25 years on, the world has changed significantly, driven by continual advancements in technology. But what have we learned from the rapid development of AI? And what role does it – or should it – play in our society?

Thanks to our panellists:

• Chuck Martin, Editorial Director, AI & IoT - Informa Tech
• Priya Krishnan, Director, Product Management, IBM Data and AI - IBM
• Garry Kasparov, Chess Grandmaster / Former World Chess Champion / Chairman - Human Rights Foundation
• Murray Campbell, Distinguished Research Scientist - IBM

 

The Transcript

 

00:01 Chuck

Good morning. Oh, come on. Good morning. Okay, they're awake. This is a good thing. I want to set the stage for set the table basically for 1997 things that were happening in 1997. CDs were king. There was no cloud. Netscape was the web browser. How many here use Netscape? Oh, this is depressing. The NASA Pathfinder landed on Mars. In 1997. The google.com domain was registered

 

00:33 Gary

with Alta Vista was the main

 

00:35 Chuck

Alta Vista. Yes, very good. Very good. The first paywall was introduced by the Wall Street Journal. In 1997 facebook.com, came online. Beanie Babies were the hottest consumer product that's depressing too. In 1997, Netflix was established. Steve Jobs returned to Apple as a consultant. And the first Grand Theft Auto video game was released. In 1997, IBM's, Deep Blue defeated chess grandmaster and former world chess champion, Garry Kasparov

 

01:17 Gary

I was the former at the time, the second match, not the first was the fourth at the time when,

 

01:23 Chuck

yes. So I wonder if we could start by Gary telling us at the time not knowing what you know, today, but what you knew then why you decided to play technology in chess,

 

01:36 Gary

Curiosity. I always wanted to do to test the machines and to understand what are the limits, both for computers and for humans. And we would go back, you know, in history, so the founding fathers in computer science, when you look at what they did, Alan Turing, Claude Shannon, Norbert Wiener, they all believe that chess will be some kind of ultimate test for machines intelligence. And, and since they had no access to to machine power. So this is the the, it was so big, so they believe wrongly, I don't feel comfortable in criticizing founding fathers of tech, but of computers, but that it's machines had to emulate human cognition. And again, it was it was understandable. But as the match was the blue proved and many other experiments. Now we'd say it's been Deep Blue was as intelligent as your alarm clock. Yeah, it's a very expensive one $10 million piece. But still, you know, I didn't need to have to be intelligent, because what we actually realized there was the, it's all about making fewer mistakes. It's not about perfection, it's about solving the game. It's not about emulating the way we think. But it's, it's simply no reducing number of human mistakes. And that was the one of the most important lessons we get to know at the time when I played it was, you know, it was probably a bit too early for us to understand that that was the main lesson of the match.

 

03:03 Chuck

So at the time, did you expect that sooner or later, technology would be a chess player?

 

03:10 Gary

You know, it's this match, you know, how public says By the way, it was a rematch because I won the first one to six, just you know, just the history. But we also played many games against other chess engines from the beginning of the 90s We played matches by the way I you know, this will remember that our first encounter was in 1989, when he was working with was was that his the team from Carnegie Mellon independently and they had this the the machine called a deep thought. And this machine actually made history by beating first Grandmaster that Larson in the tournament, game. So we played two games in New York, I won them quite handle it. And then we played other games. And we already suffered some losses in Blitz, five minutes game or in rapid chess, 30 minutes game. But the mistake we all made, we said, oh, five minutes, too quick, 30 minutes. But if we played real, you know, just quote unquote, real tournament chess six hours, then we would have enough time, what understanding that yes, you have more time machine will have more time. So it's this. So we ready actually, if we know from 2022 If we look back at the beginning of the 90s, I think it was quite predictable. And for me, this is the match. That's that we played the night and I still do, I still believe that, you know, I was in the position probably to compete and in even winning. After that match, I realized it would matter of time. So it still took about eight years for machines to dominate. But it's you know, it's it's it showed the trend, right. And also, I thought it was you know, it was my duty as a chess champion to accept the challenge because that's, you know, yeah, I lost the match, like, neither first or second place, but I think it's it's inspired. I don't know how many 1000s 10s of 1000s of computer young computer experts to  join not only IBM, but many, you know, just many other companies. I think it's it's always a milestone, not just you know, for AI, but I think it's for open just opening up new horizons, though again, I said this, the practical value probably was not that high. But it was just demonstrated that, you know, we actually move into the new era.

