Deep Thinking: AI and Human Collaboration - Garry Kasparov

So we don't have time to go into the details, but it's not just the opening stage of his career or damage career, but it's also the program AlphaGo that later was transformed Alpha Zero that eventually became the very foundation for his phenomenal work that earned him a Nobel Prize.

But as the world champion I the top chess player, I was kind of a target for these computer experts because as I already said cracking the secret of human intelligence it's it's needed some kind of metrics and what could be better than the game of chess. And my first introduction to the world of computers actually began much earlier. So it's even before I won my title in November 1985. So it's still 1985 but it's Hamburg June just few months before I played my match with Anatoly Karpov and won the title.

And as you can see on this picture I don't know how many of you just familiar with the game of chess it's one where the professional player goes around moving the pieces and in circles and I uniqueness of the event is that I faced 32 chess computers. Judging from the age of the audience I assume some of you still maybe have these antique pieces of computers that can only play chess. So there were four manufacturers each had eight models I played 32. Of course humans executed moves but machines made suggestions. So I won all the games all 32 games. And I have to say it was not a surprise. The surprise would be the opposite if I lost or even if I drew one game.

So when I look at this picture it's just this I feel bad because it reminds me of the golden era of human machine relations. Machines were weak but my hair was strong. And now I face on the one computer. But that was a fight for my life. Chess life, of course.

And I always remind people that in 1997, we played the rematch. Not just for the sake of historical accuracy. And in 1996, we played first match which I won. So that it was a rematch. And yeah, it's famous match. I lost this match. And in many instances it's just it's it's I heard it at so many places that it was the dawn of artificial intelligence. Uh-huh. Deep Blue was as intelligent as your alarm clock. Now, a very expensive one, $10 million a piece, but still it was not intelligent.

And as a matter of fact, it didn't have to be intelligent. Because here is something that we ignore, forget, or just don't want to know even today. For machines to excel in any field of human activities, they don't have to be perfect because there's no perfection in this universe. You don't expect 100% perfection. All machines have to do is to make fewer mistakes. That's it. And naturally, in this match, Deep Blue made fewer mistakes.

It's also ironic that the match that was trumpeted as the oh, near perfection. Garry Kasparov made this move, great move, and Deep Blue found the combination. All this celebrated as the greatest player ever. If you have a chess app on your phone now, so you find out how many mistakes we made. A chess engine today will be laughing at the at the game we played then. So, and by the way, there's also one thing to remember that is the chess app in your machine is much stronger than Deep Blue. So, it's just as just a fact.

And I just remember this is my experience there just going through this. And I thought at the end of the match that it would be curse. Being a first human being losing to a machine, it would stick to my name and just ruin my life. Now, almost 30 years later, I believe it was a blessing. Because I was the first one to understand that the future is not competition. We should not move look into the future as, "Oh, it's dominated by the Hollywood images, terminators, Matrix." No. It's about cooperation. And also just it's it helped me to get a more realistic understanding of these relations.

Because we hear so many polarized opinions. And it's typically when you when you hear somebody talking about computers, it's always about it's it's very emotional. "Oh, machines they will open us. They will it's open us new horizons. It's It buys us the free ticket to heaven." Mhm. No, it doesn't. "Oh, it opens gates of hell." No, it doesn't. It's not a harbinger of utopia or dystopia. It's a tool created by us, humans. And we should treat it without again being over excited. Just try to be objective.

Of course, it's not easy because machines that made us stronger in the past or faster, it's it's one story. But this is this is cognition. This is something that we believe was sacred. So, it's But I tried to actually figure out.

So, what's what's next? Again, that's why it's a unique experience. Back in 1997, I had enough personal experience and data to recognize that no, no, I was stronger than Deep Blue in 1997. Actually, I played two more matches against other computers, not as publicized as IBM match in 2003 in against the Jewish Israeli program Deep Junior and then against a German program and Fritz. Both ended as a tie.

But I knew we would have very limited timing left. And machines would do would get better. And it's a normal cycle. First machines it's a it's a laughing stock. Then, they just they can compete. And then there's a very short window on it's on the history timeline where we it's a real competition. And then, stage four, machines are much better forever after. So, just recognizing that back in 1997 helped me to think about the future and how we can establish the most effective form of collaboration.

