 Ddweud amser, mae'n gweithio'n gweld y ddweud sy'n gweithio'n hynny, ychydig y ddod yn y ddweud y ddweud yma yw'r iawn. Mae'n gweithio'n ddweud, mae'n gweithio'n ddweud, a ddweud yma, mae'n gweld y technaud. Yn y gofyn, dyma'n Ymgylchedd Aberbryd Cymru, a'n ymgylchedd MC i'w ddweud, sy'n tyfu ddim yn ystod i'n gweithio i fan hynny, yn ymdweud yma am y cyfnod. They think that I'm here to keep the conversation going. Those of you who have known them longer than I will know that my role will be irrelevant very briskly. It's an absolute pleasure to welcome Pamela McCordock back to the Carnegie Mellon campus. She is an author of books about the history and philosophical significance of artificial intelligence, the future of engineering, the role of women in technology as well as a number of novels. Pamla a hwnnw yw'n ddiolch yn cyfrannu'r llyfodol ymlaen arnyn i ei ddech chi i yw'r cyfrannu. Mae'n ddweud i ddrech llyfr o'r cyfrannu i'r llyfr. Fe yna'r ffordd o'r ddweud i'r llyfr o'r cyfrannu llyfr yma, lle mae'r llyfr yn gweithio'r llyfr yn cael ei ddech chi'n gwneud ymlaen i hyn. Mae'r llyfr yn cyfrannu o'r cyfrannu a'r dechrau i hyn o'r cyfrannu, ac yn bargod yn credu yma o'r cyfrannu. We have also some wonderful collections that Pamela gifted to us. A collection of pre-digital computers including a couple of enigmata machines. There is a card in your chairs. Promoting the enigmata machines, do come and look at them in our final rare book room on the fourth floor of this building. A couple of weeks ago some of our computer science faculty took the rotors out of the enigmata machines to get the serial numbers, an important part of the historical record. Fascinating to see that there are recordings and stories about that particular exercise on the university's web page. I also keep a kind of journal, it's more a series of scrap notes, but I was able to go back to August 1984, when I was thankfully still a teenager. ac yn fwy o'r ffordd y 8 oed 1984, oedd yn y bwysig o'r Llywodraeth oedd Cymru, ac yn Cary Lewis, o'r ddau, yn y 200m gŵl meddwl. O'r ddau, oeddwn i'n gweithio'r llyfr o'r llyfr, a'r profiad yn ymgyrch gael mewn yma, gafodd yn y bwch iawn iddyn nhw, y 5th gennaeth, cofodd gan Pamla yma yma Edw Fhaginbair, ddechrau'r byd yn y bydd ar y ffordd eich cyfath, ac mae'r ydych chi'n mynd yn y ffordd. Mae'r ffordd i'r ysgrifennu panlwyddon, mae'r ysgrifennu ei ffordd o'r ysgrifennu mi passengersu, yr uned yng Nghymni, o'r ysgrifennu ymlaenau lleolau. As well as pointing to the growing prominence then of Japan in this space and a call on the United States to step up and take a more active role. There's probably a Brexit joke lurking in there. Those of you who have seen me speak recently will know that I'm kind of pained by all of this but I'm today not finding much humorous insight at all. But in Pamela's new book there's a wonderful point where she says accidentally or by design in the early 19th century universities had become a ghostly simulacrum of the British class system. Bellethr at the top, the study of music and painting, history and philosophy just below and so on all the way down to the contemptibly practical like science and engineering considered no better than intellectual shop keeping. And there are numerous references in the book to Pamela's sojourn next door at the University of Pittsburgh and some of her comments about professors of humanities. The word lobotomy appears in there somewhere but I'll leave you to dig that one out. Perhaps one of the most opposite quotes for this evening in the book is a single sentence even in a department of stunning visionaries Raj Reddy stood out. Raj is the most of international university professor of computer science and robotics in the school of computer science here at CMU, an absolute giant in the field of AI winner of the Turing Award and many other distinctions. And we've invited Raj and Pamela who have lived much of the history of this book and Raj was counting up a few minutes ago how many years together you have in artificial intelligence and I'll leave you to unpick that. But we thought rather than having a traditional book launch we might invite Raj and Pamela to reminisce to talk about the early days of AI at Carnegie Mellon and more broadly and we'll just see where it takes us. I'm not going to set a clock too much but I do want to make sure this time for all of you to interact it's very much a family conversation this evening. So I'm going to dismantle my new best friend, pass it to Pamela. Thank you. Take it away. Thank you, thank you Keith. Do we really need to go back that long? No, let me kind of say I'm glad each of you are getting a copy of the book. If you take a few minutes to open the first few pages I promise you won't put it down until you go to the last page. It's 500 pages long and I managed to get through some of it but it'll take me some more time because every time I read a page I say I better go refer to that and see exactly what's happening. So this is a celebration of life and legacy of Pamela and she has been in a unique position in the history of artificial intelligence going back to 1960 when she was the 20-year-old editor of Computers and Thought, which was a classic book by Fagan Baum and Feldman which if you haven't seen it you can probably still get it maybe used book copy of the book or something which has all the great original papers collection going back to Turing's article on intelligence and so on. Many of the papers you will not find anywhere else. Steps towards artificial intelligence and she was the one