 

05:25 Chuck

So the photo that we're looking at here, this actually is Mark Campbell, terrifyingly moving the piece. They're not enemies just for the record. It was really the technology and Mario was involved in often, the obvious question is, why did IBM take a look at chess as a challenge?

 

05:44 Murray

Well, as far back as 1950, people had proposed chess as a grand challenge for computer science for artificial intelligence. And there's, there's a lot of good reasons for that. One is, as Gary said, When humans play chess, it requires various aspects of intelligence, it requires planning, reasoning, pattern recognition. And so the thought was getting computers to do that will be, you know, something special. But the good thing about chess is that, although it's extremely complex as a game, it's made a lot of simplifying assumptions, which is the way we scientists like to go about tackling a problem, we simplify and then understand the simplified version. So why is why am I saying this chess is simple, it's simple, because for a couple of reasons, one is it's a perfect information game. Everything you need to know about making the next move is right there in front of you. And there aren't, as it turns out, there aren't too many problems in the real world that are have exactly that character. Usually, there's hidden information or uncertainty of various sorts. A second reason is that it's a zero-sum game means what I win, you lose. And again, in the real world, we tend to look for positive sum games when hen kinds of situations. But if we make those simplifying assumptions, then we can develop an approach to tackling this problem. Now, it turns out, we, again, as Gary said, we decided, by the 1970s, that trying to emulate the human approach was not going to work. And we took a more computer approach. And that's how we actually made progress with deep blue and eventually produced the deep blue system.

 

07:42 Chuck

But and then later, Jeopardy..

 

07:45 Murray

Jeopardy adds an element when soon as you involve language, it opens up the possibilities a lot more, and it's no longer a closed system like chess. And that's the next level of of AI challenges.

 

08:00 Chuck

Love it. For you fast forward 25 years. Now you're dealing with data scientists of today versus the data science model back then what does it look like today? And then we're going to go through a little some history.

 

08:13 Priya

Yeah, actually, when Deep Blue came out, I was telling Murray, that I was still in my undergrad. And we had a room with computers, which was air conditioned, this was back in India, and only some of us were allowed in and out right? From then to now. Just even thinking about the possibilities that we can do. It has exploded, it has exploded in terms of the skill sets. It's exploded in terms of the technology, but comes with it, the fact that it has exploded is that it's unmanageable, as well. Right. So it's increased tremendously right now. In fact, the other day, I was reading an article that said, somebody is asking AI to predict what the trends will be in AI in the future. You know, that's kind of where we are right now. But it's amazing, because as the technology changes, it was to a select few in the past, like people like Murray could do this, right. But now I think it's when it comes mainstream. It's really democratized and made it more of a level playing field. So there are so many people that have come to the forefront to learn this. And to use AI, actually, in very good ways as well. It's I think that's the change that I've seen.

 

09:29 Chuck

So Murray, back in the day, the back for a second to 25 years. You're talking about AI and the fear factor in IBM. You talked about that a little bit.

 

09:38 Murray

Yeah, back in in 1997. We weren't generally using the term artificial intelligence, there was the movie 2001 had come out. Everybody knew that artificial intelligence, you know, had some concerns. So we often use the term data analytics or data science other terms like that. out in order to soften it. But in fact, Deep Blue was an artificial intelligence system as we would now call it.

 

10:09 Chuck

So Gary, you're have been following technology now since that, I mean, you've written books and so forth. And you've become sort of a an evangelist now for where technology fits. Let's talk about where technology does fit. We have all these fear movies of and fear stories of technology replacing humans. And obviously, that's not to say, obviously, it's not happening. Can you talk about that a little bit that we're humans fit in technology fits?

