DEFCON 25 conference. Garry Kasparov. "The last battle of the brain." Part 2

Original author: Garry Kasparov
  • Transfer
DEFCON 25 conference. Garry Kasparov. "The last battle of the brain." Part 1

I think that the problem is not that machines will replace a person at his workplace, including in the intellectual sphere of activity, and not that computers are allegedly taking up arms against people with higher education and a Twitter account. The introduction of AI is not at all fast, but rather too slow. Why? Because this is a normal cycle of human development, and we simply do not realize that the destruction we see means the introduction of new technology, which, before creating new jobs, destroys old ones.

Technology destroys obsolete industries and creates new ones, such is the process of creation, such is the development cycle. If you try to drag out the agony by inserting old technologies into this process or creating some advantages for outdated technologies, you will simply slow down the process and make it more painful. This will happen anyway, but the problem is that we are “regulating” the process, creating rules that deliberately slow it down. I consider this a greater problem than those that we are more clearly aware of. This is more of a psychological problem when people ask themselves: “how can you feel safe being in an unmanned vehicle”?

I looked into history and learned that one hundred years ago, one of the most powerful unions in New York was the elevators union, uniting 17 thousand workers. By the way, at that time there was already a technology when you could just press a button - and you're done, but people did not trust her! It’s just awful - press the button yourself to call the elevator! Do you know why this union "passed away" and people began to use buttons themselves? Because once the lifters decided to go on strike. They went on strike, and then people who needed to climb to the top of the Empire State Building risked pushing buttons themselves.

Remember what they said about children or grandchildren 20-30 years ago when they got behind the wheel of a car: “this is terrible, you just look at the statistics, because cars are one of the main causes of human death, how can they risk their lives?”

So, all this is pure psychology. We practically do not pay attention to how many people die in car accidents, but if one person dies from an unmanned vehicle, this event swells up to heaven. Any glitch, any mistake of artificial intelligence technologies is immediately covered on the front pages of newspapers. But look at the statistics, look at the number of incidents, and you will see how tiny this percentage of the total number of accidents is. Therefore, the human community will triumph only if it can move forward without being paralyzed by such fears.

Another question arises when we talk about fake news or cybersecurity, these are very politicized topics, and I get a lot of calls asking how I deal with AI haters. For example, I have a regular blog, and in my new article, which will be published in a couple of days, it is about hate and that salvation from hate consists in knowledge, in training. We just need to understand that this problem existed long before all these things were invented, it’s just that its importance has increased thanks to the Internet, which covers millions and billions of people.

I think it’s the other way around when someone is trying to stop progress by trying to outlaw the AI, and you know that it won’t work, because we have Putin and other bad guys, wherever they are, who use our own technologies created in the free world are against us. Therefore, I believe that we should just take it for granted.

The essence of the problem lies only in us, and the answers to the questions are within us, in our own strength and our own confidence. I argue that intelligent machines cannot make us "obsolete." However, it must be remembered that there are certain restrictions on the cooperation of man and computer, and to a large extent these are just rumors that existed before. As always, these are just new opportunities that destroy the old world and create a new one, and the further we move forward, the better we will find ourselves.

Now it is most reminiscent of a movement into the world of science fiction. The paradox is that if we look back 50-60 years ago, we will see that in those days science fiction was absolutely positive, it was a utopia. However, then a gradual transition from utopia to dystopia took place, so that we no longer want to hear anything about the future of mankind.

This did not happen in one night. There was a time when people decided that space exploration was too risky. This is really a big risk, but imagine that in 1969, when the Americans landed on the moon, all the computing power of NASA was less than the power of any modern computer device that fits in your pocket. This device is a thousand times more powerful than a supercomputer that existed 40 years ago. Just imagine how much computing power you carry in your pocket! However, I'm not sure that the Apple iPhone 7 has the same processing power that Apollo 7 had, that is, it can produce the same effect.

However, machines have provided us with many great achievements in space exploration or the oceans, and we must understand that computers provide us with the opportunity to take great risks.

I would like to finish my speech on a positive note. Isn’t that positive image shown on this slide? The photo in the lower right corner is not “photoshop,” I actually met with Terminator in 2003.

He also loves chess since childhood, but he didn’t specially deal with it, so he lost very quickly. Therefore, I was very surprised when, 6 months later, he ran for governor of California and won!

