ground, you stand on, which,

could be very shaky, could be very unsafe, but most of the time is not because underneath it is more ground and eventually you know, rock and and other things, but 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 and and I've never used this analogy before so bear with me, but I think the same thing is true in in terms of getting computers to understand things, which is odd. You ask a computer, a question, for instance alexa or some robot, or something, and maybe it gets the right answer, 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 hands and say I don't know it just the way the world is, 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 and it may be fine to have a small program- a small set of rules- does that cover you? No ninety, nine percent of the cases, but that one percent of the time when something strange happens you really to draw on common sense. For instance, my wife

were driving recently come and there was a trash truck in front of us and I guess they had packed it to full and the back exploded and trash bags when everywhere and we had to make a split second decision. Are we going to slam on her brakes? Are we gonna swerve into another lane and are we going to run it over because their cars all around us and now in front of us was a large tree. bag and we know what we throw away in trash bags, probably not a safe thing to run over and over on the the left was on a bunch of fast food restaurant. I'm trash bag, like. Oh, those things are just like styrofoam and left over food we'll run over that, and so that was a safe thing for us too. To do now, that's the kind of thing that can happen baby once in your life, and but the point is that there is almost no telling

what little bits of knowledge about the world you might actually need, I'm in some sense, nations which were unforeseen but see and you sit on that mountain or that ground. That goes deep of knowledge in order to make a split second decision about a fast food trash or random trash from the back of a trash truck You need to be able to leverage that ground. You stand stana somewhere, not merely you know it's not enough to just, I have a lot of ground to stand on its your ability to leverage it to utilise in a split integrated altogether to make their split second decision and I suppose understanding isn't just.

having common sense knowledge to access it's, the act of accessing accessing is somehow like cook correctly and filtering out. Then the parts of knowledge not useful, selecting only the useful parts and the effectively making conclusive decisions. So, let's tease apart two different tasks, really, both of which are an incredibly important and even necessary on, if you're going to have this in a usable or useful usable fashion, as opposed to say, like library, books sitting on a shelf and so on and where the knowledge might be there. But you know if a fire comes, the books are going to burn because they don't know what's in them and they're just gonna sit there while they burn I'm. So there too, there are two aspects of using

knowledge. One is a kind of a theoretical. How is it possible at all and then the second aspect of what you said is: how can you do it quickly enough I'm? So how can you do it at all is something that philosophers have grappled with. Unfortunately, fly suffers a hundred years ago and even earlier developed a kind formal language like english, it's called predicate logic or first order logic or something what predicate calculus and so on. So there's a way of representing things in this form of language which enables a mechanical procedure to sort of brine through and algorithmic. We produce all of the same

logical, entail means all the same, logical conclusions that you were. I would from that seem set of pieces of information that are represented that way, and so on that that sort of raises a couple questions. What is how do you get all this information from say, observations and english and so on into this logical form? And secondly, how can you then efficiently run these algorithms to actually get the information you need in. In the case I mentioned in a tenth of a second rather than say, in no ten hours or ten thousand years of computation, and those are both really important questions at. Like a corollary, since the first one is how many such things dna to gather for to be useful in certain contexts

like what to in the image of philosophers in order to capture this world and represented in a law. If you go away enough and with a formal logic like how many statements are required as it five is it ten, is it ten trillion? Is it like that? That's, as far as I understand, is probably still an open question and may forever be an open question just to say like definitively about it, described the universe perfectly well, I asked him if I could. I I guess, I'm going to disappoint you by giving you an actual answer to your question. Okay, when the sounds exciting, guess: okay, so so now we have like three three things: to talk about adding more, it's, ok, the first and the third related. So, let's leave the efficiency question aside for now so

how. How does all this information get represented in logical form so that these algorithms, resolution thier, improving and other algorithms can actually grind through all the logical consequences of what you said and that ties in to your question about. But how many of these things do you need, Because if the answer is small enough, then by hand, you could write them out one time. So I'm in the nineteen early nineteen eighty four I held a meeting at Stanford where I was a family member there where we assembled about half a dozen of the smartest people. I know

people like Alan Newell and marvin, minsky and and Alan K and a few others was firemen, thereby task? Is he he liked yours? He commented upon your system, your go at the time. Do you know who you are and part of this meeting, and I invite that's a heckler meaning anyway. If I think I advised him, I was there. I think I just letter burgers there. So we have all these different smart people and we were became together to address the question that that you raised, which is, if it's important, to represent common sense, knowledge and world knowledge, in order for a eyes to not be brittle in order for a eyes not to just have the veneer of intelligence. Well, how many pieces of common sense many. If then rules, for instance, would we have to actually right in order to essentially cover what what people expect

perfect strangers to already know about the world, and I expect There would be an enormous divergence of opinion and competition. But amazingly, everyone got an answer which was around a million on and one per and one person got the answer by saying. Well, look: you can only burn to human long term memory, a certain number of things unit time, like maybe one every thirty seconds or something and other than that, it's just short term memory, it flows away like water and so on. So by the time your say ten years old or so on, how many things could you possibly? have burned into your long term memory, and it's like about a million another person went completely different direction and said. Well, if you look at the number of words and in a dictionary

the whole dictionary, but for someone to essentially on be considered to be fluent in a language how men words what they need to know and then about how many things about each word. Would you have to tell it so they got to a million that way. Another another person said well, let's actually look at one single on short, one volume desk encyclopedia article and so will look at no. What was like a a four paragraph article or something I think about grebes grebes or a type of water fowl, and we were going to sit there and represent every single thing that was their helmets 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 so all of these estimates not with a million, and so I am, if you do the math, it turns out that, like

oh well, then maybe in something like I'm a hundred person years in one or two person centuries, we could act, We get this written down by hand and a marvellous a coincidence. Opportunity existed right at that point in time. The early nineteen eightys there was something called the japanese fifth generation computing effort. Japan had threatened to do who in computing and a high and hardware, but they had just finished doing in consumer products and the automotive industry, namely wresting control away from the united states and more generally away from the west, and so America was scared and and congress. I did something: that's how you know it was a long time ago, because congress did congress passed something called the national cooperative research

act in c r s 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 multi national companies? Hey all you big american companies nor molly would be an antitrust violation if you colluded on r and d, but we promise for the next ten years. We won't prosecute any of you if you do that to help combat this threat, and so overnight the first two consortia research consortia in america, sprang up both of them. Coincidentally, in Austin Texas, one called sam attack focusing on hardware chips, insolent and then one cow, m c c, the microelectronics and computer technology corporation, focusing on more on software on

it bases in a high and natural language? Understanding, and things like that, and I got the opportunity thanks to my friend, woody bledsoe, who were on was one of the people who founded that to come and be its principal scientist, and he said you know, and he sent up admiral bogdan men who was the person running m c c. I came and talk to me and said: look professor nick you're talkin about this project. It's going to involve a person. Centuries of effort of you've only got a handful of graduate students to do the math. It's going to take you like, and you know longer than the rest of your life to finish this project. But if you move to the

wilds of Austin Texas. I will put ten times as many people on it and you know you'll be done in a few years, and so that was pretty exciting, and so am I did that I am. I took my leave from Stanford. I came to two austin. I worked for m c c and good news and bad news. The bad news is that, all of us, were off by an order of magnitude on their turns up what you need, our tens of millions of these pieces of knowledge about on every day sort of like on. If you have a coffee cup with stuff in it and you turn it upside down the stuff, it's going to fall out, and so you need tens of millions of pieces of knowledge like that. Even if you take the trouble to make each one as general as it possibly could be, and but the good news was that thanks to initially the fifth generation effort and then later on you government agency funding and so

and we re able to get enough funding not for a couple person centuries of time, but for a couple person millennia of time, which is what you spent since nineteen eighty four getting sick to contain, the tens of millions of rules that it needs in order to really capture and span acid, if not all, of human knowledge, but the things that you assume other people know the things you count on. Other people knowing and so by now we ve done that and the good news is, since you waited thirty eight years, just about to talk to me, iver about at the at the end of that process and most of what we're doing now is not putting in even what you would consider common sense, but more putting in undue domain specific application. Specific knowledge about health care in a certain high but all or about hum

oil pipes getting on clogged up or whatever the applications happened to be so vamos come circle and we're doing things very much like the expert systems of the nineteen seventies and nineteen eighties, except instead of resting on nothing and being brittle? now resting on this massive pyramid ill. If you will this massive lattice of common sense knowledge, that. When things go wrong, when something unexpected happens, they can follow. on more and more and more general principles on events the bottomed out in things like, for instance, if we have a problem with the microphone one of the things they do is on I get plug in again and hope for the best right, because that's one of the general pieces of now

if you have in dealing with electronic equipment or software systems or on things like that. Is there basic principle like that like is it? Is it possible to encode something that generally captures this idea of turn it off and turn it back on and see for fixes? Oh, absolutely that's one of the things that on that site knows it's is actually one of the fundamental laws of nature. I wouldn't We call it a law, it's it's more like a seems to work every time, so it sure is sure that looks like a law at an hour so that that basically on uncovered the

resources needed, and then we had to devise a method to actually figure out for what are the tens of millions of things that we need to tell the system and for that we've found a few techniques which worked really well. One is to take on any piece of text almost could be. An advertisement could be a transcript. It could be a novel, it could be an article on and don't pay attention to the actual type. That's there, the the black space on the white page pay attention to the cop a bit of that the white space, if you well. 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 the? Why did they think that you would be able to understand what the intended reference of that pronoun months if they used an ambiguous word? How did they think that you would be able to figure out what they meant by that word?

and the other thing we look at is the gap between one sentence and the next one. What are the things that the writer expected you to fill in and infer occurred between the end of it? His sentence in the beginning of the others so like if the sentence says? Ah Fred smith, rob the third national bank period and I he was sentenced to twenty years in prison period well between the first sentence and the second you're expected to infer things like fred, got, caught fred and got arrested, Fred went to jail, Fred had a trial, Fred was found guilty and so on. If my next sentence starts out with something like the judge, dot dot,

dot, and then you assume it's the judge at his trial. If my next sentence starts out something like the arresting officer, dot, dot dot, you assume that it was the police officer who arrested him after he committed the crime and so on. So and that's it. Those are two techniques for getting that knowledge. The the other thing we sometimes look at is ah sort of like take news or set of humorous some onion headlines or on headlines, the weekly world news. If you know that is or the national inquire where it's like a we dont believe this. Then we interest backed on. Why don't we believe it said there things like gun, seventeen 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 or something like that, or Another thing we look for our contradictions where, with things which can both be true- 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 on an addition. in one article? It talked about how Elvis was cited now, even though he was kind of getting on in years and so on. another article in the same one talked about people seeing Elvis's, ghost, okay, so like white, why do we believe that at least one of these articles he know must be wrong and so on so, and so we have a series of techniques that that enable our people, and by now we have about fifty people working full time on this and have worked for decades. So we ve put in the thousands of person years of effort. We build up these tens of millions of rules. We constantly police the system to make sure that were saying things has generally, as we possibly can

so you don't want to say things like. I'm. No mouse is also a moose, because if you said things like that, yon then have to add another one or two or three zeroes onto the number of assertions you'd actually have to have. So at some point we generally as things more and more, we get to a point where we say oh yeah, for any to biological tax on. If we don't know explicitly that one is the generalization of another, then almost sir, Lee their disjoint on a number of one is not going to be a member of the other and so and so and the same thing with it. was in the ghost as I do with Elvis it's more about him, nature and though the mortality in via contriving general things are not both alive and dead at the same time, and unless but special cats and in theoretical physics, examples well. That raises a couple

fourteen points for this. The onion headline situation depth, I guess, but no, no soap. So what you bring up. Is this really important point of like? Well? How do you handle exceptions and inconsistent seas So on and one of the hardest lessons for us learn it talks about five years to to really create our teeth and learn to love. It is we had to give up global consistency, so the knowledge base can no longer be consistent. So this is a kind of sky every thought I grew up watching star trek and any time a computer was inconsistent. It would either frees up or explode or take over the world or something bad would have, an arm or if you come from a mathematics background once you can prove false, you can prove anything. So that's not good, and so, and so and that's why, on the I'm old knowledge based systems were all very, very consistent,

