He slaps the air as he speaks, and is outraged when questioned by an idea he is as sure of as Marvin Minsky’s name. He has arrived 45 minutes before the agreed time because his wife wants some free time to see El Escorial, but he lingers on each answer, he doesn’t seem to care about the clock. She stares behind a column and from time to time leans on her cane to change position, as if she had chosen the role of an old companion who follows her husband wherever he goes and continues to enjoy his digressions on artificial intelligence, intelligence human, cybernetics, science fiction, neurology, robots … They look like a couple of mature tourists, although their vitality corresponds more to that of a couple of youngsters. They have just participated in a neuroscience congress sponsored by the Complutense University of Madrid and the Technological Institute of the Canary Islands, they have barely had time to land from their flight from the United States and they are already confirming their reservations to travel to Italy this afternoon, not before having walked through the merciless slopes of San Lorenzo de El Escorial.
–How old are they?
-Well, our age is something that changes every moment. It is not worth reviewing. When I was a young researcher and started studying, my teachers were 50 years old. Then I thought: How is it possible to live half a century and still have clear ideas? Well now I know how …
–Yes, of course, and some of his ideas are very controversial …
-I do not care. I do not intend to convince anyone, I like to be controversial. Not always the best scientific ideas are the most common.
–For example, you are critical when assessing the current state of artificial intelligence.
-Yes, because the only thing we have today are capable machines, but not intelligent ones. They are not capable of learning in a million different ways, as humans do.
–Human beings learn from mistakes. The machines too?
-Yes, of course, a machine must learn from its failures. To make a machine progress it is not much use to tell it that something is correct. If you have, you know how to do it. But if she does something wrong and you tell her, then she changes. Most of the time, mistakes are better to learn than successes.
–There are people who think that even if we get a machine to do the same as a man, that cannot be considered intelligence. What difference will there be between artificial intelligence and human intelligence?
-I don’t see that there is any difference. Machines are still not very good at problem solving, but I think human beings do amazing things – just because of the way our brains work. And we can simulate with computers everything we know about its operation. So there will be no difference.
–Can we get a perfect simulation, without any flaws?
-Why not. If a difference is found, all you have to do is create more powerful programs.
–Even having the consciousness of being smart …
-I think consciousness is something simple, intelligence is more complicated. We can make very dumb but very conscious computers. If being aware of the actions means remembering what has been done, that is easy to do. But intelligence is made up of hundreds of other factors. Still, there are people who mistakenly believe that consciousness is the fundamental building block of intelligence.
–One of those who think so is the famous physicist Roger Penrose.
-Yes. I think he has no idea what consciousness is. Of course, he has a lot of verbiage, but he doesn’t make helpful suggestions.
–What about the emotions? Do they have something to do with intelligence?
-The same thing happens with consciousness: they are too simple. Emotions are nothing more than a concrete way of solving problems. For example, when one chooses to be angry it is to solve a problem very quickly and to let go. You don’t think about the cost, you don’t care if you hurt other people. It is a very primitive reaction. A mouse may have stronger emotions than a person, but it is not intelligent. Including emotions in a machine would solve nothing. Surely it happens to you like everyone else, who believes that emotions are a great mystery. But no, the thought is the real mystery, the emotions are nonsense.
–We can say the same, then, about creativity …
-Yes. It is easy to build a machine that is capable of creating all possible solutions to a problem. The difficult thing is that she does not do useless things, that she only chooses the right solution. Selection is important, not creation.
–For putting intelligence on such a high altar and believing that it all boils down to a programming problem, some accuse him of being too reductionist …
-Well, that may be an insult to them. For me, being a reductionist is good. They would insult me if they thought I was superstitious or a fanatic … Science is understanding and teaching how complex things can be done with simple solutions. People who think that I cannot do that is because they believe that there is something magical that cannot be explained. For me, that would be the insult.
–Do you think machines will one day be able to do science on their own?
-Yes, I believe that machines will be able to do whatever people do, because people are nothing more than machines.
–And what will the machines investigate then?
-They will be very intelligent and we cannot even imagine what discoveries they will make.
–What exactly are you, a philosopher, a scientist, an artist?
-I’m a kind of theoretical psychologist. But this title is not ideal for people to think that you do something truly avant-garde.
–When I was a kid I read a lot of science fiction …
-Yes, and also science.
–And do you think your work has something to do with science fiction?
-I think science fiction writers were the best thinkers about the future. I think ordinary fiction is dead, it always seems the same to me. But science fiction is about a world that is constantly changing.
–So today’s scientists have to learn anything from writers?
-Yes, many good scientists have learned a lot from literature. But beware, there is good science fiction and bad science fiction.
–Is it fiction to think that in the future intelligent machines will be as everyday as personal computers are now?
-No, it is not. And I’ll tell you more: we will use them to improve our brains. Today we are limited beings, we have a single brain with a certain speed and with a specific memory. In the future, there will be stores to recharge our brain, just as we do now with PCs. There will be a machine that analyzes the intelligence of each one and gives him the right product.
–And that day we will have achieved immortality …
-Yes. I believe that some may choose to renew their intelligence eternally and be immortal and others will prefer to die.
–And what will you do?
-I would like to be immortal, why not. I have a lot of work to do and I am too busy to die. Only those who have nothing to do in life want to die.
–Wouldn’t it be dangerous for machines to control the world, for us to depend on them to learn more, to be immortal?
-Do not. It is more dangerous not to be immortal. The alternative is for everyone to die. And the weird thing is that people settle for it. I do not understand people. Everyone seems to be happy because they are going to die, they just settle for saying, “I hope it’s not today.”
–But the world will be controlled by the machine …
-And that? Now by whom do you think it is controlled?
–By the human being?
-There is no human being. We are controlled by people, by specific individuals. That really scares me: that Hitler can get control, not a machine.
–Will there be smart and dumb machines, just as now there are smart and dumb humans?
-Yes, surely there will be a population of machines. Some of them will accidentally evolve into better forms than others.
–Then a history of machines will begin …
-There may be machines that only care about their history and others that only care about doing new things. We cannot imagine what will happen, but if a single gigantic machine were made, it would end up collapsing and it would not be interested in doing the simple things, so it would have to create smaller machines and start reproducing itself. Then the history of machines would begin.
–Do you think there is extraterrestrial intelligence?
-It’s very likely, but we still can’t know. Imagine: two hundred years ago, we didn’t have a radio and previously remote humans couldn’t even communicate. In just one hundred years science has evolved incredibly. Perhaps in another hundred we will be able to create computers more powerful than our brains and to replace ourselves with children more intelligent than ourselves. The probability that we can now communicate with extraterrestrial beings is small, but who knows what may happen in the future.
–What, according to you, is the most stupid manifestation of human intelligence?
-The sport. How is it possible that there are machines as wonderful as the brain, with billions of neurons capable of producing thousands of synapses each, wasted in the useless function of tossing balls into a hoop?
–And the most intelligent?
-Well, it is difficult, but I think that dedicating oneself to making theories about science and trying to build machines and discover new things. That has the best future.
–And music…? Are you still composing?
-Sometimes, but only when I can’t think of something better to do. As soon as a good idea comes to mind, the music seems empty.
And Marvin Minsky goes directly to the piano that adorns the hotel hall. He must not have anything better to think about, because he begins to play a melody with melancholic overtones. Emptying his mind of ideas, he waits for his wife to arrive, with whom he wants to enjoy a quick walk on the slopes of San Lorenzo de El Escorial.
Jorge Mayor
This interview was published in October 1996, in number 185 of VERY Interesting