that did all the legwork and all the homework and collected all the papers and she should have been a co-author of that book. But anyway, since that time, it's amazing that she was in the right place at the right time for the last 60-50 years interacting with all the, in a senior people in AI, starting with Noel and Simon and Minsky and McCarthy and everyone else. That's a unique perspective and also she's blessed with a very creative, articulate mind. She will be able to express all those things in ways that most of us are unable to say. And that not only that, I find her perspective especially about Noel and Simon and McCarthy and Fagan Baum are amazing. There are lots of things that I did not know but I'll tell more about it in a minute. But there are other people here like Mary Shaw and Dan Suwrik and all of us were here when she was the first lady of the computer science department and she would host all these receptions and dinners and so on where we would all kind of have a great time together. And so I'm, you know, it's great to have her write this book and I'm glad ETC Press and CME Libraries have kind of created, helped to bring it forward. And I was asking her when is the book tour, she says that book tours are 20th century, book 21st century, it is online book tours, podcasts and so on. Pamela, you're? Okay. I was an English major at the University of California, Berkeley. And I was working my way through college typing course outlines, yeah, typing course outlines and book lists and things like that. And I was expected to graduate in January and then in the fall I would go to, I thought then law school, mercifully that didn't happen. But in that interim period where I had these nine months to sort of do nothing, this young guy at Fygenbaum came to me and he said, Pamela, we know you're graduating in January and you're going to go to graduate school in the fall, would you like to work on our book in the meanwhile? And I said, oh, yes. And then I said, what's it about? Now, working in the business school, I had just spent an entire summer typing a textbook on accounting. I had no idea what was coming. So Ed said, well, it's about artificial intelligence. Now, I'd heard the term 1960, it's not widely, widely used. And I said, fine, what's artificial intelligence? And he said, this is behaving in such a way that if humans did that, we'd say that's intelligent behavior. Excuse me, please, I've been drinking water. So that led me to this wonderful adventure of putting together, I didn't put it together. No, not an editor. I was told which articles should be in, and it was my job to go all around the campus to, you know, libraries I'd never heard of, departments I'd never heard of, and pick out these particular papers which eventually ended up in computers and thought. And that was my introduction to AI. And somehow that kind of bit me. I mean, I was an English major. I wanted to study literature and all that stuff. And it sort of came about but not immediately that Ed Feigenbaum again went to Stanford and he called me and he said, you know, Stanford has a real computer science department. Why do you come and be my assistant? And I thought, yeah, why not? So that was more of being acculturated into AI. And I spent two, two and a half years as Ed's assistant, really getting, getting to know what AI might be. And because this was very early, we're talking 1965, 1966, and nobody had heard about AI. And I had a wonderful time. And then I thought, okay, enough of this, I'm going to graduate school after all, and I'm going to study writing. So I went to Columbia, where I was in the writing program. And I would always stay in touch with Ed and say, oh, what's new in AI? You know, we would phone back and forth. This was long before the days and you could have, you know, an email. Well, then didn't exist. I mean, that was just for the ARPA people. And he'd tell me and I'd be fascinated. And at that point, my husband came here to Carnegie Mellon to take over the computer science department. It was a department then. And gosh, I got hooked again. And my husband, by the way, was not in AI. I wore these socks in homage to him. He was a mathematician. But we both were fascinated by AI. I kept a journal, as you've heard. Every night, I'd sit down and write what happened during the day. And it turned out that often what happened during the day was tea or sherry, more likely sherry. With Herb Simon. Now, how did Herb and I become sherry pals? We would walk along Northumberland. You'd walk up that street that's just behind the library. And then over to Northumberland and walk along there. And our house was on the corner of Northumberland and Forbes. So I would just be putting my cover on my typewriter. And I'd see his beret or his chuwya in the winter. He liked these Peruvian things that had tassels and whatnot. And I'd lean out the door and I'd say, hey, Herb, would you like a sherry? And you know, Herb would almost always like a sherry. And these were wonderful afternoons. We just had so much fun together. And it wasn't until much later, when my husband's reading one of the first drafts of this book, he said, you were having sherry once a week with Herb Simon? Well, I was working my pedukie off. Oh, on the up and up, which it was. But he was a little jealous. So it so happens. I was a graduate student at Stanford when Pamela was there. So we met each other long before Carnegie Mellon. And so when Joe and she arrived here, I was pleasantly surprised, but was very happy to see her after having lost touch for a few years. And there are two people here who have spent more time with Joe and Pamela. Mary Shaw and Dan Suwarek. Dan was hired by Joe into the department. And Mary Shaw was already here before any of us. And she was just a lowly student, of course. And the amazing thing about Mary is that she and Joe worked on some papers in mathematics and computational theory. And I said, Mary, I thought you were a software engineer. She was working with Alan Perlis, who was mainly in software. But she's universal. She knows lots of things about lots of things. But so we had the pleasure of having Pamela and Joe here for about 10 years until the end of the 70s. And in the book she talks about one of the more recent advances, deep learning and neural network based learning by Jeffrey Hinton. She may not have known because I didn't read it carefully. Jeff Hinton was a faculty member here from 1980 just after you left for five, six years. And not only that, he invented back propagation here, published the paper here when he was a student faculty member here. And then he left for Toronto. And I tried to convince him. I said, why are you going there? It's the backwards. But it turns out Toronto has had a very strong computer science program. As long as I knew, but it was not CMU. So I said, why would you go there? So anyway, there are a number of stories I was kind of reminded of. One of them is about Herb Simon. Herb might have had cherry with her, but he would constantly baffle me. He would say, I never read newspapers. I said, why? Why wouldn't you read this? It's a waste of time. And I said, don't you want to know what's going on? He says no. If it's important, you will tell me. More importantly, his view was he had a famous statement. We said, we have a wealth of information and scarcity of attention. Even today it's the same problem. We have too much data and too little time. And the question is, how do we provide appropriate attention to the right topics at the right time in the right place? And there's a famous Bill of Rights from Jaime Carbonell who's not here. One of the research agendas of LTI is to get the right information to the right people in the right language, in the right time frame, in the right level of detail. And he called it Bill of Rights of Information Age. If you think about it, each one of these is a major research problem in AI, every one of those phrases. And again, it's there in Pamela's book somewhere. But the important thing is we have had amazing luck, being both of us, to be living in this age with these people. I can't imagine any other place I would be if I had the opportunity. Nule is the same. Alan Nule was a giant of a man, but he was very empowering. He and Alan Perlis also, the three people. Alan is not in the book. Maybe he is. No, not Alan Perlis. Alan Perlis. What I mean is Alan Perlis left by the time they came here. I don't know if Pamela and Joe had interactions with them. Over 50 a period, every computer scientist meets every other computer scientist. That's the great thing about this particular age, because over this period I came here in 69, and before that I was at Stanford from 63, before that I was at IBM. So it turns out from 59 onwards I was working on computers, vacuum tube computers and all kinds of things. But my introduction to AI happened at Stanford in 63. That was the year DARPA funded MIT, Carnegie Mellon and Stanford to do AI research. Little, little later. And Pamela was there from 65 at Stanford. It was the place to be, everything. As I read it, there are at least 90% of the detail that I forgot. I said to myself as reading the book, thank God she kept a detailed journal of every detail of all the things and the discussions. That is amazing. Thank you for doing it. You're welcome. I said earlier that when I was going to prepare this book, I went through these journals. They're just spiral bound notebooks. Not anything fancy. Morocco leather stuff or anything like that. Just plain old spiral bound notebooks. And I would read something and I'd think, I don't remember that. Did that happen? Yeah, it happened. I wouldn't have written it down. I wasn't making it up. In fact I wasn't sure whether these would ever see the light of day. And yet there was a sense I had that these early days of AI were going to be very important. And what I could capture was really important. Now, I'm not a technologist, remember? English major. But I heard the discussion all around me and knew what some of the issues were. So there I was. That's what I did. So I think we should get comments from everybody here and you open any page in the book and read a few words and then you'll be able to ask a question. Can you say more about this? But before we start there, she kind of defines what AI was from Pygambam, which is close to what I usually also say, namely to me, artificial intelligence kind of gives everybody, they imagine what it means and they kind of go on to make pronouncements based on their imagination. To me, AI is nothing more than getting computers to do things that human beings do. We see and we walk and we hear and talk and we think and we prove theorems and whatever. And that is the simplest definition, namely if a human being does something, you can ask, is it possible that a computer could do the same thing? Nothing more. We shouldn't add all kinds of imagined capabilities to it. And what we are finding is, if there was AI alerts to this today, even today understanding natural language has become a big problem and even the most recent advances in Bert and Robert and