 

10:34 Gary

Yeah, let's talk about the progress. As is today, if you have chess app on your phone is better than Deep Blue. That's just to understand, so where we are today. So and, and when you start analyzing games and played 1997, with chess engines that you can just download on your laptop. So everything that 20 years ago looked phenomenal Kasparov made this move, the blue file and escape machine will be laughing in 30 seconds, we'll tell you how it's a mistake, it's a mistake. So machines are just much better, because they're much much, much, much faster. Now. I you know, I always tell people that, you know, we should consider you know, all these technologists, you know, as as a product of human ingenuity. At the end of the day, those are the tools. It's AI is not a magic world, but it's not a terminator. It's not a harbinger of utopia or dystopia. It's something that we invented and as machines in the past made us stronger and faster, I believe AI will make us smarter. So what about about the threats? Yeah, I'm very skeptical about all these singularities and and doomsday predictions, for a simple reason, because, as Murray has pointed out, it's it's everything that we have been dealing with now, we can probably argue about Jeopardy. It's, we're dealing with closed systems, chess, dota, I mean, all the video games, Starcraft, this is, and the problem that machines are yet you know, to find a way to transfer data from one closed system to another, if you do a computer game, and you train a machine to play one map, you want a similar map, you'd have to start from the scratch. So this is, so I believe there's always room for humans to actually to play a pivotal role. So as long as we recognize that we belong to last few decimal places. So yeah, machines can do better, maybe 95%. Of War, but still there was a room. And the the future is very much depends on our ability to work with, with smart machines. So exactly recognizing, so what this machine needs for this task, and what is missing. So what is the human contribution to make this the Centaur combination that will be as close as possible to 100% , because 100% is is not achievable? there's no perfection in the universe.

 

12:56 Chuck

machines of the day of a technology are going to be remained task oriented for a long time. where it's going.

 

13:04 Murray

I think that's where we stand right now is that AI systems tend to be quite narrow, as Gary said, focus on a particular task. And in the past, say two to three years, I think that trend has started to shift, where AI systems, particularly based on large language models, can do more than one thing. But for large part, that hasn't changed much in 25 years, all the changes that we've gone through in terms of computer speed and algorithms and so on. AI systems, for the most part, tend to be as narrow now as they were back in 1997.

 

13:48 Chuck

Is that what you're finding Priya, that things some things are not changing?

 

13:51 Priya

They are not. And actually, Murray and I were talking, I think the technology was still back there, it was with a select few. The explosion of data is something that's increased quite a bit, I think, in the 25 years that we never saw before the kind of data that we see today. And with the kind of data that we see, there were a number of problems, being able to use that data is also increasing. So I think that's a challenge that we see when we actually take it in into a real world application. Those are the problems that we're seeing, the data is exploding, the computing power has also increased, but it's still a challenge out there. In terms of you're gonna say something, Gary, for them.

 

14:33 Gary

It's about data explosion. Yes, absolutely. But what kind of data we're dealing with now. So the picture of your lunch, you know, it's probably, you know, takes more space than the entire works of Shakespeare. So, it's the equation is that there's so much data that is now available, and it's and it's being added in, I don't It's every day, you know, we have, you know, terabytes show, but it's, it's again, and that's exactly what I believe humans could play an important role. because this is it's ocean of data, you have to know. Okay, stop. Let's do it. So it's about direction. Again, I think it's we're still, you know, contemplating what is our role, we could see the data explorers, we could see machines getting faster, smarter, but we are struggling to understand. So how we, you know, make sure that, you know, our contribution to decision making process remains unique. Yeah.

 

15:20 Chuck

So this technology basically becomes for,

 

15:22 Murray

I just want to make a comment there. I think Computer Chess gives us a good preview of what we're going to see in many other fields. In the there, maybe you can stay up, there are four phases, the first phase, computers were too weak to help a strong player. In the second phase, they did certain things quite well, but overall, not as good as the human. So the human takes charge, and they drive, but the computer can help on certain things. In the third phase, it flips on the computer is very, very good, but it has gaps in his knowledge. And that's where a human can influence it and make it better. And in the last phase, which we're probably in right now, for chess anyway, it's very hard for a human to influence a computer and make it play better. So I think Gary had a proposed way back in the 90s, that the collaboration of humans and computers working together, is an interesting phenomena. And in fact, there's a window of time, maybe 1015 years where that was, in fact, the way to produce the best chess playing entity on the planet is human and computers working together.

 

16:42 Priya

And I think that is going to stay even beyond chess, can you imagine going into surgery and just having nobody but robots working on you. I mean, just thinking out loud, I It's, it's, it's, I think the intersection of human computer is here to stay, and I strongly do believe that them working together is the best one replacing another is not the way to go.