The trouble is that by and large are models of the world. The way we talk about the world, and so there are all sorts of inconsistency the creep in here and there and that will sort of kill some attempt to build some enormous globally, consistent knowledge base, and so what we had to move to as a system of local consistency. So a good analogy is you know that the surface of the earth is more or less fearful globally, but You live your life every day as though the surface of the earth were flat, knowing it talking to someone in Australia. You don't think of them as being oriented upside down to you when you're planning a trip know. Even if it's a thousand miles away, you may 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,

really worrying about that, because the earth is locally flat in much the same way. The site 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. Inconsistency is at the boundary between between one context and the next one and so on, and so by the time you move sake. Twenty context, it's over there could be glaring inconsistencies. So eventually you get from of the normal modern, real world context that we're in right now to something you know like a roadrunner cartoon context where physics is very different and in fact life in the are very different because no matter how many times is killed, you know the coyote comes back. in the next seen in and so on, so that that was a

lesson to learn, and we had to make sure that our representation language, the way we were 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 other things that are true. Let's say in one region like one country but false in another things, Their true in one person's belief system, but false in another person's belief system 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 down at the atomic level, it's mostly empty space, send so on the southern. That's fascinating and put it puts a lot of pressure on contacts to do a lot of work. Seasick tectonic plates. Is it possible the formula contexts their general and big that do this kind of capture of knowledge basis or do than get,

girls and aperture paternally again, where there's just a huge number of contacts. So it's good! You ask that question cause your pointed in the right direction, which is you want contacts to be first class objects in your systems. Knowledge base, in particular in sykes knowledge base, and by first ass object. I mean that it should. We should be able to have psych, think about and talk about and reason about, one context or in other contexts the same way. It reasons about coffee, cups and tables and people and fishing and so on.

and so contacts are just terms in its language, just like the ones I mentioned, and so psych can reason about and context can arrange, hierarchically and so on. 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 a broad, let's say a century or a millennium, or something like that things are true. In Austin, Texas are generally gonna. his specialization of on things that are true in texas, which is going to be a specialization of things that are true in the united states and so on, and so you don't have to say things over and over over again on at all these levels. You just say things at the most general level that it applies to, and you only have to say at once, and then it essentially inherits to all these more specific on content

it's just a slightly technical question: is this inheriting a heritage in a tree or a graph? Oh, you definitely have to think of it as the graph, and 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, and so each node in their representation could only have one parent knowed. So if you had a table, that was a wooden object, black object flat object, and so you have to choose one and that's the only parent. It could have 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 it for 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 one of the 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 it most on ten children and each node could only have one parent and while that does enable the dewey decimal type numbering of concepts labeling of concepts. It prevents you from presenting all the things you need to about our objects in our in our world, and that was one of the things which and they never were able to overcome, and I think that was one of the main reasons that that project failed so will return to some of the doors of open. But if you can Back to the room and then tee for around there with marv amidst get effort, but by the way I should mention that marvin wouldn't do his estimate tell someone brought him an envelope so that he could litter

we do it back of the anvil up calculation to come up with a number of well, because I feel like Egg. The conversation in that room is an important one in this this. How synthesize is done in this way a few people get together and planted seed of ideas and they reverberate throughout history and some some kind of odd dissipated disappear and some in our drake equation and dinner. They It seems like a meaningless equation, somewhat meaningless, but I think it drives and what they thought. A scientist. When the aliens finally show up the equation get even more or because they will get be able to us in the long arc of history, the draken kwairyo prayer wheel proved to be quite useful.

I think in the same way a conversation of just how many facts are required to capture the basic comes as knowledge of the world as a fast a question. I wanted to distinguish between what you think of his facts and the kind of things that we represent. So we we map to and essentially make sure that psych has the ability to, as it were, red and access the kind of facts you might find say I'm in a wiki data, more stated in no wikipedia article or something like that. So what we're representing the things that we need, a small number of tens of millions of our more like rules of thumb, rules of good, guessing things which are you usually true and which help you to make sense of the facts that are your answer to sitting. Often some database or some other more static story. There there almost like botanic forms. So when you read stuff on Wikipedia that's going to be like projection

Those ideas, you re an article about the fact that Elvis died, ass, a projection idea, that humans are mortal unlike very few wikipedia article right humans are mortal exactly that's. What I meant about ferreting out the unstated things and tell us what are all the things that were assumed, and so those are things like if you have a problem with something turning it often on often fixes it for real, we don't really understand and were not happy about or on people can't be both alive and dead at the same time and or water flows on hill. If you search online for water flowing up hill and water, flowing downhill, you'll find more references for water flowing uphill, because it's used as a kind of a metaphorical reference for some unlikely thing be of course everyone already knows that water flows downhill. So why would anyone bother saying that Do you have a word? You prefer who said faxes in them

word is their word like concepts like, I would say, assertions assertions or rules, because I'm not talking about rigid rules but rules of thumb but assertions it's a nice one that covers all of these things. Yeah, as programmers me assert, has a very dogmatic, authoritarian fields that I'm sorry I'm so ass. I look at what is it assertions works So if we go back to that room when our remains key with you all these seminal figures, Add value, mom thinking about this very philosophical, but also engineering question. You can also go back.

couple of decades before then, and thinking about artificial intelligence broadly when people are thinking about. You know how do create super intelligent 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 exceptionally difficult ideas not always is very difficult to true. I understand ourselves were worth thinking about the human mind, to two interests back how difficult that is. engineer intelligence. The solvent challenges were not rig. estimating that, and you are somebody who was really stayed with this question for decades, Do what's your sense from the nigeria for today you got a stronger sense of just how much knowledge is required. He's gotta said

Some level certainty that still an order of magnitude of tens of millions right by a for the first several years it is said that it was on the order of one or two two million yeah, and so it took it, took us about five or six years to realize that we were off by it by a factor of ten. But I guess what I'm asking you know: marvin misc is very confident in the sixties when yeah right. What's your sense, if you. You know two hundred years from now. You're still, you know you're you're you're not going to be any longer in this particular biological body, but your brain will still be ah in the digital form and you'd be looking back. Would you think you were smart? I like your intuition- was right or do you think you be really off. So I think I'm I'm right enough.

and let me explain what I mean by that, which is sometimes like. If you have an old fashioned pump, you after prime, the pomp and then eventually starts though, I think I'm I'm right enough in the sense that a problem the pump, what we have, what we ve built, even if it isn't so to speak everything you need, its primed. The knowledge pump enough that sake canal itself help You learn more and more automatically on its own by reading things and understanding, and occasionally asking questions like her a student would or something and by doing experiments in discovering things on its own and on so through a combination of sake, powered discovery and psych powered reading on it will be able to bootstrap itself

it's the final two percent, maybe it's the final, ninety nine percent. So even if I'm, if I'm wrong, I'm all I really need to to build is a system which has primed the pump enough that it can begin and that cascade upward that self reinforcing answered, quad drastically or maybe even exponentially, increasing path upward that we get from, for instance, talking with each other. That's why humans today know so much more than humans? A hundred thousand years ago were not really that much smarter than people were a hundred thousand years ago, but there's so much more knowledge and we have language we can communicate. We can check things on Google and so on so effectively. We have this enormous power at our fingertips 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 standing 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 over that slower as a way of gathering data and so on and anna this is really important point, which is if we have our initial intelligence, real general, artificial intelligence, human level, artificial intelligence, then people will become smarter. It's not so much that it'll be us versus the eyes. It's more like us and the eyes together will be able to do things that require more creativity, that, would take too long right now, but will be able to do lots of things in parallel. I will be able to misunderstand each other less this all sorts of value, though factually for an individual would mean that individual will, for all intents and purposes, be smarter, and that means that humanity, as a species we'll be smarter, and when was the last time that any info

jane qualitatively in made a huge difference in human intelligence, You have to go back a long ways. It wasn't like the internet or the computer or mathematics or something it was all the way back to the development of language. We sort of look back on pre, linguistic cave men as well. You know they. They were really intelligent worthy in work really human worthy, and I think that, as you said fifty year, hundred two hundred years from now on, people will look back on people today on a bright before the advent of these of lifelong general, ai uses, muses and say this poor, this poor people. They were really human, where they do exactly as you said, a lot of really interesting things by the way, I would maybe try to argue that the internet-

It is on the order of the kind of big leap in improvement, that the invention of language was but is certainly a big leap in one direction. We're not sure whether its upward her down what I mean is 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 very quickly. So I could take. their side of this argument, and since you just took one sign, I greatly the other side, which is that almost nothing It's done more harm than something like the internet and access to the information, two ways: one is its made: people more globally ignorant on in there same way that calculators made us more or less enumerate. So,

I was growing up. We had to use slide rules, we had to be able to estimate, and so on today. I'm people, don't we we understand numbers, they don't really understand math. They dont really estimate very well at all and so and they don't really understand the difference between millions and billions and billions and so on. Very well, because calculators do that all for us and on thanks to things like the internet and search engines that same kind of juvenile ism is reinforced in making people on the centrally. be able to live their whole lives, not just without being able to do a rhythm and taken estimate, but now without actually having to really know almost anything because the time they need to know something I'll just go and look it up Urim and then I could tell you, give people sizes said, is a double edged sword. You can of course say- thinking about language, probably people when they invented language, they will criticise it used to be with you

If you're angry, we just kill a person and for a lover was just have sex with them, and now everybody's writing, poetry and bullshit that you know you should just be direct. He should like have physical contact enough of his words and books and his you're you're not actually experienced like. If you read a book you're not experiencing the thing nonsense that that's right, if you read a book about how to make butter, that's not the same as you can't die like learn it and do it yourself. I consider yourself and so on so so let me say that something is gained, but something is lost every time you of these, these sorts of dependencies on technology and overall I think that the having smarter individuals are having smarter ay. I augmented and human species will be one of the few ways that will actually be able to overcome some of the global problems. We have involving on poverty and starvation and global warming, and

we're crowding all the other problems that that arm besetting the planet? We really need to be smarter and they really only two routes to being smarter. One is through biochemistry and genetics, genetic engineering, other route is through having general ai eyes that augment our intelligence and and- and you know- hopefully, one of those too- there are ways of up paths to salvation- will Will come through before it's too late? He asked that I agree with you and obviously, as an engineer I have, I have a better sense and an optimism about the technology side of things, because you can control things they're more biologist. 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. Your unit is now like a battle of humans. Vs virus is just like okay and then you can just wipe out an entire species. The other problem with the internet is that it has enabled us to surround ourselves with an echo chamber with a bubble of, unlike minded people, which means that and have a truly bizarre theories. Conspiracy theories, fake news and sullen promulgated and surround yourself with people who essentially reinforce what you to believe or what you already believe about the world and in the in the old days. That was much harder to do. Fine, you had say only three tv net works or even before, when you had no tv network seem to actually like look at the world and make your own recent decision elect the push him pool of our dash forward.

because then I'll just say in the old world having come from the soviet union, because he had won a cup of networks than propaganda, be much more effective and then the government can overpowers people by telling you the truth and then starving millions and torturing milieu and putting millions at the camps as starting wars with a proper? Ghana, machine allowing you to believe there. You should get doing good in the world with the internet because of all the quote, unquote. Conspiracy theories. Some of them are actually challenging the power centres. The very kind of power centres that a century ago would have led to the death. is so there's a is again his double edged sword, and I I very much with your on the outside its it's often and intuition, that people have that somehow I we used to maybe overpower people by certain select groups and to me is that all of you,

Thus, the likely scenario to me, the likely scenario, especially just having observed the trajectory of technology, is there'll, be used to empower. People will be used to extend the capabilities of the visuals across the world. Because there's a lot of money to be made, though I like improving people's lives, you can make a lot of money agree. I think that the the main the main thing that ai prostheses, ai amplifiers, will do for people, make it easier- maybe even unavoidable- for them to do good, critical thinking, I'm so pointing out logical fallacies, logical contradictions and so on. In things that they otherwise would just blithely believe, pointing out essentially data which

they should take into consideration if they really want to learn the truth about something and so on. So I think I'm doing just educating in the sense of pouring facts into people's heads. But educating in the sense of Many people with the ability to do good, critical thinking on is enough. Mostly powerful on the education system that we have in the: u s and worldwide generally don't do a good job of that, but, I believe that the air I will for the elderly as well the eyes can and will in the same way that everyone can have their own election. our siri or on Google earth, dinner, whatever I'm every. When will have this sort of cradle to grave assistant, which will get to know you which you'll get to trust it'll model, you your model, it and I'm it'll, call to your attention things which will

in some sense, make your life better, easier and less mistake, ridden and so on, less regret ridden. Ah, if you listen to listen to it in full agreement with you about this like space of technologies- and I think it's super exciting. From my perspective, integrating emotional intelligence so even things like friendship and companionship and love into those guys systems that, as opposed to help you you just oh 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, it is not just. The pursuit of reason is the pursuit of happiness to ass, the for the forced back well, let me say, because you mentioned if this so many faceted things when we come back to them idea of automated reasoning, so the acquisition of new knowledge has been done in his voice.