all kinds of things, those of you don't know, you'll find it. If you get AI alerts from AAAI, if not send me a message I'll forward you the link, there's a very interesting set of papers about natural language problems that is still not yet solved. And when people ask me, I say, it's probably going to take 10,000 years. Human evolution intelligence evolved over 10 million years. This may take 10,000. This weekend I was with Andy Van Dam and they said, no, you're under being very pessimistic, normally you're not like that. We think it'll happen in 100 years. I don't think it'll ever happen. To just give you an idea. Imagine multiplication and division. We didn't, in 2000, 5000 years ago, nobody knew how to count. Nobody knew how to read or write. People were all illiterate and they invented language and they invented number systems. It took them 3,000 years and even then they just got to the positional numbering system. And another 500 years to just understand what zero is, how to use it. So you've just that one algorithm took them 3,500 years. Imagine the algorithms for speech understanding, imagine for the algorithms for image understanding, imagine for walking. Everything, if you go to YouTube and say walking robots, you'll see lots of robots. All that they do is fall down after 50, 60 years. We had the best people here, Mark Rebert and so on. So the question is, what is AI and how is it going to impact the future? The last few chapters, which I haven't had time to kind of go through in detail, kind of provide a lot of additional detail. There's a section on art and AI. We had here Harold Cohen for a year on sabbatical. And we funded him there in the School of Fine Arts and Computer Science funded him jointly. And I have a computer drawn painting in my office signed by Harold. And if any of you want to come and see it, you're welcome to see it. And I'll probably give it to CMU at some point. The one thing I said in the 1960s, there was a big discussion about what computer science was. To me, computer science and AI were synonymous even then. And there were papers by Newell Simon and Pat Perlis about computer science is a phenomenon around computing and computers. I said computer science and AI have the same role to play as engineering. Engineering is a science which enhances the physical capabilities of the human being. And computer science and AI help you to enhance the mental capabilities. There's a very nice section in the book where Pamela talks about what Elon Musk and others talk about the thing and says that's stupid. What it is going to do is increase our capabilities, human capabilities. In some sense, we talk about super human AI, we should not be afraid of it. If you think of physical capabilities, just because we are now able to fly it a thousand miles an hour or 600 miles an hour, doesn't suddenly make us any less human, any less in control. But we are creating super human capabilities through engineering. The same thing will happen in the mental regime, anything we use our brain for. We may have super human AI that will kind of enhance it by a factor of a hundred or a thousand. One of the things we work on is guardian angel technology. That's also in the book. The guardian angel technology is an intelligent assistant which knows everything that's going on in the world. Then he's able to reason about all of that with respect to yourself because it's your agent and saying, how is it going to affect my life and then warns me about it. If there's going to be a problem or a hurricane or a tornado or even a traffic accident or anything, I have no way of knowing, but it knows. We, as human beings, can know things that we cannot know now. It's not a mystery and anybody can create it. It will happen. It's just a question of time. We should not be afraid of AI at all. There's no reason. Let me just add that Raj is being very polite. I was not so polite to Elon Musk. I said he was coming on about calling up the demons and all that stuff. I thought, Elon, have you given up your smartphone? I didn't think so. So on and so forth. I said to my friends, which I reported in the book, these guys have always been the smartest guys on the block. All of a sudden, here comes something that's going to be smarter. Whoa! No wonder they're all upset because there's a lot of ego involved. Having said that, I will not buy a Tesla because I have the feeling that Elon might mess around with my software. Can I tell a story about this involves Raj? This is the early days of language understanding, continuous speech understanding. Everybody is sitting in an auditorium in Science Hall. Raj gets up and says, OK, now I'm going to show you how my program works. Of course, like all pieces of software, it fails the first time. We laughed, we were used to it, we kept on chatting. I happened to be sitting next to Herb Simon. I noticed how we talked to each other. We didn't finish sentences. We waved our hands. We smiled to say, you know what I mean, kind of thing. Eventually this program wins a DARPA award. Joe cannot get anybody in Pittsburgh to pay attention to this. It's really a big deal. John McCarthy in Silicon Valley says, hey, this is a big deal. He tells the San Jose Mercury News, look what's happening. The Merc News is all over the place. Great stuff. Raj called Joe and he said, you know who else wants to talk to me? National Enquirer. And they say, if I don't talk to them, they're just going to break the door down anyway. OK, says Joe. Don't worry about it. I'll talk to them. And he gets on the phone to the National