 

17:04 Gary

I, I probably should argue as Murray it's, it's, again, it's about last few decimal places, it's the the human role has been shrinking, but it will never disappear. Even if the most scared machines, you know, you have a computer that starts from scratch, let's say take game of chess, and it comes up with its own system, like, you know, Alpha zero. So this is plays, you know, millions and millions of games, and it has its own patterns or system of evaluation, but they're always, you know, inaccuracies gaps there, for instance, you know, speaking about chess, the state could be a gap between the value of Bishop a knight, so this is, and because of big number of games, statistically, you know, it values Bishop much higher than it had to be. So, for machine to actually understand that there's the, there's an inconsistency in after playing 50 60 million games. So it will mean 10s of 1000s of losses, before it gets in to the right, adult human could actually somebody immediately and start. Again, it's, it's a little tuning, it's like, you know, you have a very powerful rifle, and that it could shoot, you know, one mile, and all you can do is just to tune just, you know, a millimeter in it, or just, you know, point a millimeter in a barrel means, you know, probably, you know, 10 feet, or just, you know, it's, I don't know, maybe 50 feet, you know, a difference mile away. So that's why I'm going back to this point, it's about understanding what is our role, the danger is that we'll try to expand or all to enter the territorial machines already doing better. And that's why sometimes you better deal with an expert, not the top Grandmaster or just the top professor because psychologically I know, it's, it's difficult to accept but the superiority. So and, again, maybe an experienced nurse could do better in assistant for in radiology than top professor because she will look she will look at the at the screen, and she'll recognize that this is a little bit that, you know, that that could be done to improve she'd know to stop arguing about what she's overall conclusion.

 

19:13 Chuck

So is there an end to that? Or does it keep going like that, where it's the human is always involved, no matter what the technology is not going to get perfected?

 

19:23 Murray

I have an example. There was a paper that I saw that just came out today about somebody who taken a very top level go program go is this game this many played in Asia, but very difficult game is more complex than chess in some ways. And they show that this highly proficient system can be defeated by a very simple system by using a technique showing that there are gaps in even this very high-level system. So I'm concerned, whether it be games or software development or other fields, no matter how good these AI systems get, there are going to be gaps that a human can see just from our experience and world knowledge that they can't. The AI systems can't recognize themselves.

 

20:15 Chuck

So what's the challenge of the future? It was chest 25 years ago for technology. What were you? I mean, what are your data science says? Saying what they what they need? Because these are, these are the scientists. I mean, that's what you're doing. I mean, that's you're researching, what is it that the people are looking for?

 

20:35 Priya

I think, thank you for asking the data scientists of the world. They actually love the idea of automation and AI to actually do more and develop more of these models. So they would rather have the systems do the mundane tasks around them, and they want to be focused on their high skill, jobs that they have to do. For the future, I think, I mean, you're still asking about the interference in the intersection of human and technology here. Are we still on that? Yeah. personal view is that at least for quite some time, I do see the intersection of human technology there. But more and more that it actually makes us more efficient. Like the data scientists of today are still wrangling with jobs that they don't want to do. They're still struggling with data, whatever data, you say, Gary, they're still struggling with it right? Today. They don't want to do that. Like why can't I have a system do some of that stuff? Like why can't I make the system better? So I think that's where the head is. And that's where it's going to be for some time. So the more and more that we think about it that yesterday, one of the talks that I had was around, okay, can I actually put a governance around? What's exploding out here? Because once it goes into the world, it's very hard to start controlling some of this in the future.

 

21:57 Murray

Yeah I have a great example. Today, software, it's software engineering is you could say it's in the first phase of what we saw in computer chess, where computers are very good at doing certain things, and can help software engineers develop their systems. So right now, there's, there's a system called copilot, and many software engineers writing code, use this to help suggest ways of writing little snippets of code for them. It identifies library functions that might be useful for what it looks like you're trying to do right now. I think that's a trend that will continue and make software engineers much more productive in the coming years.

 

22:39 Chuck

So is that the role of technology for the foreseeable future? It's really an assistant.