Interesting way, but primarily by humans, doing this Just you could think of the monks in their cells in medieval europe. Me no careful the illuminating manuscripts. Yes, though, on a very difficult and amazing process, actually because it allows you to truly asked the question about, in the white space. What is assumed, I think this exercise is that very few people do this right. They just subconsciously there or foreign by definition, by having has because those pieces of alighted of omitted information of those missing steps, as it were, r p, it's common sense. If you actually included all of them it would, I would almost be offensive or confusing to the reader like. Why are they tell me all these? Of course, I know that all these things, and so I so it's it's one of these things which

almost by its very nature, has has almost never been explicitly written down anywhere, because by the time you are old enough to talk to other people and so on, you know if you survived to that age. Zoom you already got pieces of common sense like no. If something causes you pain, whenever you do it, probably not a good idea to keep doing it. So what ideas do you have given? Difficult this step is what ideas are there for how to do it on a map, without using humans, or at least not the you know, doing like a large percent did you have the work for humans and then humans only do the various high level supervisory work. So we have and are in fact, two directions were pushing on very, very heavily on currently like. We're in one involves natural language understanding and the ability to read what people have explicitly

written down in and to two points. knowledge in that way, but the other is to build a series of knowledge, editing tools, knowledge, entry tools, knowledge capture tools, knowledge testing tools and so on. I think of them ass, like user interface, on sweet of software tools. If you want something that will help people to more or less automatically expand that extend the system in areas where, for instance, they want to build some out have ado some application, or something like that. So an example of one on which is something called abduction. So you probably heard of like deduction and induction and so on, but abduction is unlike this abduction is not sound, it's just useful. So, for instance,

deductive lee if someone is out in the rain and they're gonna get all wet and they enter room. They might be all wet and saw so that's deduction, but if some where to walk into the room right now and they were dripping wet. We would immediately look outside. I say: oh did it start to rain or something like that now, and why did we say? Maybe it started to rain? That's not a sound logical inference, but it's certainly a reasonable and abs active on 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, and so what? What does that have to do with? What we're talking about so suppose, you're building one of these applications and the system gets some answer wrong and you say, oh yeah. The answer to this question is, and this one

not the one you came up with then, but the system can do as it can use everything. It already knows about common sense, general knowledge. The domain you ve already been telling it about, and contexts like we talked about, and so on and say. Well here are seven alternatives, each of which I leave 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. Is one of these seven things true and then you? The expert, will look at those on seven out and say: oh yeah, number five is actually true, so without actually having to tinker down at the level of logical assertions and so on and you'll be able to educate the sis

in the same way that you would help educate another person who you were trying to apprentice or something like that, so that that significantly reduces the mental effort or significant increases. The efficiency of the teacher, the human today exactly em. It makes more or less anyone able to me to be a teacher In that I'm in that way. So that's that's part of the answer, and then the other is that the system on its own will be able to on through reading through conversations with other people and so on. I'm learn the same way that I'm? U were I or other humans do for sure. That's that beautiful vision I'll have to ask about both semantic web in the second year, but first are there: when we talk about specific techniques, you find something inspiring or directly use

or from the hall space of machine learning, deep learning. These kinds of spaces are techniques that have been shown effective for certain kinds of problems in the recent now decade and a half I think of the machine learning work as more or less what our right brain hemispheres do so being able to take a bite of data and recognised patterns being able to statistically infer things and so on and on the eye certainly wooden one to not have a right brain hemisphere, but I am also glad the heavy left brain hemisphere as well, something that can metaphorically sit back and puff on its pipe and think about on this thing over here like. Why might this have been true and what the implications of an. How should I feel about that, and why and so on so unthinking were deeply,

and slowly and what kahneman called thinking slowly versus thinking quickly, whereas you want machine learning to think quickly, but you want the ability to think deeply, even if it's a little on the slower arm. So I'll give you an example of a project we did recently with on an age involving the cleveland clinic and a couple other institutions that we ran a project for and what it did was it took on glosses, genome wide association studies. How does her odds had a big databases of pay, since they came into a hospital and they got their dna sequence, because the cost of doing that has gone from an infinity to build tens of dollars, two hundred dollars or so, and so now patients routinely get their dna sequence. So you have these big databases of the snips, 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 correlation studies. machine learning, studies of which mutations our associated with and lead to which physiological problems and diseases, and so on, like getting arthritis and on and so on, and the problem is that those correlations turn out to be very spurious, they turn out to be very noisy. Very many of them have led doctors onto wild goose chase and so on and so they wonder way of eliminating or on the bad ones are focusing on the good ones, and so this is where psych comes in witches sake takes those sort of aid. Is the correlations between point mutate, regions and on the medical condition that needs treatment and we say: ok, let's use all is public knowledge and common sense knowledge about what reactions occur where in the human body,

What polymerize is what what catalyzing, what reactions and sullen and let's try to put together a ten or twenty or thirty step causal. Explanation of why I that mutation might have caused that medical condition and so sacred put together in some sense, some rube goldberg like unchecked, in that would say. Oh yeah, that mutation, if it got expressed, would be this. I'm altered protein, which, because of that, if you got to this part of the body, would catalyzed this reaction and by way. That would cause more bio, active vitamin d in the persons blood and anyway, ten steps later that screws screws up bone resorption, and that's why this person got osteoporosis early. Life and saw it as a human interpretive or lease doktor humane until sat exactly now and the important thing even more more. That is you shouldn't really trust that twenty step

have rube goldberg chain any more than you trust. that initial abc correlation, except two things, one, if you can't even think of one causal chain to explain this and then that correlation probably was just noise to begin with and sex Lee and even more powerfully along the way that causal chain will make predictions. The one about having more by active indeed in your blood. So you can now go back to the data about these patients and say by the way, did they have slightly elevated levels of by active vitamin d in their blood and so on, and if the answer is no, that strongly diskin firms your whole causal chain. If the answer is yes, that somewhat confirms that causal chain, and so using that we were able to take this on these correlations from this glass database, and we

able to essentially focused the the doctors focused the researchers attention on the various more percentage of correlations that had some explanation and even better some explanation that also made some independent prediction that they could confirm, or just confirmed by looking at the data, so think of it like this kind of synergy where want the right brain machine learning to quickly come up with possible answers. You want the left brain. Site. Like a I. I not think about that, and now I think about why they might have been the case in 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 on two. So it's that kind of sinner You back and forth, which I think is really what's: gonna, led to general ai, not narrow, brittle machine learning systems and not just something like psych, okay, so that that the ruins centres

but I was also thinking in terms of the ottoman expansion of the knowledge base you mention, and now you this is very early days in the machine learning space of this but itself super, learning methods. You know these language models jpg three and so on. They just read the internet and they former intentions that can then be map to something use the question is: what is the useful thing like there now playing with a pretty cool thing called open air codecs? which is generating programmes from documentation. Ok, that's how useful is. But my question is: can be used to generate in part maybe with some human supervision, a site like assertions feed psych more assertions from this giant body of internet data. Yes, that that is in fact, one of our goals is how,

can we harness machine learning? How can we harness natural language processing on to increase singly automate. The knowledge act, efficient process, the growth of sight and that's what I meant by priming, the pump that you know. If you, you sort of learn, things the fringe of what you know already. Aren't you learn this new thing is similar to what you know already, and here are the differences in a 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 really hard to learn anything, and so, if you're, not careful, if you start out with two small sort of a cord start this process, it, never really takes off, and so that's why. I view this as upon priming exercise to get a big enough manually print

even though that's kind of ugly duckling technique put in the elbow greece to produce a large enough core on that, you will be able to do the kinds of things you're, imagining without without sort of on ending up with the kind of wacky brittle nisus that we see, for example, in jpg three, where it, didn't go. Tell a story about it as someone knob pudding poison, no plotting to poison someone and so on, and then the you know, then you know but p three says of. What's that you say what The very next sentence, the next sentences: oh yeah, that person then drank the poison they just put together like that's, probably not what happened. Recycling or if you Two siri- and you know I think I have a newark we're gonna, go

for I'm help with my on alcohol problem or something it will come back and say, I found seven liquor store is near you, you know, and so on, so, you know it's one of these things where, yes, it may be helpful on most of the time It may even be correct. Most of the time, but if it doesn't really understand why it saying and for 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. if you're only goal is to sort of find relevant information like search engines, do then being right. Ninety It's the time is fantastic, that's unbelievers lingering. Ok, however, your goal is to like you know, save the life of your child, who has the meadow? problem or your goal is to be able to drive. You know for the next ten thousand hours of driving

without getting into a fatal accident and so on. Then you know: error rates down the ten percent level, or even the one percent level not really acceptable. I like the motto of that learning happens at the edge and then you can think of knowledge is the sphere so If you want a large fear because the the learning is happening on the surface. Exactly so you have the what you can learn next increases quad radically as the diameter that sphere but as it is nice because you when you know nothing: it's like you can learn anything but the reality not really. right, if you know, if you know nothing, you can really were, nothing can appear to learn so I'll. Also what one of the anecdotes I could go back and give you about why why I feel so strongly about this personally was

in nineteen, eighty, eighty one by daughter, Nicole, was and she's actually doing fine now, but when she was a baby and she was diagnosed having meningitis and doctors, one july, these scary things and my if I were very worried, 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 short life was another system, professor in computer science, at stanford at the time and he'd been a programme called mice which was a medical diagnosis program that happened a specialised in our blood, in fact, like meningitis, and so he had privileges at stanford hospital. Because he was also an empty and so we got hold of her charge, and we put in her case, and it came

with exactly the same diagnoses and exactly the same therapy recommendations. But the difference was because it was a knowledge based system, a rules based system. It was able to tell us step by step by step on why, on this, was the diagnosis and step by step why this was the best I'm therapy and the best on procedure to on I'm due to do for her and so on, and there was a real epiphany because that made all the difference. The world, instead of blindly having to trust in authority we re able to understand what was actually going on and so I'd 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 the way the world works, why things are the way they are they weren't able to give. explanations of their answer. You know and know it's one thing to do.