Enquirer. I guess then in Florida they say, oh, this is Joseph Trowb, Raj's chairman. I understand you're very interested in the things we're doing. Now, can I tell you the latest things about the Cwm Trowb algorithm? How about multivariate analysis? We're very good at that. I think they hung up on it. OK. Amazing. Thank you for being here. Pamela, your books I think espouse a particular view of history, the kind of history that says individuals matter. I think it's called the great man approach. And the alternative approach is there are big forces, social forces, economic forces that cause change. I think an interesting question to ask is in AI, in the intellectual area of AI, do the individuals you focus on really matter, or would we have basically had the same thing if somehow we had a parallel universe or something? The equivalent question is, what kind of AI are space aliens going to have? Many people invented calculus, so you can make the argument Newton didn't matter. And I would argue many people invented neural networks. And so we can point at any individual and say, well, maybe they didn't matter. Yes. I'm not a big fan of the great man theory of history. If it hadn't been Newell, Simon, McCarthy and Minsky, it would have been other people. Because it was in the air. These people aren't even talking to each other. And all around them, people are saying, you know, there's stuff that computers do. It's kind of like thinking. And so, yeah, it would have happened. But I was lucky enough to be around the guys who did make it happen. So I write about them. So if I can add to that, there are two interesting anecdotes I wanted to tell about. I was with Feigenbaum more recently about maybe six months ago or whatever. And he was saying, Raj, there is something going on. I was on a committee with a CMU faculty member and he said something. And he said, what would Simon say about that? And he said, Simon who? So the memory is not going to be 100 years from now. Most of us will not even be thought of or mentioned. It is irrelevant that we existed. Maybe Simon, even that, I doubt. Basically if I asked you who are the Nobel Prize winners in physics or chemistry last year or a year before, you may not. 100 years from now, most of the only people that you may remember is Wright brothers and a few other people like that. Maybe Steve Jobs as a person and a creator. But most of the rest of us won't matter. But the ideas that we help to create will be the foundations and the building blocks for things that will come in the future. And that's all we can say. Agree. You've had a unique opportunity which I appreciate very much in having had these decades of experience with these extraordinary people. I also think it's quite important that you came from humanities and your first book was in Fletch to me. And I wonder how you would say what you think the future will be. This AI thing, how's it going? If you get to the last part of the book, you see this person who is dying to connect AI and the humanities, sciences and the humanities. And so I have a lot to say about the digital humanities which for me is the bridge that's going to come across. No, it's really important. Really important. It's not just engineering. I've been sitting in the day and I'm in a slightly confusing situation there. I was wondering, but I refused to leave that you did not. There should be a lot of interesting ideas in those conversations. And I'm wondering if you can remember instance? I think I would ask awkward questions. I don't think I had great ideas. You will see that there's a period of about two years where Herb Simon and his wife Dorothea Simon, Alan Newell and Noel Newell, Joe and I, and a novelist who lived here, Mark Harris and his wife, we gathered every month and this had kind of been cooked up between Simon and me because I said ah, you know Herb, my students have the really interesting questions, what's the meaning of life and things like that. I'm stuck with how much can we spend on the photocopying. It should romantic poetry be two semesters or only one and on and on. So it was Herb's idea, he said oh well let's get together and we'll try it once and if it works we'll do it monthly and that's exactly what we did and you'll see the accounts of two of those meetings but there were lots more than I wasn't able to put in the book because it was already so long but those wonderful afternoon evenings we met after supper. They were lots and lots of fun. Do you have any advice for aspiring authors who want to write about various topics? I'm a master's student in AI but I do have an interest in philosophy as well and I was thinking of writing a philosophy book. Oh my, publishing is in great flux right now. Raj mentioned that I'm not on a book tour, I'm talking to people who have podcasts and some of them I think gosh this guy's as dumb as a box of rocks. He has 500,000 people listening to him so who am I? And this is how ideas get communicated these days. It's unusual for an author to be able to be in person with a group like this. It just doesn't happen anymore. The publishers can't afford it for one thing and the publishers are really up against a lot of problems. They don't have the resources they once had. Writing, publishing is not the central thing it once was in our intellectual lives. So we do our best. Maybe I could interject. Pamela is spot on. There are 0.1% of the 0.01% who make it in terms of being at New York Times. But there are tremendous opportunities I think for people who have a voice who want to be hired to publish