But, before of course I'm I was a world champion. By the way, it was so painful in 1997 because it was not the first match I lost to against a computer. It was the first match I lost period. Then, after 12 years of winning, winning, winning, so that's it's Trust me, it was quite shocking. So, that's why I wanted to take a revenge. And I tried. I just I even spoke to Lou Gerstner.

I understand they made a good business decision because they knew after this victory and by the way, some of my coaches said my worst mistake was not actually to insist on having IBM shares because they jumped to 22% in 12 days. So, yeah so they knew that it would not add anything to the value. And they turned me down. And I was desperately trying to find out what's happened with the blue. I As a matter of fact, I was lucky. I found You recognize the place? It's making sushi in JFK in terminal five. I love sushi, but I never ate there. I would just go around.

And So I had to lick my wounds and contemplate all this ruminate and what's what's coming next. And the next year in 1998 I came up with what I thought was just a very important step forward, a concept that what I called advanced chess. So if we can't compete, maybe we should get together.

So, and chess became a very important field to actually test the human machine collaboration. Because we have something to add to the game. So, yes, machine can do many things better. So, it's it's it's it's all about this is the recognizing patterns, calculate brute force. But there's also human element of that.

And we had plenty of games played with humans assisted by machines playing other humans assisted by machines, and also the humans and machines together with machines playing much stronger machines. And we actually found out that human element is very important. As long as we understand that's important. That we belong to last few decimal places. We do not have to compete with machines expertise. Maybe 95% 96% out of this 100% decision-making formula.

As long as we understand what we can add to the specific machine for the specific problem, it's it makes all the difference. And what is also we found out that is when you have humans competing against each other with machines assistance, the formula of success it's not necessarily is with the human expertise or with machines speed the machines power, but it's a process synergy.

And often we saw as a weaker player with a slower machine could actually prevail over stronger player with a faster machine. Because he or she or a group of people could actually find the best combination. It's a synergy. It's a centaur. I think that was quite an important experience.

And again, there's a shortage of time just to share all the data that have been accumulated, but there was one of the research made in the University of Milano about radiology analysis, and they proved that it's not about top professors expertise. It's about an expert even without the world recognized name, who knows how to get the best out of computer. Again, not to duplicate, not to compete.

And the reason I think it's psychological explanation why sometimes you don't want a top expert to work with a machine because psychologically you don't want to recognize machine's superiority anywhere. And trust me. So, the moment you start competing, the moment you have this kind of anger or just human pride standing on the way, you just you're losing the this uniqueness of this of this cooperation.

So now, of course, we all live in the world where nothing happens without machines. And natural question is, so what about all these new LLMs, just chat GPTs, whatever. You give these names. It actually fits my formula. Because while machine contains all the data and just goes straight and gives you quick answers, I still can manipulate it. And I can show you how.

Of course, the question I ask a computer, who is this best player of all the time? Yeah, that's yes. That's the first one. And you could see, it's a very diplomatic answer. Yeah, this is this. Kasparov of Carlsen is just say yeah, it's this. Fischer, which is by the way, it's it's a correct assessment. If I have to pick up top three, that would be my top three. No, no, I'm not here just to argue. I know it's it's it's a little bit difficult because it's this we all have a preferences, so but still this is it's it's good. It's but it's diplomatic.

But if I don't like it because I'm too proud, so I just I'll do a little tweak. Now, it's too long. Give me one answer. So question matters. So, we still have just it's it's not a switch on the wall, but it's just a little tweak. So, now again, the list could be very long. I just give you one more example. You can read it as machine explanation. No, it's it's of course it doesn't it doesn't freeze and at 27° F because it's 32. Again, it's it's it's it's you can always find a way to go around.

So, and it's it's a there was so many funny stories happened to me. It's back in I think in 2018, I wrote an article in Science magazine called chess a Drosophila of reasoning. He's explaining how chess can be just the ideal field to test some ideas. What? I mean, every second month I receive an invitation as an entomologist because I'm a professor. Because from AI's perspective Drosophila, so that says, "Yeah, reasoning professor." So, AI generated invitations so to the conference. I have to drop them, unfortunately, as I not to disappoint the hosts of these conferences.

So now, the answer to this problem. Pablo Picasso had no ideas about computers, but he said something which sounded odd when he said it, but this is true. Because question, and he knew it as an artist, is a beginning. And by asking the right question or a wrong question or whatever question, you can just make a little difference.