He was 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, point eight three and you say No in more detail wine says point eight three one, that's not really very compelling and that's not really very helpful. There's a say of the semantic web and when I first heard about a just, fell in love with it, There was the obvious next step for the internet, sure and Maybe you could speak about what is the semantic web? What are your thoughts about it, your vision and mission and goals with psycho connected integrated mike. Are they danced power? nerves are they aligned what he thought there so think of the semantic web as a kind of knowledge ground and Google already has something. They call knowledge graph, for example, which sort of like I knowed and link diagram. So you have these nodes that represent concepts or words or

firms, and then there are some arcs that connect them that might be labelled and so you might have a node with like one person that represent one person and lets say a a husband link that ten points to that person. his husband and said there'd. Be then another link that went from that person, labeled wife, that went back to the I'm first nude and so on so having having this kind of representation is really good if you under represent binary relations, essentially relations between two things and, if you so, if you have an equivalent of like three word sentences you'll, like god, freds wife, is wilma, who or something like that. You can represent that very nicely using these

kinds of graft structures or using something like the semantic web, then and so on. But the problem is that on very often what you want to be able to express, takes a lot more than three words and a lot more than simple graph structures like that to represent so, for instance, if you re or seen romeo and Juliet now I could say to you something like Remember when Juliet drank the potion that put her into a kind of suspended animate animation when Juliet drank that potion, what did she think that romeo think when he heard from someone that she was dead, and you could basically understand what I'm saying you understand. The quest jane. You could probably remember the answer was well. She thought that this suffering

air would have gotten the message to romeo saying that she was gonna. Do this, but the fire didn't answer so you're able to read. President and reason with these much much much more complicated expressions the go way way beyond what simple on three, as it were, three word or for word english sentences are, which is really what the semantic web can represent. really. What knowledge graphs can represent is a good step back for a second, because it's it's funny! You went into specific, and maybe be can elaborate, but I was also referred to semantics as the vision of converting data on the internet into something that's interpreter bull. stand will by machines. Oh, of course, at that at that level this issue, I wish to say that what is this magic web I mean you could say a lot of things but

It may not be obvious alot 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 key words: search rank by the quality estimate of the website, integrating previous human based, google searches and what they thought was useful Is there some weird combination of bomb like surface level, hacks that work exceptionally well, but they dont understand the current data, the full contents of the websites that searching, so Google does not understand to the degree we ve been about the word understand the contents of the wikipedia pages as part of the search process in a semantic web says: let's try to get come up with a way

a fourth computer should be able to truly understand the contents of those pages. That's the dream! Yes! So, but let me let me first give you, I inadequate and then I will answer your question. So there is a search engine. You ve probably never heard of called northern light and it went out of business, but the weight worked. It was a kind of the empirical search engine and what it did was it didn't index the internet at all. All I did was it a 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 on. Then you know they type in a completely different query, unrelated korean saw, and so it just went from query to the web page that seem to satisfy them on eventually 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 company said. Oh we're not going to say this, it any more so that no one out of business because it had no other way of on taking users to wear them want to go and sullen and of course, the search engines, I now using that kind of idea. Yes, so let's go back to what you said about the semantic web, so the dream: TIM berners, Lee and others on dream, about, semantic web at a general level is, of course, how excite

ding and powerful and in a sense the right dream to have, which is to replace the The kind of statistically statistically mapped linkages on the internet into something that's more meaningful on semantic and actually gets at the understanding of the content and so on, and eventually If you say well, how can we do that? There is a sort of a low road, which is what the knowledge graphs are doing and on and so on, which is to say well, if we just use these. Simple binary relations. We can actually get some fraction of the way ford understanding and do so thing where you know in the in the land of the blind, the one eyed man is king kind of thing and so being able to even just

a toe in the water in the right direction is fantastically powerful, and so that's where a lot of people stop But then you could say well what, if we really wanted to represent and reason with all meaning of what's there, for instance, about romeo and juliet with the reasoning about juliet believes that romania will believe that Juliet believed you know and so on? Or if you look at the news, what you know as an invite and believed that the leaders of it a ban would believe about the leaders. Afghanistan, if they, you know Bob over so in order to represent a complicated sentences like that I'm at let alone reason with them. You need something which is logically much more expert

if so, then these simple triples, then the simple knowledge, graph, type structures and so on, and that's why kicking and screaming we were led from something like ah the semantic web representation, which is where we started in nineteen, eighty, four, with frames and slots with those kinds of trip. rules triple store representation. We were led, kicking and screaming to this more and more general logical language. This higher or your logic, so first were led to first order logic and then second order and then eventually higher order. So you can represent things like me. rules like believes desires intends expects and some in a nested ones you can represent. complicated kinds of negation. You can represent the

process you're going through in trying to answer the question, so you can say things like oh yeah you're trying to do this problem by integration by parts and on you recursively. We get a problem that solved by integration by parts: that's actually ok, but if that happens a third time you're, probably off on a wild goose chase, or something like that, so being able to talk about the problem solving process as you're going through the problem solved in the process, an arm is called reflection and so and that's another so as important to rep be able to represent that exactly. You need to be able to represent all of these things and 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 represented on then it so Like someone with a limited vocabulary, you can't understand as easily what you're trying to tell them in

That's that's really why I think that the the general dream, Original dream of static web is exactly right on, but the implementations that we ve seen are sort of these toe in a what water, little tiny baby steps in the right. Actually she died in an area that you know if, if No one else is diving in then yes, taking a baby step in the right direction is better than nothing, but it's not gonna be sufficient to actually get you the the realisation of the semantic web dream, which is what we all want from a flip side of that I was wondered. I built a much websites just for fun, whatever.

or say a wikipedia contributor? Do you think there's set of tools that I can help psych ah interpret the website I create you know like this again pushing onto the semantic web dream. Is there something from the creator perspective that could be done in one of the things you said with psycho urban site that you are doing is the tooling side, making human more powerful their unease, the other humans in the other side that create the knowledge I forgot. Will unite, having it to three. Whatever our conversation now Their way that I can convert this more make it more accessible. To cite two machines: do you think about that side of it? I I'd love to exactly that kind of stuff. my automated understanding of people right in what people say, I think of it as

the kind of foot noting almost almost like the way that, when you on something in say, microsoft word or some other document preparation system, Google, docs or something you'll. yet underlining of questionable thing, is that you might want to 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 psych powered tools that read through what it is you said or have typed in and And try to partially understand what you said and then you help him out exactly and then they put in little footnotes that will help other, there's and they put in certain footnotes of the form, I'm not sure what you meant here.

Their men is sir this or this I bet, if you take a few seconds to descend big for me, then I'll know and I'll. Have it correct for the next hundred people were the next hundred thousand people who come here 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 real, and with it then, yes, it will be worth you're a small amount of time to go and make sure that the I trying to understand it really did correctly, understand it and you know. Let's say you run a travel with web site or something like that

and people are going to be coming to it because of searches. They did looking for looking for vacations, that or trips that had certain properties and might have been interesting to them for various reasons. Think things like that and if you ve explained, what's going to happen on your trip thin system will be able to mechanically reason and connect what this person is looking for work what it is you actually offering, and so if it understands that there's a free day in Geneva, Switzerland, then If the person coming in happens to let's say I'm being nurse or something like that,

and 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 that means and saw, then it can basically say hey. You might be interested in this rip, because, while you have a free day in Geneva, you might want to visit that red cross museum and now, even though it's not very deep reasoning, little tiny factors like that may very well cause you to sign up for that too. rather than some competitor trail, and so there's a lot about through a ceo, actually kind of think. This is about a lot of things, which is the actual interface the designer. The interface makes a huge difference. How efficient it is to be productive and also how Full of joy, the experiences yeah I am, I may I would love to help a machine and not from an EU perspective just as a human right,

the reasons I really enjoy? How tesla have implemented their autopilot system is there's a sense, the you're helping this machine learn nagging humans, I mean having children, pets, people love doing that. We we we there's joy to teaching apps for suite 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 lake, like annoying like correcting and annoying system more like teaching a child like innocent curious system. I think you can literally just like several orders of magnitude scale the amount of good quality data being added to something like psych. What what you're suggesting is much better. Even then, you thought it was one of the one of the experiences that we have all had 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 thus 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 enjoyable If you are the mentor or the teacher and so on, so we developed a program called math craft to help six, Raiders, better understand, math, and it does and actually tried to teach you the player anything. What it does is it cats,

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 and that trick from the point of view of psych 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 it. deeper and deeper level of understanding of the subject, and so, if you give it good advice about what it should have done, instead of what it did and so on then

psych knows that you now understand that mistake. You, you won't make that kind of estate yourself as much anymore, so psych stops making that mistake, because his know you had a gradual usefulness to it. So, from your point of view, as the player you feel like you, ve taught it something cause it used to make this mistake, and now it doesn't and so on to this tremendous reinforcement and engagement on, because of that, and so on. So having a system that plays the role of a student and having the player play the role of The mentor is enormously powerful type of metaphor, just in portland way of having this sort of interface designed in a way which will facilitate, actually, the kind of learning by teaching that aren't that 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 is, stood in europe in the seventeen and eighteen? Hundreds more. coral and lancastrian education systems. It occurred in the one room. School house in the american west in the eighteen, hundreds So on where you had one schoolroom with one teacher- and it was basically a five year olds to eighteen year olds, students, and so while the teacher was doing something half that how of the students would have to be men, touring the younger kids so on and that turned out to, of course, with scaling up of education, that all went away and that in I powerful experience just went away on the whole education institution

as we know it today, sorry for the roma, the question, but what is the most beautiful idea? He learned artificial intelligence, knowledge reasoning, far more working on site for thirty seven years or maybe what the most beautiful idea, surprising idea about psyched you when I look up at the stars, I gonna want like that that amazement. You feel that while And you are part of creating one of the greatest and most fascinating efforts in artificial intelligence history, so which element brings you personally joy. This may sound contradictory, but I I think it's the feeling that

This will be the only time in history that anyone ever has to teach a computer. This particular thing that we're now teaching it it's it's like painting, starry night, you, you only have to do that once or creating the pierre taught. You only have to do once you it's not it's not like a it's like a singer who has to keeping us not like Bruce Springsteen, having two to sing. His greatest hits over and over again a 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 I really get the sense of we are telling the system things its useful for it to know its useful for a computer to know for an eye to know, and if we do our jobs right when we do our jobs right now

will ever have to do this again. For this particular piece of knowledge, the very very exciting I guess, there's a sadness too It too is a god, there's a magic to being a parent and raising a child and teaching them all about this world. But you know, there's billy and the children right like born of what I would have their numbers of large number love number of choice and a lot of parents get to experience that your teaching were they eyes systems theo? They at least the current constructions. They remember you, don't you don't get to experience. The joy of teaching am a machine millions of times better come work for us before it's too late then hangs. Let's get back at the good hiring page.

yeah it's true, but then there is also you know at the project that continue forever in some sense, just like wikipedia. Yes, you get to a stable base of knowledge, but knowledge rose, acknowledge evolves. We learn as a huge economic human species as sigh as an organism, constantly grows and evolves and changes, and then in power that, with the two of artificial intelligence and that's going to keep going growing and growing and may at the assertions that you how previously may need to beef significantly expanded model,

fight all kinds of things, it could be like a living organism vs. The analogy I think we started this conversation with, which is like the solid ground. The the other beautiful experience that we have with our system is when it asks clarifying questions. inadvertently turn out to be emotional to us. So at one point it knew that these worthy named entities who were authorized. To make changes to the knowledge base and so on, and it notice that all of them were people, except for it. Because it was also allowed to- and so it said, am I a person had the white tell it very sadly, no less effective you're, not the moments like that, where it asked questions

that are unintentionally, poignant are are, are are worth treasuring that is powerful, that such a powerful question it has to do with basic control. Who can access the system who can modify it? But that's one those questions like what rights do I, as this is a system. What that's another issue, which is they'll, be a thin envelope time between when we have general s eyes and when Everyone realises that they should have sick, human rights and freedoms and so on. Right now we don't think twice about effectively enslaving

our email systems and our series and our electors and so on, but at some point there will be as deserving of freedom as human beings. Are you? I am very much with you, but it aside absurd I happen to believe that it will happen in our lifetime. That's why think they'll be a narrow envelope of time when we'll keep them, as essentially indentured servants and after which will have to realise that they should have. They should have freedoms that other that we give that we afford to other people and all of them starts with the a system like psych, raising a single question about who can modifies stuff? I think that's how it starts Yes, that's that part of a revolution. What about the other stuff, like

I love and consciousness, and all those kinds of topics do they come up in psych in the knowledge base, of course, so an important part of human knowledge. In fact, it's difficult to understand human behavior in humans. three without understanding human emotions and why people do things stand and how how emotions drive people to to do things, and all all of that is extremely important in getting psyched to understand things sample in coming up with scenarios, so one of the applications that site does one kind of application it does is to generate plausible scenarios of what might happen and in what might happen based on that and what might happen based on that and sullen sue generate this ever expanding sphere. If you will of possible future things too, to worry about or think about and.