using innovative approaches. We here in CMU libraries are working with the ETC Press. Brad is sitting there with the ETC Press t-shirt. We are thinking about the notion of a CMU publishing type framework to help people who have something to say relevant to our space to get their voice out there. So talk to Brad, see how we can get things going. There are lots of people at CMU working in the interface of ethics and philosophy and AI. If you don't know them, take the time to meet people like David Danks and Alex London and others who can help you explore your interests more fully. This is a phenomenal community to be part of in terms of being able to stitch together interdisciplinary and cross disciplinary ideas. But Brad can be your agent and your publisher. I'm wondering who came up with the term artificial intelligence and if you would like that term after all these years. I like it very much because I like the sense of artifice and all those things. The term was invented by John McCarthy at Stanford. He and his colleague, Marvin Minsky, who was then at MIT. I know there was some moving around, but I guess he was at MIT. Anyway, they had a summer where they went around asking people to contribute papers to this thing called artificial intelligence. Claude Shannon, Mr Information Theory, it was under his auspices, but he said, I'm not going to do any of the work, so you guys put it together. That's how AI was coined. Now for some years, some decades, it was considered a kind of bad term. You'd hear all kinds of funny ways of computers doing things, which if humans did them, we'd say that's intelligent behavior. I have a list of them in my book because AI had gotten itself into a bit of bad odor. I feel it was mostly on account of people who wanted to sell things. AI has come back and is respectable now. Great. The term is unfortunate, maybe, but it is with us. But it turns out over a period of last 50, 60 years, we've seen these ups and downs of artificial intelligence, because that name is ambiguous. It creates all kinds of imagined things, depending on who it is that hears the phrase, you think of something. All I say is, think of trying to get machines to do things that you do, whatever they are. You multiply. Can you get a machine? Yes, of course. But you speak. Can you get a computer to speak here? Those kinds of things are my definition. When I hear the phrase, I'm not offended by it. For a while, Newell and Simon didn't like the phrase. They used the phrase complex information processing. Oh, really? That's too long. Too many syllables. Nobody will use it. Finally, they also accepted the term AI, because at least it communicates roughly the area that they were working on. Other people tried to use the phrase cognitive science, which is probably closer, and information processing psychology. All of them are approximations to what people were trying to do. AI seems to somehow embody all those different ideas. Rather than trying to change it, we can say, this is what we mean. Jim? Both of you touched on a different part of the power of AI. You said once maybe our brains are free, some of the stuff that we currently have to be thinking about, we may achieve some higher intelligence or come up with some new processes or ways. Pamela, you brought up the point that, get over it, this stuff's going to eventually be smarter than many of the things that, you know, it is the way it is. You're describing two potential paths here. I would imagine some folks who are freed up from their drudgeries of what they have to use their brains for today might start watching the more television they've got to fit, and then others might then be trying to think about the problems that are in it. I think, you know, Simon has always been right. Basically, the human brain is kind of bombarded with too much information, and we are in society as a whole is not able to deal with it. None of us. There are lots of things we ought to know and deal with. We are not able to cope with the information glut, and that's where I think computers are most useful. Already we've seen it, whether it's smartphone or Google search or all of those things, are giving us tools to kind of do more with less. In fact, that is one of the slogans of Microsoft in the last five years, you know, which is do more. That's it, you know, whatever that means. That is, the goal of Microsoft is to enable everybody to do more. And so, in some sense, all of whatever might happen in the next hundred years will be in that realm, including super human capabilities, because just because we are able to move faster, just because we are able to build submarines to go under water without waiting for evolution to give us, you know, fins, there's nothing to be afraid of. That's all I'm saying. My feeling is whatever advances come will be empowering human race in ways that we will all be happy that it happened. I'd like to say something about the term intelligence. As AI was getting started, there was no question, but intelligence was a human property. Nobody else had intelligence, and we furthermore thought that intelligence meant these very high pollutant things, like playing chess and so on and so forth. Well, guess what? Intelligence has all kinds of manifestations, and I was fascinated to hear a talk by Franz Deval, the great primatologist. I'd just gone to find out what was new with the chimps, you know, and he said AI taught us the questions to ask. We did not know how to ask whether chimp is intelligent or not until AI came along and guided us. That was pretty