So, now this is this important thing to understand. Back to what I said a few minutes ago. Our role is just okay, it's a time of war now, so I use them military metaphor. So, you have a very powerful gun that can shoot 2 mi from here. And all you can do is just a it's it's a tiny correction. It's a tiny correction of the bullet in the barrel. 0.1 mm. 2 mi from here, it will be probably few meters. That's exactly our role. And recognizing how to do this little tweak, so little correction, that could give us a enormous competitive advantage.

And again, it says it's the it's Yeah. I don't want to sound callous. There will be jobs that could go on the chopping block of automation. But it's happened before. In 18th century, beginning of 19th century, it was the agricultural society. These jobs were these jobs, they've gone. Then the manufacturing, these jobs gone. So, we are undergoing a revolutionary change, but this is not unique. It's It's It's changing the landscape. It's This is It's reshaping it, but there's always room. As long as we understand what is required.

Say, humans are plus machines. So, this is a very simple illustration of a cycle. And that's what my genesis I preferred the word augmented to artificial. Because it's it's it's it's basically augmenting our ability to think and to advance the process. So, we start with brute force. And by the way, most of the computers, it's still brute force. And then you have smarter machines. Yeah, that's it's when we just It's like advanced chess. And we are just entering the phase the last phase. It's when you can imagine the human shepherd just is the pushing the flock of algorithms. So, so in the fields of grass. Just in your imagination.

So why I picked telescope, the old telescope? For a very simple reason. What does telescope do? It augments our vision. But there's one trick. Or just it's depends how you position it. If you look down, you will see nothing but dirt. If you look up, you'll see stars. So, that's exactly how we roll. So, that is we have to decide how we use the technology that helps us to augment whether it's our vision, our hearing, or our cognition.

So, I hope that this gives you a little bit more comfort in finding your spot, your room in this in this fast-changing world. Oh, I mentioned Hollywood movies, and of course, they had tremendous impact on us. And it's the images of these Terminators, and they had quite a chilling effect. So, by the way, I never understood why Skynet wanted to kill people. I just did see I remember just watching these Terminators.

But, actually I think that we can we can look at this at this movies at the Hollywood production from a different angle. So, and just to find out how does it fit the theory that I'm just sharing with you now. By the way, that's not Photoshop. It's a real one. So, where I visited the Terminator in his office in 2003. So, and he had, by the way, big respect for the game of chess. All his kids how to study chess. So, we played the game, a very short game. I offered draw instantly. Not stupid enough to push hard. I was under the impression that this very short game gave him such a confidence that 6 months later he run for the governor of California and won.

Now, you ask me why this is important. By the way, it's Terminator 2. I think it's it's going back to the big screen soon. Because what's happened in Terminator 2, human plus inferior machine beats more superior machine. And the human component was very important. Huh? So, that's it again. So, everything depends on the angle of observation. So, I think that in the end, we just we more we analyze it, And again, just brush aside the emotions. Just try to be objective. Again, I'm just it's not I'm not Santa Claus to tell you fairy tales. I'm just trying to be objective. I'm just not an optimist or just pessimistic. But I think it's very important for us to push aside the army of doomsayers. It's easy to sell the doom. The doomsday scenario. No.

So, we and then Let's by the way just think about just what we what we discussed. So, unfortunately, the screen is too big, so I Machines don't have to be perfect, only better. Very quickly. Humans define the problems. Question comes first. Human-plant machines need superior collaboration process. Again, process is the key. Understanding your computer is like a four-wheel one. Now, machines made us stronger, made us faster. Yeah. If we know how to use it, we get smarter. So, this is important for me. I'm 63, so I'm just I'm dumb compared to my kids. So, they laugh at me when I looking for the just a button. So, they It's easier. More sophisticated the program is, it's easier for us just it's dinosaurs to actually work with this.