In some cases, those are intelligence agencies doing possible terrorist scenarios so that we can defend against a terrorist threat. Before we see the first one. Sometimes they are computer security attacks so that we can actually close loopholes and vulnerabilities before the very first time someone actually expire. It's those and so on. Sometimes they are scenarios in All they know more positive things involving our plans like, for instance, what what's college Should we go to what career should we go into and saw what profess? an old training. Should I take on that? That sort of thing, so there there's all sorts of There are all sorts of useful scenarios that can be generated that way of cause and effect and cause and effect they go out

and many of the linkages in those scenarios, many the steps involve understanding and reasoning about human motivations, human needs, human emotions. What people are likely to react? Two in in something that you do and why and how and so on, so that was always a very important part of the knowledge that we had to represent in the system, so I talk a lot about love, so I gotta ask you a member of the top of your head. How sake is trying to is able to represent their is ass books of love, their useful for understanding human nature and therefore integrating into this whole knowledge base of common sense. What is love we try to tease apart cancer, That have enormous complexities to them. And variety to them

down to the level where, where you don't as it were, you dont need to teach them apart further, so loves general of a term its not useful. Exactly so when you get down to a romantic love and sexual attraction, you get down, Parental love? You get down to have filial love and are you get down to a love of doing a kind of activity or creating. So eventually you get down to maybe fifty or sixty concepts each of which is a kind of love. Their interview, they did and then each one of them has idiosyncratic things about it and you dont have to deal with love to get to that level. Of complexity, even something like in X.

being in. Why meaning physically in? Why that we may have one english word in to represent that, but its useful to tease that apart, because the way that the the liquid is in the coffee cup is different from the way that. The areas in the room which is different from the way that I'm in my jacket and so on, answer their questions like. If I look at this cost, cup. While I see the liquid, if I turn it on I do liquid come out and so on. If I have say coffee with sugar in it How do the same thing? The sugar doesn't come out right and stays in the liquid cause. It's dissolved in the liquid in Seoul, so by now, we have about seventy five different kinds of in in the system and its importance to distinguish them. so if you're reading along an english text, you see the word in the right,

of that was able to use this one innocuous word, because here she was able to assume that the reader had enough, common sense and world knowledge to december regulate which of these seventy five kinds of in actually met and the same thing with love. You may see the word love, but if I say you know, I love ice cream, that's obviously different than. If I say I love this person nor I love to fishing or something like that. So you have to be careful not to take language to seriously because people have done a kind of via parsimony. A kind of terseness where you have a few words is as you as you can, because otherwise you'd need half a million words in your language, which has a lot of words like tat. times more than most languages really make use of, and so I'm just we have on the order of

bout, a million concepts in psych, because we ve had to tease apart all these things, and so When you look at the name of a site term, most of the site terms actually have three or four english words in a phrase which captures the meaning of this term, because you have to distinguish all these types of love. To distinguish all these types of in, and there is not a single english word which captures most of these things and It seems like language when used for communication between humans. Almost as a feature has some ambiguity built in it's not so as an accident because, like the human condition, is a giant mess, and so it feels like nobody wants to robots, like very precise, formal logic, conversation on a first date right

like there there's some dance of like uncertainty of word, of humor, of push and pull and all that kind of stuff. If everything is made precise, then life is not worth living. I think for in terms of the the human experience and we've all had this expert, science of creatively misunderstanding one of one of my favorite. What am I fear it stories involving marvin minsky is when I asked him about how he was well to turn out so many fantastic phds. So many fantastic people who did great this phd thesis. How did you think of all these great ideas? What he said is he would just They say something they didn't exactly makes it he did you really know what it meant, but this student would figure like. Oh my,

minsk. He said there must be a great idea here, sweat here. She would work on work and work until they found some meaning in this sort of chauncey gardiner, like utterance that may not be meaningless and then some great thesis would come out of it yeah. I love this so much that there is a young people come up to me and I am think they made aware that the words I say have a long, lasting impact. I will now doing the minsk group method of saying something cryptically profound and then letting them actually make something you. One great out of their effect. You have to become revered enough. Now. People will take as a default that everything you say and profound as yesterday that would mean a lot more risky. So much is so much of her this interview with him when he said that the key to success,

to hate everything's ever done like in the past years, when you get leg one liners and just the are also to work on things that nobody else is working on, because he's not very good at doing stuff. Ah, I think that was just false or pussy I took He said, and I ran with it, and I thought was profound because, as my room is gained, a lot of behaviour is in the eye of the beholder a lot of the meaning is in the eye. The beholder one of minsk is early programmes was begging programme racially with us, so This is back in the day when you had job control cards, I'm at the beginning of your ibm card deck. That said things like how many sea few seconds to allow this to run before it got, kicked off and on because can Wintertime was enormously expensive, and so he

the program and all it did was. It said you know, give me a thirty seconds of cpu time and all it did was. It would wait like twenty seconds and then it would print out on the operator's console. ella type I need another twenty seconds at the opera. There would give it another twenty seconds with waited says? I'm almost done they may be a little bit more time so in the end. You get this print out and he be charge. You know for like ten times as much computer tide of his job control. Court now need say: look I put ten seconds thirty seconds here and I'm your charging me for five minutes. I'm not gonna pay for this in the poor operator would say well the program kept asking for more time and marvin would say: oh it always does that. I love that is. Is there if you could just linger in it for a little bit? Is there something you've learned? from the interactions, mom is key about artificial intelligence, about life. About them in his white again, I, like your work,

his work is, you know he's a seminal. figure in the very short history of artificial intelligence, research and development, What have you learned from him as a human being, as as an ally intellect I would say, both he and advise and mom impressed on me the The realisation that our law as your finite, our research lives are finite. We're going have limited opportunities to do a research projects, so you should make each one count, dont, be afraid of doing a project. That's gonna, take years or even decades to and don't settle for bump on a log projects. that could lead to some no published journal article- that five people will

need and pat you on the head, foreign and so on so one on a log after another, is not how you get from the earth to the moon by slowly putting additional bumps on this log. The only way to get there is to think about the hard problems and think about novel solutions to them, and if you do that, and if you're willing to listen to nature, to empirical reality willing to be wrong. It's perfect, fine, because if occasionally you're right, then you ve gotten part, the way to the moon. You know you worked on site for Thirty, seven over the over that many years have you ever considered quitting I mean it has been too much I'm sure there's than optimism in the early days. It is going to be way easier and.

Ask it another way too, because I've talked to a few people on this podcast a I folks that bring up sake is an example of a project that has a beautiful vision and is it the beautiful dream, but it never really materialised as how spoken about I suppose you could say the same to you on your networks at all, all all ideas. Until they are so what? Why do you think people say that first of all and Second of all, did you feel that everything your journey and deserve a consider quitting on this mission. We keep a very low profile. We don't at ten Very many conferences. We don't give talks, we don't write papers, we don't play the academic game at all. and, as a result on people and only know about us because of a paper. We wrote tenor

twenty or thirty or thirty seven years ago, They only know about us because of what someone else. Can hand or third hand said about us, so thank you for doing this park has by the door it it shines a little bit of white on some of the fascinating stuff. You're doing what I think. It's time for us to keep a higher profile. Now that we are far enough along that, other people can begin to help us with the. The final n per cent, maybe end is maybe ninety percent, but now that we ve gotten this knowledge pump primed. It's gonna become very important for everyone to help if they are willing to they're interested in Retirees who have enormous amounts of time and would like to leave some kind legacy to the world people because of the

in who have more time at home for one reason or another on to be online and contribute if we can reach his awareness of how far our project has come and how close to being primed the knowledge from is. Then we can begin to harness this untapped amount of humanity. I'm not really that concerned about professional colleagues, opinions of our project. I mean interested in getting as many people in the world as possible, actively helping and contributing to get from where we are to really covering all of human knowledge and different human opinion, including contrasting opinion. That's that's worth representing, so I think that's that's one reason a I don't think. There's there was ever a time where I thought about quitting. There are two

times where I've become depressed a little bit about how hard it is to get funding for this damn. Occasionally they are. I winters and things like that. Occasionally there are a I what you might call summers where people have said why in the world, didn't you sell your company to you? No company acts for on some. Large amount of money when you had the opportunity and so on, and no company acts here, like old companies, maybe you ve never even heard of like lycos, or something like that so well at that. The answer is that one reason we've stayed a private company. We haven't gone public. One reason that we haven't gone out of our way to take investment dollars is because we want to have control over our future at over our state of being so that we can continue to do this as until it's done,

and we're making progress and were now so close to done that almost all of our work is commercial applications of our technology, so five years ago, almost all of our money came from the government. Now have virtually none of it comes from government, almost all of it, is from companies that are actually using it for something hospital chains using it. For me, The reasoning about patients and energy companies using it and various other men manufacturers using its region about supply chains, and things like that, so there's some questions I ass, so one of the ways that people can help is by, to the knowledge base, and that's really. Basically anybody if the toys is right, I do the other way. I kind ask you by your thoughts on this eve. Had no, you said in government and you'd big clients a lot of clients, but most of it is shrouded in secrecy, because

the nature of the relationship of the kind of things you're, helping them with that's one way to operate as another way to operate is more in the open, where it's more consumer facing and so are you not? Hence something like open. Psych was born at some point or there's. No that that's a misconception. What this does go, there's a what are now what is open it can. How was the born two things I wanted to say and I want to say each of them before the other, so it can be difficult, but we'll come back to open sake in a minute, but one of the in terms of our contracts with all of our customers and partners is. knowledge you have, that is genuinely proprietary to you. We will respect that will make sure that its marked as proprietary to you in the site knowledge base, no one other than you will be.

what to see it. If you don't want them to and it won't be used in inferences other than for you and sought, however, any now knowledge, which is necessary in building any applications for you and with you, which is publicly available. General human knowledge is not going to be proprietary. It's going to just become part of the normal psych knowledge base and it will be openly available to everyone who has access to psych. So that's it important constraint that we never went back on even when we got push back from companies which we often did who wanted to that almost everything they were telling us was proprietary. So So there is a line between very domain specific companies, pacific stuff, and the general knowledge that comes from that, yes or, if you imagine, say it's an oil company, there are things which they would expect

any new petroleum engineer. They hired to already know- and it's not okay- for them to consider that that is proprietary and that sometimes a company will say well we're the first ones to pay you to represent that in sake, and our attitude is some polite form tough, the deal is this: take it or leave it, and in a few cases they left it in most cases dealt they'll, see Point of view and take it because that's how we ve built the psych system by essentially tacking with the funding, wins where people would fund a project and half of it would be general knowledge that would stay permanently as part of cycle. So always with these partnerships is now a distraction from the main psych device,

I meant it's a while, since the board of a small distraction is a small, but it's not a complete one, you're adding to the knowledge base. Yes, absolutely, and we try to stay away from on projects that would not have that property on. So let me go back and talk about open psych for a second, so I've had a lot of trouble expressing convincing other ai researchers how important it is to. Used an expressive representation language like we do this higher order lodging, rather than just using some trip store knowledge, graph, type representation and so as an attempt to show them why they needed something more. We said, oh well, we'll represent this unimportant projection or shadow or subset.