 

22:46 Gary

No, I think one of the problems we have to deal now is it's I think it's a public misconception about the power of a machine. You know, it's we're still dealing with this, you know, with this notion that unless machine does it perfectly, it's no good. But again, it's about making fewer mistakes. So this is I think that's what's it's kind of a roadblock on the way of bringing more technology into our lives, like, you know, driverless cars. Yeah, the Yeah. Can you live it in a world where driverless cars, you know, on the streets, but you know, it's not 100% Perfection, there will be accidents. I think that's, that's, you know, it's about statistics. But the problem is that, you know, while 40,000 or so, also being killed in car accidents in the United States every year, so it's pure statistics, but the one accident was driverless cars, the front page of a newspaper. So, so again, we need just to simply accept the fact that, you know, it's, it's yes, it's the mistakes inevitable, machines will never be 100 isn't perfect, but we have to actually just to remove this roadblock, I think it's very important for us to actually to start, you know, taking, yes, it's a risk. But I think the bottom line eventually, you know, it's it's, it's woody should be very attractive for us to actually do to cope with this notion of imperfection.

 

24:13 Chuck

So how does society get there? You're right about the car crash of an autonomous vehicle, autonomous vehicle crashes in 10,000. Others crash it, there's one story, the autonomous vehicle, how does society get over that?

 

24:27 Priya

I think it's it'll take time, it'll take more of the maturity as, as we interact more and more with the system's AI, like Gary said, I think the idea that, Oh, my God, this is something out there, right. As we get more and more comfortable, I think the intersection gets a little bit better. So then we start to stop thinking about what you said, this is one incident that we're going to be reporting on. That's happening, I think, and it's going to keep happening. So I would say it's a matter of time. More than that, but I'm just curious, do you in the future? Do you would you think people are going to watch two machines just playing chess? would that even be interesting for somebody?

 

25:11 Gary

Well, they're playing now?

 

25:13 Priya

There's, there's always I mean, it's,

 

25:15 Gary

it's but it's I don't think that you know, watching machines playing chess, I mean, it's as exciting as watching cube of Legends, because are their emotions, you know, when you see humans? Yeah, you're looking for mistakes, again, are they it's about the game is lost and won, you know, because once I'd made mistakes, actually, machines, games are much longer, bad. But watching these games, you can also understand, you know, that's the interstates certain things about the game of chess that we didn't understand before. I mean, sometimes it's very painful, because you have now all these machines or positions, you know, like endgame positions started was four pieces now, five, six, I think seven. Now we're moving to probably eight but seven pieces positions when you look at this position, and it's, it's the machine says, made in 499 moves. I mean, it says, and I can tell you that it's when you look at the solution, quote, unquote, is solution. So the first four and 50 moves, I wouldn't understand what's happened. pieces moving around the pool, but there's just not much has changed, but and then you start understanding. So but somehow, it also tells us that you know, about imperfection of our knowledge, let's imagine if many endgame positions could be made in private and moves, and the average duration of human games about 45-50 moves, it gives an idea that probably, you know, we there's a lot of room for improvement. So that's what I believe, you know, it's it's a future for us to actually learn from machines to understand. So what is missing? Yeah, it's and I think just as as, as long as we will, we will go all our pride, you know, this just, it's just a false sense of, oh, no, no, no, no, no, we know better. Okay, fine. Let's learn. And that's why I'm, you know, I'm, I'm happy to learn now, that the many the world of chess whether we can recognize that, and let's get the biggest challenge now, it's actually it's not, you know, two machines playing each other. It's as a human strength to cheat. Yes, everyone talks about chess, I'm curious why nobody talks about other things like bar exams, because we are dealing with so many things where, you know, financial stakes are just, you know, in place. And it all depends on data. In chess, at least we have some systems to verify, but it's everywhere now. So this is the the cheating actually having access to data and outperforming your competition. This is the big issue. I think that's what we should we should concentrate. Not about, you know, this is the it's It's Hollywood, brainwashing stories about machines, you know, there's going after humans.

 

27:41 Priya

Yeah. But that's the point, right, which is, as you get more and more comfortable, as more and more comes with it, you start interacting with it. I think that is where you get the comfortable working relationship, but also, where you brought up, which is I want to highlight that again, right? There are ways in which I mean, it's not all for good, it's not all out there, we need some controls. Otherwise, it's gonna go into what Gary was mentioning.