interesting. He didn't know I was in the audience, and there's a line that was cut out of machines who think. You know, one of my snottier things along with I'm sorry, was. I said, and you know, this chess and theorem proving and stuff, the patriarchal hit parade, it's all, my editor said, you can't put that in. Okay, okay, you're right. I will, that will, but I still think so. So there are things that we do that computers are nowhere close to doing. We get angry. We laugh at a joke. You know, we can create jokes and we create intuitiveness and you know, and what we call creativity. You know, Simon, the work comes creativity, how people discover things. There's a book on scientific discovery. But nobody is working on emotional computers that I know how a computer can exhibit anger and how and why and when. And so there are lots of things. To me, they're not insurmountable. Basically, all it takes is enough money and somebody to work on it. If we can convince our authority to start a program in emotional computing and fund it for 20 years, we'll make a lot of progress. That's all it is. Peter, okay. Photo editor decided she didn't want to have any formal portraits in the book, which I eventually agreed with. But there was that thing with me in a football outfit with a football in my hand. I think I've handled football twice in my life. And that was a trading card for the computer history museum's fundraiser called the Computer Bowl. You can still find it with all of us sitting there being not too serious, but serious enough. And I saw this trading card and I thought, okay, fine. But then there it is in the book. And I'm thinking, really? I mean, I'm a dignified author. No, that's how come it's there. So when I arrived here in 1977, AI was all about knowledge based systems, expert systems. You want to do something like diagnose diseases or play chess. You interview a bunch of doctors or chess players, pull up with the gazillion production rules and figure out how to resolve conflicts among them and so forth. And that's what people thought about it. At least that's what I would hear about. Now, there are other ways of doing things. And, of course, the DARFASERF program of the 1970s, which of course, by a gradual remember began the hearsay and the heartbeat systems, which to my mind were kind of the archetypes for knowledge based approaches and statistically based approaches. Now, 42 years later, when you think about AI, when you hear about AI, what I always hear about are systems based on deep learning, now called deep learning, which are almost completely statistically based. And it feels to me like AI seems to have done a 180 in terms of what we're thinking. Nobody talked about statistics then. Nobody talks about knowledge now. So I'm wondering what the two of you think about all that and also whether there's any hope for knowledge representations. I'll start. You're quite right about how things have changed because we now have access to all this knowledge, all these decisions that they can look at and say, oh, one million people made the following decision. It must be good. So on. Now, if you look at these particular programs, they are almost all in the part of the intelligence continuum that we share with the rest of the animal kingdom. Including slime molds and slime molds. Come on, no, really. Whereas the part that talks about symbolic stuff, that's pretty empty right now. And that's waiting for some really smart researchers to go in and fill that part of applications up. I'll leave it to Raj, who actually knows what he's talking about. No, I think Rich is exactly right. Namely, we seem to have come 180 degrees, but there were inklings of it even in 1970s. If you remember, we used hidden Markov models. As you know, that phrase hidden Markov models comes from statistics. If you remember, we learned the Gaussian parameters acoustic models by using data. The main difference is Jeff Hinton and company said, we're not going to anyway human code the learning of the thing. And we are going to let the system discover by itself. And even today, you know, one of the people, Benjio, in his talk, said, I believe there are limitations in what we are doing with deep learning. And somehow we have to go back and figure out how to introduce other aspects of reasoning. How can reasoning happen in deep learning? One of the questions they're all asking, I'm sure we'll find something. But the bottom line, though, is the importance of statistics and big data statistics. You know, that is most statisticians until recently have not been looking at data items of billions and trillions of data items. Now they're all studying. There are some new theories coming out from statisticians and big data. And I'm very hopeful that that will give us some new insight. I've heard all things about the different definitions and how it's changed. I like to think that mine is timeless and that it's programming techniques that are not yet standard. It tends to annoy people. So you said that you don't believe that if it weren't for the people that you do, it would have been someone else. And that there was something in the air I think you said. You're going to talk about that? What was it in the air? What do you characterize that as? It was the advent of the computer itself. So John McCarthy sat at Caltech and listened to people saying, this is what we can do with this computer. And he thought, that's a lot like thinking. We could make these things think. And across the country, there was Marvin Minsky looking at what the computer did and