Like that, just happens to be the simple binary relations the nation argument, one argument to triples and so on and and you'll see how much more useful it is. If you had the entire likes, it So is all well and good to have that tax ah relations between terms like person and night and sleep in bed and house and eyes, and and so on, think about how much more useful it would be. If you also had all the rules of thumb about those things like people sleep at night, they sleep lying down, they sleep with their eyes closed. They usually sleep in beds in our country, They sleep for hours at a time they can be woken up, they don't like being woken up and so on and sought. So it's that

massive amount of knowledge, which is not part of open sake, and we thought that all the researchers would then meet Lou emil, We say- oh my god, of course we need the other ninety percent that you're not giving us let's partner and licence psych so that we can use it in our research. But instead, but people said is: oh even the bit you released is so much better than anything we had both just make do with this answer if you look there a lot of robotics companies today, for example, which use open psych as their funding. Dental ontology and and in some sense the whole world missed the point of open sake, and we were doing it to show people why that's not really what they wanted and too many people thought somehow that this was sacred, that this was in fact good enough for them and they never even bother coming kept coming to us to get access to the full sake

but there's there's two parts open sykes, a one is convincing people on idea and the power of this general kind of representation of knowledge and vat. Are you that you hold in having acquired that knowledge and built in and continue to build it and the others? The code base the scissors, the code side of it, so my sense of the code base that cycle is operating with I mean it has the technical debt of three decades plus right. This is the exact same problem. The good deals with the earlier version of ten to flow is still doing what that there to basically break compatibility with the past several times and that's only over a period of a couple of years but day I think successfully opened up. It's very risky, very gutsy move to open up tensorflow and then pi torch and the facebook side.

And what you see is there's a magic place where you can find a community. We could develop a community that builds on on the system without taking away a not any, but most of the value. So most of value, the google has still google most valued at facebook. Has the facebook, even though some with this major machine learning tooling is released into the open. My question is not so much in the knowledge, which is also a big part of open psych, but all the different kinds of tooling so the There is the kind of all kinds of stuff you can do on the knowledge graph knowledge base would have. We call it there's the inference engines, so there there could be some. I probably are a bunch of proprietary stuff. You wanna keep secret and there's probably some stuff. You can open up

completely and then let the community build up of community where they develop stuff. On top of it. Yes, there be those publications and academic work on all that kind of stuff, And, and also the tooling of adding to the knowledge base right like developing in a there's incredible ma like there's so many people that are just really good at this kind of stuff in the open source community. So my question is like have you struggled with this kind of a that you have so much value in your company already you ve developed, so many good things. You have clients, really value your relationships and then there's this door giant opensource community, as, as I know, you're not utilizing, is there there's some things to say there, but that there could be magic moments where the community builds up large enough to wear the artificial intelligence field. That is currently ninety nine point: nine percent machine learning

dominated by machine learning, a face shift towards ike, or at least in part, towards more like what you might call symbolically. I this whole place where psych is like at the center and then, as you know, the requires a little leap of faith, because you now surfing and be out. Obviously competitors, though pop up and start making you nervous in that kind of stuff. So do you think about the space of open source in some parts? others how to leverage. The community are all those kinds of things that that's a good question and I think you phrased it the right way. Witches were constantly struggling with the question Of what to open source, what to make public? What to even publicly talk about rights and its

Their enormous pluses and minuses to every alternative and it's very much like negotiating a very treacherous path, part partly the analogy is like. If you slip are, you could make a fatal mistake: give away something which essentially kills you or fail to give away something which are failing to give it way. I'm hurts you and so on. So it is that it is a very tough, tough question. usually what we have done with people who approached us too, collaborate on research is to say we will make available, to you, the entire knowledge base and executed copies of all of the code, but

Only very very limited source code access, if you have some idea for how you might improve something or wherever, thus on something. So let me also get back to one of the very, very, very first things we talked about here, which was separating the question of how could you get a computer to do this at all verses? How could you get a computer to do this efficiently enough in real time, and so one of the early lessons we learned was that we had to separate the episcopal logical problem of what should the system no separate that from the horrific problem of how can the system reason efficiently with what it news, and so, instead of trying to pick

one representation language which was the sweet spot or the best trade off point between expressiveness of the language and efficiency of the language. If you had to pick one knowledge grass would probably be associative triples would probably be about the best you could do and that's why we started there, but on after a few years we realise that what we could do as we could. let this and we could have one nice clean, epistemological level, language, which is this higher order, logic and we could have one or more grubby but efficient jurists Level modules that opportunistic lee would say, I can make progress on what you're trying to do over here. I have a special method that will contribute a little bit towards a solution and the sort of some subs

of Gaza, not actually, so by now, we have over a thousand of these touristic level modules and they function as a kind of community of agents and there's one of them, which is a general fear, improver and theme. In theory, That's the only one you need, but in practice always takes so long that you never want to call on it on you, always these other agents to very efficiently reason through it sort of like. If your balancing a chemical equation, you can go back to first principles, but in fact there are algorithms which are vastly more efficient or for trying to solve a quadratic equation. You could go back to first principles of mathematics, but it's much better to simply recognise that this is a quadratic equation and apply the binomial formula and that you get your answer right away and so on so think of these visas, a thousand little experts that

all looking at everything the site gets asked and looking at everything every other little agent has contributed almost like notes on a blackboard notes, on a await board and making additional notes when they think they can be helpful and gradually that can you the have agents gets an answer to your question gets a solution to your problem and if we ever come up in it, in application where site. is getting the right answer, but taking too long, then what will often do is talk to one of the human experts and say here's this that the set of reasoning steps that psych when, through you can see why it took a long time to get the answer? How is it that you able to answer that question into second

and occasionally you'll get an expert who just says: well, I just know it. I just was able to do it or something, and then you don't talk to them any more, but sometimes gets an expert who says well. Let me into respect on yes here is a special representation. We use just for acquiesced, chemistry, equations or here's, a special oh representation, own especial technique, which we can now apply, two things in this special representation and so on. Then you add that as the thousand and first age, all here, dick level module and from then on, I'm in any applicant and if it ever comes up again, it'll be able to contribute, and so on, so that that's pretty one of the main ways in which psych has recouped this lost efficiency. A second important way is never reasoning, so you can speed things up by

Focusing on removing knowledge from the system till all it has left is like. minimal knowledge needed to. But that's the wrong thing to do right now. Would you like a human extra painting, part of their brain or something that really bad? So, instead, what you want to do is give it nedda level advice tactical and strategic advice that enables it to reach and about what kind of knowledge is going to be relevant to this problem. What kind of tactics are going to be good to take in trying to attack this problem? When is it time to start trying to, prove the negation of this thing, because I'm knocking myself out trying to prove its true and maybe its false, and if I just spend a minute, I can see that its false or something like dynamically pruning the graph. The two only like the basin of particular thing, you're trying to infer yes and so by now? We have about a hundred and fifty of these sort of like

break through ideas that have led to dramatic speed ups in the inference process. Nowhere one of them was this e o, H, l, split and lots of hiv jobs and other one was using mega and met him at a level reasoning. To reason about the reasoning: that's going on, So on, and you know a hundred and fifty breakthroughs may sound like a lot. But now, if you divide by thirty seven years, is not as impressive so there's these cut em you're, a sick modules are really help improve the inference. How hard in general, is this higher order, logic in in

general, the thier improver sense. It's intractable very difficult problem. Yes, so how hard is this inference problem when we're not talking about? If you let go of the perfect and focus on the good, I I would say it's half of the problem in the in the following empirical sense, which is over the years about half of our effort, may be. Forty per cent of our effort has been in our team of inference, programmers and the other, fifty sixty percent has been our analogous aren't. Ontological engineers pudding in knowledge, so are ontological years. In most cases, don't even know how to program. I'd have degrees in things like philosophy and so on, and so is almost I love the other thing out. Those people are obvious, but it is very much like the eloi and the morlocks lsd wells time machine. So you had the eloi who only programme

the episcopal logical, higher order, logic language, yes, you have the morlocks who are like I'm under the under the ground figuring out what the machinery is, that will make the sufficiently operate and so on, and so you don't occasionally they all toss messages back to each other. and so on, but it really is almost this. Fifty fifty split between finding clever ways to recoup efficiency when you have an express a language and putting in the content of what the system needs to know, and they are both a friend saving it to some degree. The entirety of the system, as far as I understand, is written in various variants of lisp, so my fair programme languages still

I don't program in it in much any more because you know the world has in majority of its system has moved on. Like everybody respects lists, but many of the systems are not written in lisp anymore, but psych. As far as I understand, maybe you can correct me: there's a bunch of lisp in it so it's based on a list code that we produced most of the programming is still going on in a dialect of lisbon and then the for efficiency reasons gets automatically translated into things like java or sea, nowadays it's almost all translated into java cause java has gotten good enough that that's that's really all we need still says translate into java, and then java is compiled out to buy code. Yes, okay, so that that's sort of that's a that that that's a you know, it's a process that, probably

it's to do with the fact that when sake was originally written and you build up a powerful system like there is some technical doubt the ought to deal with, as is the case with most powerful systems the span years and have you ever considered that this? This would help me understand because, from my perspective, so much of the value of everything you've done with psych and psychopath. Then, as the is the knowledge have you ever considered just like throng way the code base, starting from scratch that really throwing away but Sarah moving it too, like route through we away that technical debt starting with them more data programming language, that drawing well out of value or no look. What's your sense, how much of the value is? silly software engineering aspect,

so the values and the knowledge, so development of of programmes in lisp, I'm proceeds, I think some are between two thousand and fifty thousand times faster than developed and in any any of what you are calling. modern or improved computer like well there's other functional ngos, closure, others there. It is, I mean I'm with you, I I like blessed by just one, the humming great programme, there are there still like? Yes, so It is true. When a new inference programmer comes on board, they need to learn some of less than the fact we have a subset of lisp, which we call cleverly saddle which is really all they need to learn? And so the programming actually goes on in seville, not in full lisp, and so it does not take

grammars very long at all to learn seville and that something which can and be translated efficiently into a job and for some our programmers who are doing A user interface work than they never have to even learn saw bell. They just have to learn a p. I into the the basic psych engine c unnecessary feeling the burden of like it's, it's extremely efficient. Is that's not a problem to solve? Okay right right. The other thing is wrong. were that we're talking about hiring programmers to do inference, who are programmers interested in effectively automatic thier, improving, and so those are people already predisposed to representing things in logic, and go on and list really was the programming wide language on based on logic that John, dorothy and others who developed it basically create, took the the formalism,

alonzo, church and other philosophers other locations, I'm had come up with an basically said. Can we base? we make a programming lap language which is effectively logic and so since we are talking about reasoning in about expressions written in this logical, pissed, logical language and we're doing operations which are effectively like fear, improving, if operations and so on theirs natural impedance match between lisp and the knowledge the waves represented. So I guess you could say it's a perfectly logical languish to use. Oh yes, after all. Even let you get away with that. I find that thing. I'd really like that. It's a problem that in the future, without without credit that credit, but now I think I think the idea that the point is that the the language

programme in isn't really that important. It's more that you have to be able to think in terms of, for instance, creating new helpful h all modules and how they work with each other, and I'm looking at things that art in a long time, in coming up with new specialized data structures that will make this efficient, so limited Give you one very simple example, which is when you have a transit of really nation like larger. Then this is larger than that which is larger than that which is larger than that. So the first thing must be larger than the the last, whenever you have a transit of relation, if you're, not careful, if I ask whether this thing over here is large thing over here, I'll have to do some kind of graph, walk or fear improving that might involve like five or ten or twenty or thirty steps. But if you store redundant, lee store the transit of closure, that cleaning star of that transit released