 

28:06 Chuck

So let's talk about time, because you're dealing with time in a big way. Because you're in your position, you're creating the ultimate challenge for IBM. Okay, we got that one. But it takes a long time. In business, people have quarterly results. And they've got different pressures for time for delivering the technology doesn't go the same speed that businesses necessarily want it to go. How does that ultimately reconcile either in business or in technology? Where does the time start to align? I'll just say that

 

28:42 Murray

you can't schedule breakthroughs. They happen when they're ready. In fact, it's a phenomenon across the world that when the field is ready, the same breakthrough can happen in multiple places independently. And so you have to be patient for some things, they just aren't going to happen when you snap your fingers.

 

29:06 Chuck

The other thoughts on time are you dealing with 25 years 25 years ago, Gary, towards out there were 25 years later now and you've seen a lot in that period of time. That thing's gone as fast as you expected or slower.

 

29:25 Gary

Oh, I'm turning 60 Next year, so I'm not sure the right person to answer the question. I can tell you that just looking at my kids. So this is daughter 16. My son is seven. So a couple of years ago, he discovered his eldest is a sister or a DVDs and he didn't know what it is. So that's what you do. So it's, it tells you about this. It's I think it's about inevitability. So that's why is this 1997 was an important milestone. For me personally, because I understood, it's a matter of time, I still thought I could do you know, it's a good job, but what playing machines, I played few other computers, so just in 2000, to 2003. But, again, I knew the day would come. And we had to prepare for that. So same as now it's this, it's time flies, and we could see new machines. But I think, again, this is still is inconsistency, you know, with the ability of machines to help us in our in our lives. Whether do routine job, by the way, I will do, by the way, tons of the of the of the jobs that you know, that we can actually, you know, subscribe to the machines, we don't, you know, it's, we, I think we have to celebrate our creativity, I mean, we have to find exactly the what is the what is the quarter where in our, our ability to, to put on display human, you know, human talents is unique, and we're still struggling with was was finding this again, the right spot for us. And, and a lot of people feel uncomfortable on machines doing doing this work perfect. You know, it says if machine is doing your job, your job is rotten. So this, this, this, that fight, so let's find find another job. And, and it's happened in history so many times, you know, machines, you know, replace humans, and then we moved elsewhere. So at this point of room for us,

 

31:18 Chuck

and in 2017, you did a great TED Talk, which has been watched by millions of people, as you know, and in that you said, you must face our fears to get the best out of technology. He talked about that a little bit, how do people face their fears to get the best?

 

31:32 Gary

Look, I mean, fear is understandable, again, because you know, this is every technology the past, you know, before creating new jobs, destroyed old industries. So when we are seeing this process, but But again, it's it's inevitable, its history, its history of progress, and I while in I don't want to sound callous, I don't want us to, to, to talk about to talk down the fears or the fear of people who might be facing this, this, this this new challenges, but at the end of the day, you know, it's, it's if we are talking about humanity, we always will, because he lets you know what I mentioned, radiology, which is, you have, you know, many jobs, maybe hundreds, maybe 1000s, of good well paid white collar jobs in America or in Europe, because of because computers AI that will require, you know, just you know, experts, not top professors to operate, but the other side, cost reduction, more in, it'll be accessible to many people in America, in Europe, and also the third world. So then with technology, you can actually offer these services to people in in Africa, in Asia. So as humanity was weak, when obviously, you know, it's not, you know, it's not easy, you know, not so a win win for everybody. But I just view it view of this, this this problem, you know, just from the global perspective.

 

32:48 Chuck

Other thoughts?

 

32:50 Priya

Now, I was thinking about the time question that you asked, and I was interesting, if you wouldn't mind touching, if I can touch on that a bit, right? It time is time it's constant. Everybody knows it's what it is. But they think the Illusion of Time is changing. And when you think about the innovation that's coming across the globe, here, yes, it's come quite a bit. But in other countries, it's coming right now. And for them, it almost feels like it's moving really fast. Time is precious. I'm doing a I'm going ahead with time, right. So I think that is a good thing. And what you said was what I wanted to call back as well like, like the example of a data scientists wanting to do what they are skilled in doing is the great thing. So giving up the other mundane tasks or something else that the machine can do better, is something that they have to get comfortable. When you get comfortable. You have that extra time to do what you have what you really trained to do and to do better, you know. So that mindset, I think, is something that we're still grappling with a little bit. We need to get there, we need to get comfortable with that.