thinking, that's a lot like thinking. We could do that. Newell and Simon had never done numerical computation. They thought it was too boring. And so they were really interested in what came to be knowledge-based systems. But yes, it could have been others. And it just happened that this stuff was in the zeitgeist, in the air. It happened that I knew the guys who were doing okay. So to me, AI, this is the phrase I think I want you to remember from this meeting. AI is about enhancing the mental capabilities of the human beings. And for the next 100, 200, 300 years, research will progress in that direction. What can we do to create intelligent assistance and whatever that will give me more power, enhancing my capabilities? And that's where the work will be. All the other things, imagining that somehow the world is going to come to an end and we're all going to be slaves of machines will never happen because nobody will work. I'm going to take this question and then fill it and then we're going to wrap up and have book signings and conversations and snacks. So please. So I might ask which humans, enhancing the capability of which humans? Because the way I see it, the most popular applications of AI in the last five, six years have not been solving the information innovation problem but actually creating capital for people that enhance it, make the information innovation problem worse. Like if you think about the Facebooks and the Twitters and those, I haven't seen examples of those new kinds of machine learning techniques being used to actually solve the synthesis problem for people to think better. I'm curious on your question. Enhancing the mental capability of the human being. Forget about individual things like Twitter and Facebook. It is accessed, we're getting information. And the question is how can I as an individual who cannot cope with, because of my brain's limitations, with billions of facts and millions of facts, how can I build a tool that will give me extra power that will give me more capabilities? What tools? Right now, the tools that we are discovering is from the big data, data analytics. We are discovering patterns in human behavior that have not been known before. For example, this is the big controversy about recommender systems. When you go to Amazon or Google or something, they're collecting all this data about each of us. They can predict what you're going to do before they even do it. People will say, my God, it even knew my birthday. How did it know that? To me, I stopped worrying about it. There's completely nothing I can do about that kind of thing. At the same time, insofar as helping me to do things better than before, I'm happy about it. It's the way things are. Thank you. That number of us got to talk to Herb and Alan and understand that. I think part of their goals is not to generate intelligent behavior, all the stuff we've been talking about, but to use AI only as a tool or at least as a tool to understand how humans think as a model, how we lost that thread. I don't think we have lost that thread. I don't. Basically, but that's where the cognitive science comes from and that's where the people in the psychology department here and other places would probably make progress. Right now, the deep learning stuff is not giving us an understanding of how the thinking happened. Rob, I'm sorry, you've been raising your hand. I'd just like to weigh in on the issue of the great man theory versus the great history theory and offer an intermediate point, which may you could speak to, but let me call it the great book theory and I'll just quickly trade through my own trajectory, which is along that theory and which is in 1965 when I encountered Niels Nielsen's book, Learning Machines, and that completely epiphimied me about computers and started the whole, that was a little snowball. Then I read computers and thought, so thank you very much for that. And that led me to Noah and Simon and then I read human problem solving and that's what ended up ringing. That's how you came here. But at the same time, I was reading about two other guys called March and Simon. And one of the greatest atyphonies I ever had was finding out it was the same Simon, March and Simon and Noah and Simon. So to build on what Raj said about 500 years from now, no one will remember we were here. But my answer to that is that's why we have libraries because 500 years from now people are going to be able to find this book. They are going to be able to find machines who think and they will find computers and thought. You could just talk to the protest about how much money so I'm going to wrap up the formal proceedings and invite you to come and chat with Pamela and Raj. Have your book signed. Quite a few people to think and I think to think. Hope I don't forget anyone. Brad King I mentioned earlier who has brought Pamela's book to life. There will be digital versions as well as physical versions between the preservation of paper and of bits and bytes. The book will survive into the future. Julia Cullen, the university archivist who helped bring together the photographs that were in the book. I'm sure that Ann Marie had a hand in turning them into digitised form. Heidi Barth did design the cover. She is our graphic designer in the university libraries. Andy Prisbyllett and Shannon Riff have made all of this possible. Mary reviewed an early version of the manuscript so thank you for doing that Mary. Raj and Pamela thank you for a wonderful evening. Ladies and gentlemen thank you for being here. Enjoy the rest of the evening.