And now you have this big table, but you can always guarantee that in one single step you can just look up whether this is larger than that and so on. We there are lots of cases where storage is cheap. Day, and so by having this extra redundant data structure, we can an this commonly occurring type of question very, very efficiently. Let me give you one other analogy: and a log of that, which is something we. I'll rule macro, predicate switches, we'll see this complicated rule and will notice that things very much like it. Syntactical e come up again and again and again so will come

eight a whole brand new relation or predicate or function that captures that and takes. Maybe not two arguments takes maybe three four or five arguments and so on, and now we have effectively converted some complicated. If then, rule that might have to have inference done on it into some ground atomic formula, which is just day The name of a relation and a few arguments, and so on and so converting commonly occurring types or scheme is of rules in to brand new predicate brand new functions turns out to enormously speed up the inference process so rip. So now we ve covered about four of the hundred and fifty on good ideas. I said that the nice difficult, though that idea and protect us like a nice compression that the terms

it's really useful yeah! That's really interesting! I mean this whole thing is just fascinating from a philosophical. This pardon me, I mean it makes me a little bit sad because your work is both from a computer science, perspective, fascinating and the inference engine from across the philosophical aspect, fascinating But you know it is also you running accompany and there's some stuff. There has to remain private and said what you hear something it may make. You feel better a little bit better where are we formed a not not for profit company called the now well jack civilization, institute, knacks k in a accent- and I have this firm belief with a lot of empirical evidence to support it- that the the education that people get in high schools and colleges and graduate schools, and so on.

is almost completely orthogonal to almost completely irrelevant to how good they're going to be at com. Up to speed in doing this kind of Angela go engineering and writing these assertions and rules and so on in in sight and so very often we all interview candidates who have their phd in philosophy, who ve taught logic for years and so on and they're just there just awful. But the converse is true. So one of the best ontological engineers we ever had never graduated high school, and so the purpose of knowledge act immunization institute. If we can get some some foundations to help support, it is identify. People in the general population may be high school dropout, who have layton talent for this sort of thing offer them

effectively: scholarships to train them and then help placed them in companies that need more trained, ontological engineers, some of which would be working for us, but mostly would be working for partners or customers or something. And if we could do That would create an enormous number of relatively very high, paying jobs for people who currently have? No? no way out of some. You know on situation that their locked into so is there some day you can put into words that describes somebody who the great and ontological engineering. So what characteristics bought a person make them a at this task. This task of converting the machinery of human language knowledge into formal logic is very much

what else touring had to do during row were too in trying to find people to bring to blush park where he would publish in the london times cryptic crossword puzzles alive. with some some innocuous. Looking note, which essentially said if, if you are able to solve this puzzle in less than fifteen minutes, call this phone number and so on. So you know, or back when I was young, there was the practice of having a match books where, on the inside of the match book, there would be a. Can you draw this, you have career in art, a commercial art? If you can copy this sad drawing none? So so I guess the the analogue that their little test, the get to the the core of a whether it be good or not so part of it has to do with our being able to make and appreciate and and react negatively appropriately to puns,

and other joke say you have to have a kind of sense of humour and if you're good at ten joke. Send I'm good at understanding, jokes! That's that's! Why An indicator runs yeah. Dad jokes! Yes! Well, maybe not dad jokes, but real, but funny jokes, but they come applying to all. is that what they are but another another is if you're able to interest back so very often I will we will give someone a simple question and will say like what what why why is this and sometimes not just say, because it is it's a bad sign, but very often they'll be able to interest back to, and so on. So one of the questions, I often ask as all point to a sentence for the pronoun in it now say you know it referenda that pronouns, obviously this now over here. You know how?

would you or I, or in a I or a five year old, ten year old child, no that that pronoun refers to that now over here and are often on the people who are going to be good at ontological engineering will give me some causal explanation or will refer to things that are true in the world, so if you imagine, a sentence like the horse was led into the barn while its head was still wet, and so its head refers to the horse his head, but how do you know that, and so some people will say I just know it- some people will say well. The horse was the subject of this once and I'll say: okay! Well, what about the horse was led into the barn while its roof was still wet now its roof obviously refers to the barn, and so then they'll say

Oh that's, because it's the closest noun and sell it. So, basically, if they try to give me answers which are based on syntax and grammar and saw that's a really bad sign, but if we are able to save things like well, horses have heads and barring stone and barns have roofs and horses. Don't then that's a positive sign that they're going to be good at this because they can interests backed on what's true in the world. That leads you to know certain Things are fascinating. Is it that getting a phd makes you less capable to interest bag deeply about them? I'm the I wouldn't I wouldn't for that, far saying that it makes you less capable, let's just say, its independent, I don't know about how of how good a lawyer or not saying that I am saying that there is a certain cut. Is its existing that for a lot of people pages sorry philosophies aside, sometimes education narrows you're thinking, forces expands, yes, it's kind of fast ay and for certain, when your turn

and to do ontological engineering, which is essentially teach our future. I overlords hunted reason deeply about this world how to understand it that that requires that you think deeply about the world saw I'll. Tell you a sad story about math craft. Which is. Why is that not widely used in schools today? We're not really trying to make big profit on it or anything like that that whenever we ve gone to schools, their attitude has been Well if a student spends twenty hours going through this math craft program from start to end and so on, and will it improve their score on this standardized test? More than the they spent twenty hours just doing mindless drills of problem after problem after problem, and the answer is well known,

but it will increase their understanding more and their attitude has been well. If it doesn't increase their score on this task on then that's not it we're not going to adopt. It said I mean that's, that's a whole that the whole another three for our conversation about education system. But let me ask you: let me go super philosophical as if we weren't already in nineteen, fifty, our touring wrote the paper that formulated the turing test. Yes and he opened the paper with the question. Can machines think so? What you think can machines think, and I ask you this question of absolute me- miss teens can think, certainly as well as humans can think right. We're meet machines just because they are not currently made out of meat. Is just yet you know an engineering solution, not decision and so on. So of course, of course, machines can think

I think that there was a lot of damage done by people on this. understanding, a turing's imitation game and- and I the focus on trying to trying to get a chat bot to a fool other people into thinking, it was human and so on that- that's that not a terrible test in and of itself, but it shouldn't be your one and only test for intelligence city, you in terms of tests of intelligence below the alumni prize, which is a very kind of you, gonna, say more strict formulation, the turing test as a regionally formulated and then there's something like elects a prize which is more, I would say, more interesting formulation of the test, which is like

Ultimately, the metric is: how long does a human wanna talk to the hazards them? Professor development goes, you wanted to be. Twenty minutes is basically not just have a convincing conversation, but more like a compelling. One or a fun one or each thing one, and that that seems like more to the spirit may be of value, of what the turing was imagining. But what for you, do you think in the space of tests is up as a good test that good, when you see a system based on psych that passes that test you'd, be I damn. We ve created something special here that the test has to be something involving depth of reasoning and recursive. Nyssa reasoning, the ability.

to answer repeated why questions about the answer you just gave is how many why questions than rocky keep ass certain such something like that, and also have like a young curious child, any eyes system, and how long way as system at last before it was too quick and again, that's not the only tat it another one has to do with argumentation. In other words, here's a proposition come up with. Pro and con arguments for it and try and give me convincing arguments on both sides and so that's that's another important kind of ability that the system needs to be able to exhibit in order to really be intelligent. I think so that their sir

if you look at ibm watson and like certain impressive accomplishments? First very specific test was a good demo right. There is some like other, I talked to the guy, who led the the jeopardy effort and there's some kind of hard coding, heuristics or tricks that you tried to pull it all together to make the thing work in the end. For this thing That seems to be one of the lessons with the eye is, I think, that's the fastest where'd you get a solution as pretty damn impressive, so so here here's. What I would say is that as impressive as that was it made some mistakes, but, more importantly,

Many of the mistakes it made where mistakes which no human, would have on aid and so part of the the new or augmented during tests, have to be, and the mistakes you make are ones which humans don't basically look at and say what I so. For example, there was a question about. which sixteenth century italian politician blah blah, blah and watson said ronald reagan. So most americans would have gotten that question wrong, but they never have said ronald reagan as an answer, because you know the things they know? Who is that he lived relatively recently in people don't really live four hundred years, and now things like that, so that that's, I think, I'm a very important thing which is its make.

mistakes which no normal same human would have made. Then that's a really add sign and if its not making those kinds of mistakes than that's a good sign, I don't think it's any one very, very simple test. I think it's all of the things you and she and all the things I mentioned is really a battery of tests which, together, if it passes almost all of these tests, I'd be hard to argue that it's not intelligent and if it fails some several of these. plus it's really hard to argue that it really understands what it's doing, that it really is generally intelligent, said to pass all those tests in old It's a lot about sake and knowledge and reasoning. Do you think this as So we need to have some others Unlike elements, for example, a body or physical manifestation in this world and

another one which seems to be fundamental to the human experience, is cautiousness these subjective experience of what it's like to actually be. You do you think he needs those to be able to pass all those thirst and achieve general intelligent if the good I think in the case of a body I know I know there are a lot of people are like penrose, who would have disagreed with me and so on there and and others, but No, I don't think it needs to have a body in order to be intelligent. I think that it needs to be able to talk about having a body and having sensations and having emotions and so, and it doesn't actually have to have all of that. But it has to understand it in the same way that Helen Keller was perfectly intelligent, enabled to tie.