 

33:54 Chuck

Murray, I wonder if you can talk about unintended consequences, because you're looking at the future virtually all the time. How does how does that fit in in the grand scheme when you're because you're obviously scientists and you're looking at technologies in a perfect world? How do you factor in a potential downside of it, if there might be one?

 

34:18 Murray

That is a big concern. The as you're developing as we're developing technologies and labs, for example, we're trying to make more general intelligence. The obvious goal for us is to create systems that can help us tackle some of the world's really, really big problems. You know, climate, economic uncertainty, etc. Right now, computers are not great at problems that are complex require multi step reasoning, and that kind of effort. We hope will pay off. Now, the fact is that such systems, we can't control how they will be used. And if they are used to manipulate people into thinking certain ways, we have to be on guard for that. I think creating systems that can help people avoid being manipulated, would be a good target for this kind of technology.

 

35:29 Chuck

So things like sustainability, you see technology playing the role in the future.

 

35:33 Murray

I think it already is playing a role. The question is, Can AI as it becomes able to encompass broader and broader parts of our society? Will it be able to help us find better solutions? And I firmly believe yes, but I don't think the systems that we have yet are there are good enough yet.

 

35:56 Chuck

Systems of the future for good.

 

36:00 Priya

Oh, it's it's important. It's super important that I mean, I strongly like I keep saying from the beginning, which is, as these keep developing, I think the need for us to stop and think you mentioned sustainability. That's a that's an excellent angle, right? Yes, I can create machines I can do. I can create models that can do a whole lot, but at the expense of what, right, always have to think and stop and ask that question. And there are guardrails that are essential. And I it's already happening, but we are creating AI systems for social good. And I really hope to see more and more of that going forward. You know, and to do that we do need those guardrails on systems that do not do AI for social good.

 

36:46 Chuck

Technology for good.

 

36:50 Gary

I'm not sure I, I liked the question. Because I actually said this wrong question. technology could be used for good or bad. Yes, as we know that. So this is this whole idea is AI ethics, it will rely on us because at the end of the day, humans still have monopoly for evil. So this is anything that is invented to improve our lives could be used against us, as we could see, now this is this, you got social networks, the idea was that if we have a free exchange of information, so it will help to spread democracy in the world, not so fast. So we're glad we could see you know, the other guy's the on the other side of the fence. So they know how to use technology invented and developed in the free world, to attack our own democracy. So that's why I think it says whatever we do, you know, we still have human at the bottom of this equation. So I think that it's, it's, we should not lose the fact that it's, it's still a human issue. So and if machine comes up with some sort of answer, we don't like it, mostly it because it looked at the data accumulated through our history. And you know, the statistical leader we did, it comes up with an answer. So when people talk about, oh, we have to fix AI ethics. I mean, come on, it's like mirror, you look at the mirror. If you don't like what you see there, you can, of course, distort the mirror before you do some work on your body.

 

38:11 Chuck

Or stop looking in the mirror. For your final thoughts on the future.

 

38:20 Priya

I think there's a lot of promise. And I like I said, I really liked the idea of the intersection of human and, and AI together. It will be there to stay. That's my strong feeling. And I certainly believe that as as these things change it we have to ask ourselves, Am I am I'm doing this at what costs every single time the onus is on us. The onus is on us to ask this question and make sure that we are progressing. Really the AI for social good, right.

 

38:53 Chuck

Final thoughts on the future?

 

38:55 Murray

I think there's already a lot of value being generated by these narrow AI systems that we have. I think if we can get to broader AI systems that can quickly pick up new tasks rather than having to be expensively trained. I think that will move us to the next level.

 

39:18 Chuck

Final thoughts on the future?

 

39:20 Gary

Technology will eventually benefit people in the free world free minds will always dominate. For those who have any doubts. Just look at the latest tragic events in the world. The virus came from China vaccine from the United States.

 

39:35 Chuck

Please join me in thanking this fantastic panel.