And this is important. Because there's so many regulations now, especially in Europe, is this Now, it has to be explainability. I mean, are we in business? Are we looking for explainability or bottom line? No, it's nice. But if it's a black box, but it improves your bottom line, do you care? No, eventually it will get better, but you can't actually push it through. And also the whole idea this one ethically I. What is ethically I? AI is just he's looking at the data we collect and put it in. And if we had social ills in our past or our present, AI will reflect it. Or you start making your tweaks, but that's not AI. That would be your worldview installed. So again it says recognizing the limitations. I mean not expecting this AI to solve our problems. Because again it's oh this is we just put it AI in and it's no. It's still us. It's as you like you look in the mirror. I'm not physically fit. I don't like it. Two choices. You start training or you distort the mirror. Oh okay the answer. So most of us prefer distortion of the mirror because it's easier.

This is important also, yes. Rote work. I call it garbage collection. And it says most of the things we doing they are it's rote work. And AI could actually unleash our energy and creativity. It's just this type of wildest dreams. As long as we understand again this is it's it stays there and it can do the all the work that is being done, but we can just leave it with machines to accomplish.

Two pictures. I guess what is this one. It's it's hard to hard to miss. July 20th, 1969. And some people were I lived in Soviet Union they didn't cover it, but I just know. Apollo 11 it's on its way to the moon. Look at the a celebration. Can you say what is this one? It also 11. It's people are waiting online for iPhone 11 to be released. September 20th, 2019. It's amazing this is they're already having in their pockets or in their hands almost identical device. Yeah, I grew up in the Soviet Union. We used to start standing long lines for bread. Yeah, not for again device that you can you bought identical one just a year ago. So, that tells us that maybe we were not effective by using this enormous power that was given to us. Because each device that's these people have in their hands is 10,000 times more powerful than the entire computing power of NASA in 1969 when Americans land on the moon. It's all about fantasy. It's all about imagination. So, we Our wildest dream can be realized as long as we know how to use it and we're not we're not shy, we're not afraid.

The last American walk on the moon is 1972, you just showed them. And the good news is now we're trying again. That's that's that's confidence. That So, we recognize that this immense power should be used for good. I mean, by the way, it's there's no guarantee of success. One of the problems is we stopped doing space exploration. It's a risk. Yeah, it is risk. And sometimes it's it's tragic risk. Yeah, so the space shuttles can explode. And anything can happen. Apollo 13 nearly died. But that's another problem of our society. So, we think that we can eliminate risk, but they keep the profits. Uh-doesn't work this way. That sounds to me like a socialism. The free market doesn't work this way. So, again, should we assume more risk? Yes, and that's a very important one because we have powerful tools. They know it's it's not win-win-win-win situation. Somebody wins, somebody loses. As humanity will always win. When people talk about many jobs lost in Yes, in I don't know just this mentioned the same radiology. Yes, it there's a potential target this of experts white-collar jobs that might not survive, but opportunities to use this technology to incorporate tens hundreds of millions of people from Africa, Asia, other continents, that tells us that humanity will always win.

So, and just to wrap up my presentation, so I would like to go back to Hollywood. To one of my favorite movies. What's just quickly anybody what's your favorite Star Wars movie? No, okay. I just we disagree, so Mine is Empire Strikes Back. Yeah, actually probably because this was the first movie I saw on a big screen when I was 17 in Rome. Yeah, as a part of Soviet squad who played in the chess Olympiad. And So, why do I have this there? As I said it's You remember this moment of course. By the way, this is the typical Hollywood mistake because technically you should do it other way around. It's not 3,721 but one. So, that's but we forgive Hollywood. So, this is when they just mix chess pieces I always forgive them. So, that's the Now, this is interesting. Computer doesn't make a mistake and C-3PO was right. The chances of surviving in asteroid field were 1 to 3,720 against So, from machines perspective, from computers perspective, statistically, to survive, to extend your life, you have to go back to be caught by Darth Vader, by the Imperial Guard, to be tortured, and killed later. We was we know the outcome. The value of this one is much higher than one because it's your only chance. And you cannot quantify it. It's only humans can understand it. So, that's why I think it's very important to recognize that certain decisions that only for us to make. We They could be killed. Okay, not in the movie, but in real life. But still is the only chance. Is it risky? Yes. But what?

It's very risky to be an attacker, to move forward. But these days is much riskier than not to. Because your chances of improving your position, you improving your business, improving your enterprise using other spaces space metaphor. It's better when you can we when you are ahead of the curve, when you can see what's coming next. Being second very often is being lost.

So, I hope that in this 30 minutes, I just I make you feel better about our future. So, and then well, let's get busy and just continue exploration. Thank you.