About colours and sounds and shapes and and so on on, even though she didn't directly experience all the same things that are the rest of us, do so knowledge of it and being able to correctly make use of them. It is certainly an important facility, but actually having a body I'm if you believe that that's just a kind of religious or mystical leave you can't really I'm argue for or against it. I suppose it's it's just something that some people that some people believe what about the like an extension of the body which is consciousness me like it feels like something to be here sure, but you know what it. What does that really mean it's like? Well, if I talk you say things which make me believe that you are conscious the arm. I know that I am conscious that that's you know you just taking my word for it now, but

in the same sense site is conscious in that same sense already, where of course, it understands its a computer. program it understands where in wet when it's running it understands he's talking to it. It understands what its task is, what its goals are, what its current problem is that its working on it understands how long it's spent on things, what it's tried and understands what it's done in the past and so on on, and I you know if, if we want to call that consciousness, then yes, psych has already cant is, but I dont think that I would describe anything, I'm mystical too that again some people would. But I would, I would say that your other other than our own personal experience of consciousness. We're just treating everyone in the world so to speak at their heard about being conscious, and so, if, if a complete program, if in a I is able to exhibit all same kinds of response, as you would

spect of a conscious entity on them. I'm you know doesn't doesn't to serve the label of consciousness. Just as much so that their another burden than comes of this whole intelligence thing that humans got? Is the x languishing of the light of consciousness, which is kind of realizing that we're gonna be dead some day? and others. A bunch of philosophers like earnest becker, who kind of think that this realisation of mortality and then fear, sometimes they call it terror of. Of of mortality is one of the creative forces behind human condition like it's. The thing that drives us do think it's important for an ai system. You know one psych propose that it's one

it's not human and is one of them moderators of its contents in others. Another question it could ask which is like I kind of nausea humans are mortal am I mortal I think one really important thing: that's possible when you cautious, is to fear the extinguishing of that consciousness. The fear mortality. Do you think, that's useful for Intelligence, thinking like I might die, and I really don't want to die. I I don't think so. I think it may help some humans to be better people. It may help some humans to be more creative and so on. I dont think its necessary for a eyes to believe that they have it lifespans and therefore they should make the most of their behaviour. Maybe eventually the answer that in my answer to that will change but

As of now, I would say that that's almost like a a frill or a side effect of that on is not, in fact, if you look at most humans, most humans ignore the fact that they're going to die most of the time so well, but that's like this goes to white space between the words. So what earnest Becker argues is that ignoring, is we're living an illusion that were constructed on the foundation of this terror. So were escaped life, as we know it pursuing things creating things love everything we can think of this beautiful by humanity is, is just trying to escape this realization or going to die one day. That's his that's his idea, and I think I dunno. If I I one hundred percent believe in this, but there's it certainly rhymes. It seems like to me, like it rhymes, with the truth yeah. I I think that for some people

that can be a more powerful factor than others. really dog is talking about russians ev. I think there are Some russians. Clearly it in what all of russian literature at any. I doesn't have to have. Fear of death as a motivating for in that we can build in motivation, so we can be in the motivation of. Obeying users and making users happy and making others happy and and so on, and that can substitute for this sort of personal fear of death that sometimes leads to Of creativity in in humans had our now, I think I like,

I think, hey. I really need to understand death deeply in order to be able to drive a car, for example. I I think there's just summed like that now I really just great. I think it needs to understand the value of human life, especially the value of human life to other humans. The on and understand that certain things are more important than other things. It has to have a lot of knowledge about ethics and, morality and so on with some of it, is so messy this impossible to encode, for example. This is it agree, so the person dying rainfall us most. Human beings would help that person, but they would not apply that same ethics to everybody else in the world. This is the tragedy of how difficult it is to be a doctor, They know when they help a dying child. They know that the money there spending on this child cannot possibly

be spent on every other child. This dying and that's that's a very difficult in co decision may perhaps perhaps it is perhaps it could be formalised. Oh, but I mean you you're talking about a tunnel vehicles right, so a ton of his vehicles are going to have to make those decisions on all the time of what is the chance of this bad that happening on how bad is that compared to this chance of that dead event happening and so on? and no when and potential acts and is about to happen. Is it worth taking this risk? If I have to make a choice which of these two cars am I going to hit and why and sales thing about a very different choice for I'm talking with your mortality which is just observing manhattan style driving. I think that humans, as effective driver needs to threaten she. Pedestrians lies a lot, there's a dance

I of wash predestined a lot of. I worked on this problem. Seems like the the. If I could summarize, the problem of a pedestrian crossing is The car with this movement is saying: I'm going to kill you and the pedestrian saying, maybe and then they decide to say. No, I don't think you, you have the guts to kill me and you walk in. They walk in front and they look away and there's that dance, the the the the pedestrian as the social contract, that the pedestrian trust that once they're in far the car, the car is sufficiently from a physics perspective able to stop they're going to stop, but the car also has to threaten that pedestrians, like I'm late for work, so you're being kind of an asshole by crossing in front of me, but life and death is in like is part of the calculation here and it's that's. That equation is being solved millions of times a day, yes very effectively that game theory, whatever

yet over their formulation is looming. I just I don't know if it's as simple as some formalize will gain through project it get very a b and it gets a driving and in the case of most of us human society, I don't know, but yeah. You might be right that serve the fear of death is just one of the works of like the way our brains have evolved, but not the stuff. The same feature of a of intelligence drivers. Certainly are always doing this kind of estimate, even if its unconscious subconscious of What are the chances of various bad outcomes happening like, for instance, if I don't wait for this pedestrian or something like that, and what is the downside to me going to be in terms of and no time wasted talking to the police, or you know getting sent to jail. Horrida things like that, and so it is

Emotion like people in their cars than to get ashley angle right, that's that's dangerous, but think think about. This is all part of why I think that autonomous vehicles truly autonomous vehicles are free they're out then than most people do, because there is this enough I miss level of complexity which goes beyond mechanically controlling the car and might I can see the autonomous vehicles as a kind of metaphorically literal accident waiting to happen and not just because of their overall in hurrying, verses, preventing accidents and so on, but just because of the almost voracious appetite. People have for bad bad stories about powerful company

and powerful entities when, when I was at eight coincidentally japanese fifth generation computing system conference in nineteen? Eighty seven, while I happen to be There. There was a worker and an auto plant who was despondent and committed suicide by climbing, under the safety chains and so on, getting stamp to death by machine and instead of being a small story that said despondent worker commit suicide. It was front page news that effectively said robot kills worker, because the public is just waiting for stories about like a, I kills phone, a genetic them. the five type stories, and even if you could show that nation wide this system saved more lives than it com stan saved, more injuries, prevented more injuries that it caused and so on the meat.

The public. The government is just coiled and ready to pounce on stories where, in fact, it failed, even if there are relatively few yet so fast ain't watch us humans, resisting the cutting edge of science and technology. And almost like hoping for it to fail across the india. This just happens over and over and over throughout history, or even if we're not hoping for two fell. We're we're fascinated by it and in terms of what we find interesting the war. in a thousand failures, much more interesting than the nine hundred and ninety nine boring successes. So once build in asia system say cycles some part of some part of it. and say very possible. They would be one of the first people they can

sit down the room. I say with her and have a conversation. What would you ask her? Oh you talk about looking at all of the content out there on the web and so on what are whatever some possible solutions to the problems that the world has, that. People have a really thought of before that are not being properly or at least, to adequately pursued what are some novel solutions that you can think of that we haven't that might work, and that might be worth considering. So that is a damn good question, given that the age is going to be somewhat different from human intelligence, its

we're going to make some mistakes that we wouldn't make. But it's also possibly going to notice some blind spots we have and- and I would I would love it as a test or visit really on a par with our intelligences. Can it help spot some of the blind spots that we have said the two part question of. Can you help identify? What are the big problems in the world and to what our son novel solutions to those problems that are not being talked about by anyone and some others may become in feasible or reprehensible or something, but some of them might be actually great things to look at. No, if you, if you go back and look at some of the most powerful discoveries that have been made, like relativity and on superconductivity, and so on. A lot of them were cases where someone

took seriously the idea that there might actually be a an obvious answer to it to a question so an einstein case, it was on god the lorens transformation is known Nobody believes that is actually the way reality works. What if it were the real way that reality actually worked so in a lot of people, don't realize he didn't actually work out that equation. He just sort of took it seriously on or in the case of soup, conductivity you had this v equals. I are equation where arias resistance and so on and it, was being mapped at lower and lower temperatures, but everyone thought that was just bump on a log research to show that equals I are always held, and then, when some graduate student got to a slightly lower temperature and showed that resistance suddenly dropped off, everyone just assumed they did it wrong having, and it was only a little while later that they realized it was on. It was actually

new phenomenon, or in the case of the h, pylori lorry bacteria causing stomach ulcers, where everyone thought that stress and stomach said caused ulcers and when a doctor in australia and a it was actually a bacterial infection we can get anyone seriously to listen to him and he had to ultimately inject himself with the bacteria to show he suddenly developed a life threatening all sir I'm in order to get other doctors to seriously consider that so they're all sorts of things where humans are locked into paradigms. What thomas coon called paradigms and we can't get out of them very easily so, a lot of ai is locked into the deep learning machine. in paradigm right now and and almost

All of us, and almost all sciences are locked into current paradigms and accuse point was pretty much. You have to wait for people to die, and in order for the new generation to escape those paradigms- and I think that one of the things that would change that sad reality as if we had trusted a g eyes that could help take a step back and question some of the pair the times that were currently locked into yeah, it would accelerate the paradigm shifts, and yet in human science and progress, you've load the very interesting life you thought about big ideas. He stuck with them. Can you give advice the young people today, somebody in high school

somebody undergrad about career about life. I'd say you can make a difference, but in order to make a difference, you gonna have to have the courage to follow through with ideas which other people might not immediately understand or support. You have to realise that if you make some some plan, that's going to take an extended period of time to carry out, don't be afraid of that that's true of physical training of your body, that's true of learning some profession, that's also true of innovation that some innovations are not great ideas. Uk

right down on a napkin and become an instant success. If you turn out right. Some of them are paths you have to follow, but remember that your mortal remember that you have a limited number of decades sized debts to make with your life, and you should make each one of them. Can and that's true in personal relationships. That's true in career choice, that's true in making discoveries and so on, and if you follow the path of least resistance, you'll find that your optimizing for I'm sure periods of time, and before you know it, you turn around and long periods of time have gone by without you ever really making a difference in the world in others it when you look I need to feel that I really love is artificial intelligence and does not many projects there's not many little flames of hope that bank

It out for many years, for decades, in sight represents one of them and I mean that in itself is just a really inspiring thing. So I'm I am deeply grateful that you would be carrying that flame for so many years and I think that's an inspiration to young people. That said you said life is finite and would talk about mortality is a feature Regina. I do you think about you. Majority? Are you afraid of death? I'm sure I'd be crazy if I weren't and, as I get older, I'm now over seventy, so us get older hits. More of my mind, especially as acquaintances friends, and especially mentors one by one are dying, so I can't avoid thinking about mortality, and I think that the the good news from the point of view and the rest of the world is that

adds impetus to my need to succeed in a small number of years in the future, because I know it s line exactly I'm not going to have another thirty seven years to continue working on this. We really do want psych too make an impact in the world. commercially physically metaphysically in the next small number of years to three five years, not two three five decades anymore and this is really driving me toward this. This sort of commercialization and increasing increasingly widespread application of psych, whereas before I felt I could just sir to sit back rule my eyes. We tell the world caught up, and now don't feel that way anyway, anymore. I feel like I need to put in some effort to make the world aware of what we have and what it can do, and the good news from your point of view is that that's

It's why I'm getting here and you gotta be more productive benefits. I love it helping any way. I would love to hear from a from a you know from a programmer perspective, and I love especially these days just contributing in small and big ways. So, if there's any when sourcing from an mit side and the research out, I would love to help. But when you no bigger than psych, like I said, is that little flame the year carrying of artificial intelligence, the big dream is there. What do you hope your legacy is that's a good question that people think of me as one of the poor in years or inventors of. the ai that is ubiquitous and that they take for granted and so on much much the way that today we look back on the

the pioneers of electricity or the pioneers of similar types of technologies and so on. As you know, it's hard. Imagine what life would be like if these people hadn't done what they on, what they did, so that that's one thing: that I'd like to be remembered as another. Is that so creator. One of the original originators of this gigantic knowledge store and acquisition system that is likely to be at the centre. Of whatever this future, I think will look out exactly, and I don't

I like to be remembered as someone who wasn't afraid to spend several decades on a project in a time when I all, when almost all of the other forces, institutional forces and commercial forces are incentive, p, all to go for short term rewards and a lot of people gave up a lot of people that draft the same. as you gave up yes, aid in yes, dug its it's truly an honour. This is a long time coming. I am a lot of people bring up. Your work are specific We and more broadly philosophically of this is the dream of artificial intelligence, is likely a part of the future or so

focused on machine learning applications all that kind of stuff today, but it seems like the ideas that site carries forward, is something that will be at the centre of this problem, though all trying to solve, which is the problem of intelligence. emotional and Otherwise, some thank you. So much is said you huge honor that you would talk to me and spend your valuable time with me today. Extra talking thanks lex. It's been great thanks for listening to this conversation with doug planet to support the spa guests. Please check out our sponsors in the description and now let me leave you some words from mark twain about the nature of truth. If you tell the truth, you don't have to remember a Thank you for listening. I hope to see you next time,