Why do I call these pictures positive? Because although in all series except the first one, old Arnold always stands on the side of the winners and never gets tired of fighting against new cars, it is in the first series that we see the combination that I spoke about - this is when a person plus an old machine plus a perfect interface defeats the latest a car.
You can say: “Yes, machines are stronger than people, because they can calculate absolutely everything!” However, the point is not that they can calculate everything. For example, in chess it is technically possible to talk about the mathematical infinity of the number of possible moves equal to 10 ^ 45, which is not difficult to calculate for any modern computer. However, in the game it is not the calculations that are important, but the fact that the computer is ahead of the person, because it is always guided by the rules. And you know the effect of these rules and know why the computer chooses the best move from the huge number of possible moves.

But if we turn to real life, then I'm not sure that a computer can always be useful. Let's look at the most typical situation - you have a computer that monitors budget compliance, you are in the store and are going to buy an expensive gift. The computer evaluates the purchase and says, “No, you cannot afford this thing because you exceed the budget.” The machine has calculated everything, but there is a small nuance - your child is standing nearby, and this gift is intended for him on his birthday. Do you see how much this changes the conditions of the problem? This changes everything, because the child is waiting for this gift.

I can start adding these little things that change everything, but I don’t think that they can be included in the conditions of the problem and get the right solution. We have a lot of rules, but we still have to ask questions because the situation is changing. This is what can be called a normal situation, but if you look at these films, you can say that the situation shown here is more dramatic and extraordinary. On this slide, you see a shot from episode V Star Wars: The Empire Strikes Back.

Han Solo directs the ship right through the asteroid field, and S-3PO panic reports that the chance to survive in this field is 1: 3122. Han Solo tells him: “Never tell me what our chances are!” This begs the question, who is more right in this situation?

The technique represented by S-3PO is absolutely right, because the chance of survival tends to zero. It is possible that from a robot’s point of view, being captured by imperial troops is a better choice that a person does not even consider than death in an asteroid field. But if the computer decides that surrendering to the empire is the best option, then we can assume that a person has no options at all. It is very important that in both cases, ordinary and extraordinary, we have the opportunity to make the final decision, and to make such a decision, human leadership is still required.

Sometimes this means that you have to oppose computer recommendations. The meaning of human leadership is not to know the odds, but to ask really significant questions, not only today or tomorrow, but in the distant future. This process can be called "human guidance" or "human intervention," exposure without the help of intelligent machines. This is what our course should be in this century.

People are sometimes surprised at my optimism about intelligent machines, given my experience with them, but I'm really optimistic. And I'm sure that all of you are equally optimistic about the future of AI. But we must remember that our technologies are agnostic. This is neither good nor bad, but can be used both for good deeds and for evil ones. Machines should become smarter and more capable. And we, people, must do what only people can do - dream, dream to the fullest, and then we can extract all the useful things that these amazing new tools carry in themselves.

As planned, we still have 10 minutes to answer questions.

Question: in your opinion, is it possible to create a machine learning system that could determine which moves are more in line with the human style of the game?

Kasparov: First of all, we do not expect from the computer that it will tell us the first move and the remaining 17505 moves. I think we should rely on the car as a source of getting the best recommendations for unique moves. By the way, top-class players use computers as a guide to help them take the most appropriate position in the game. I repeat once again - in 9 cases out of 10 the assessment of the situation by the computer far exceeds the assessment that a person is capable of making.

Question: Do you agree that real intelligence requires freedom of choice, freedom in decision-making, which can only be made by a person? After all, the Deep Blue software and other computer programs are written by people, and when you play Deep Blue, you lose not to the computer, but to the programmers who wrote this program. My question is this: is there any danger from any kind of machine intelligence as long as computers have freedom of choice?

Kasparov: here I have to move from science to philosophy. With regard to Deep Blue, everything is clear - this is the result of a lot of human work. In most cases, even in the case of AlphaGo Demis Hassabis, these are all the fruits of human intelligence. I don’t know if cars can have freedom of choice, but I believe that everything we do, if we know how to do it, cars will do better. However, doing most of the things, we don’t know how to do them in the best way, so we often can’t understand what will happen. Simply put, we have a goal, but we don’t know what it is, and the role of the machine is to help us realize this goal. Therefore, if we talk about the free choice of computers, then it should help to connect us with this goal. I think that for computers this is a very distant prospect.

Question: What do you think about such human characteristics as courage and morality, and about the decisions that artificial intelligence can make based on them? For example, what should an unmanned car do - run into a child or avoid a collision with him, crashing into a rock and killing his passenger?

Kasparov: this is what people call “feelings,” they cannot be quantified, because it is a whole bunch of different human characteristics. If we are talking about courage, then this characteristic always goes against the chances of choosing the best option. Courage, like other human feelings, by definition contradicts accurate counting.
Question: Mr. Kasparov, my question does not concern computers: what is in your flask and can I try it?

Kasparov: what do you mean?

Host: he asks what is in your pocket!

Kasparov: in my pocket? "Capital"! This is not an advertisement, if you notice, I threw it away.

Question: who, in your opinion, will become the next world chess champion and does the young Chinese chess player Wei Yi have a chance to overthrow Karelsen from the throne of the chess king?

Kasparov: Karelsen is player number 1, he is not a world champion, but simply the best chess player in the world according to the rating. This year he will be 27 years old, so he is still young, but not very young by modern standards. I think Wei And now 18 or 19 years old. Magnus has outstripped young players such as the Americans Wesley So and Fabiano Keruana, and potentially, Wei Yi could be his adversary. However, in order to become a world champion, talent is needed, it is not necessary to be young and energetic, it is enough to have a little luck. So, answering the question, I can say - yes, he has a chance to defeat Magnus Karelsen.
Question: when you talked about deterministic algorithms and machine learning, you mentioned the possibility of using machines as tools to complement our intelligence. What can you say about the possibility of maximizing resources before creating a powerful AI, or even about "putting" the human brain into a computer?

Kasparov: I do not hesitate to admit my ignorance when I'm not sure that I am not able to answer the question correctly. I try my best to understand what the human brain is, if we look at it separately from the human body, what functions it performs. Because it is difficult to imagine how the brain will behave separately from the body. It may be possible to carry out such an experiment in the future, but I’m sure that the combination of the human brain, human feelings and emotions with a computer will form a “mind” that will be much more effective than an extracted and frozen brain used as a device filled with neurons.

Question: Is there a universal basic approach to the problem of replacing human jobs with computers?

Kasparov: I think this is a very important question, because it is clear that we are approaching the moment when many people can be left without work. This is a paradox of technological progress: on the one hand, we have the latest technologies that provide enormous competitive advantages to the young generation who deal with these devices and technologies. On the other hand, we have made progress in medicine and healthy nutrition, which prolongs human life and gives people the ability to work for many years. In this sense, the generation of the 50s, 60s or even 40s cannot compete with today's youth. We must find a solution to this paradoxical situation where the generation gap is so wide. Historical experience says that such a gap always leads to a big bang.

This is a problem that politicians prefer to put off until the next election. No one wants to talk about it because it is a painful issue. It’s very easy to print money, hoping that someone will pay for it sometime in the future. So in this area, there are many paradoxes, for example, the accumulation of debts for providing social guarantees to the older generation in the expectation that the burden of paying these debts will fall on the shoulders of the younger generation. There are many questions for which I have no answers, and many questions that I could ask, I hope that AI will help me with this.
It is very bad that for decades politicians have been trying to ignore the problems that we have just discussed. They are always ready for statements, they always have plans, but they do not want to understand the counterproductiveness of hushing up the problem of the conflict of technology and society. Thank you for your attention!

Thank you for staying with us. Do you like our articles? Want to see more interesting materials? Support us by placing an order or recommending it to your friends, a 30% discount for Habr users on a unique analogue of entry-level servers that we invented for you: The whole truth about VPS (KVM) E5-2650 v4 (6 Cores) 10GB DDR4 240GB SSD 1Gbps from $ 20 or how to divide the server? (options are available with RAID1 and RAID10, up to 24 cores and up to 40GB DDR4).

Dell R730xd 2 times cheaper? Only we have 2 x Intel TetraDeca-Core Xeon 2x E5-2697v3 2.6GHz 14C 64GB DDR4 4x960GB SSD 1Gbps 100 TV from $ 199 in the Netherlands! Dell R420 - 2x E5-2430 2.2Ghz 6C 128GB DDR3 2x960GB SSD 1Gbps 100TB - from $ 99! Read aboutHow to build the infrastructure of the building. class c using Dell R730xd E5-2650 v4 servers costing 9,000 euros for a penny?

Also popular now: