Igor Aleksander

Speaker 30:00

Hello, I'm Kirstie Young, and this is a podcast from the Desert Island Discs Archive. For rights reasons we've had to shorten the music. The programme was originally broadcast in nineteen ninety nine, and the presenter was Sue Lawley.

Presenter0:31

My castaway this week is an engineer. He doesn't build bridges or power stations, but machines that can work like the brain. He was born in the former Yugoslavia, but fled from the Nazis and ended up in South Africa studying electrical engineering. In the late fifties he came to Britain, where working on what was regarded as the lunatic fringe of computer science, he developed Wizard, a machine which could recognize patterns or faces in a crowd.

Presenter0:58

Then in nineteen ninety came Magnus, a computer which could think, a form of artificial consciousness. Inevitably his work has attracted criticism, but he doesn't mind the flack and refuses to bow to the prevailing British view, which he says wants engineers to know their place and stick to fixing washing machines. He is the Garbor Professor of Electrical Engineering at Imperial College London, Igor Alexander. Would you be any good, I wonder, Professor at fixing washing machines?

Igor Aleksander1:25

I've always thought that perhaps it's something I should have learnt, and I'd be a much richer person these days if if I could fix a washing machine. Any attempts I've had, as it were, were absolutely hopeless.

Presenter1:37

What you do, though, effectively is is cross the line. You move from your electrical engineering into neuroscience, and that's what upsets people, isn't it?

Igor Aleksander1:45

I think it's also to do with the fact that most of the people who work in things like consciousness, psychology, neurobiology feel that they don't really want to mess with mathematics and engineering. But in fact, the systems that they're dealing with are enormously complex. And if there's anything that an engineer learns, it's how to deal with complexity by building things that are complex. And so I make no apologies for crossing that line.

Presenter2:13

That's more exciting than that.

Igor Aleksander2:14

Absolutely, and working on the fringe between various fields is the most exciting of the lot.

Presenter2:20

But it's also got something to do with with the nature of you, I suspect, because I read that you even got bored with computers back in the sixties when they were just beginning to take off. You said, Oh, this is a limited business.

Igor Aleksander2:30

Well, computers are enormously boring if all if if your entire life has to do with programming computers. So I tend to use computers as a tool, and that's uh that's the sort of disregard one should have of them.

Presenter2:42

But the the difference between the computers that we know on our desktops and the computers that you deal with is that the one is limited and the other isn't.

Igor Aleksander2:53

Well, I think they're they're probably both limited, but the computers that everybody knows are programmed, and whatever they do has to be dreamt up by a programmer beforehand. I like working with computers that can develop their own ways of doing things, and develop their own programmes, if you like, which in a sense learn from the world, in a sense build up their own experience.

Presenter3:16

So if you were talking about playing chess with such a computer, an ordinary computer simply plays it in as far as it's programmed to play it. Your computer might just be able to extrapolate and think for itself.

Igor Aleksander3:27

It it would try and build up its own experience in chess and probably would end up playing a lousy game of chess, but uh it may be able to do one or two other things which are more interesting than chess.

Presenter3:39

Can it talk to you?

Igor Aleksander3:40

We're very much interested in computers that can communicate with us. They probably at the moment do this in writing rather than speaking. But the the whole question of language is is enormously interesting.

Presenter3:55

So in a very, very broad and I know I'm oversimplifying here, but in a very broad and primitive form, it is consciousness.

Igor Aleksander4:04

Yeah.

Igor Aleksander4:05

It is consciousness, but I like to call it artificial consciousness because it's about as different from real consciousness as an artificial arm is from a real arm. But it does tell us quite a lot about real consciousness, and that's the object of the exercise.

Presenter4:23

I'll tell you something before we start. You can't take it to your desert island. Tell me about your first desert island is.

Igor Aleksander4:28

Well, I would like to listen to Arturo Benedetti Michelangeli playing the second movement of Beethoven's piano concerto. This is because at an early age in South Africa I met him. He came to our house when I was showing some films of Laurel and Hardy, and this is what he really wanted to see. But I was impressed by his enormous hands. They would stretch for about a foot when he stretched them out. And it's only later that I learned how well he could play the piano.

Presenter5:54

Arturo Benedetti Michelangeli and the Vienna Symphony are playing part of the second movement of Beethoven's piano concerto, No. five, The Emperor, conducted by Carlo Maria Giolini.

Presenter6:04

Magnus, then, Igor Alexander, can imagine things. What kinds of things can it imagine?

Igor Aleksander6:13

What we're trying to do is to replicate the ability of the brain to have visualization, imagination. If I were to say, think of a blue banana with red spots, no one would have difficulty in doing this as a result of putting together one's experience of a banana, one's experience of blue and red. But the way that a computer might do that or an artificial device might do that is quite complicated. Our brains are made of masses of cells, ten billion cells, which somehow or other talk to one another in order to create these experiences for us. And I'm trying to see some of the mechanics of how this might happen and do this on a computer.

Speaker 46:43

Hmm.

Presenter7:01

And what's the power of Magnus in comparison with the human brain, then?

Igor Aleksander7:04

Oh, it's absolutely tiny. The human brain if you imagine the main part of the human brain as an enormous handkerchief about one metre square or one metre by one metre, the size of Magnus is about half a thumbnail with respect to that. But what I find absolutely fascinating is that that half a thumbnail can do quite sophisticated things.

Presenter7:27

And the principle you believe is the same as the principle on which our brain works?

Igor Aleksander7:31

Well, I think so, but I still need to prove that.

Presenter7:35

You've got to guess first, I suppose, and then try to prove your guess is correct.

Igor Aleksander7:38

That's quite right. And I think we also need to prove it from a rigorous mathematical point of view.

Presenter7:44

That of course can explain intelligence, the ability to to rationalize, to reason. But what about emotion, instinct, feeling, humour? Can it will it, do you believe, in the end account for all of those two?

Igor Aleksander7:58

Probably not. The thing that one must distinguish very clearly is a living object with an object like Magnus, which is just silicon inside a computer, if you like. Now, the consciousness of a living object is the consciousness of a living object. The best that Magnus will ever be able to do is to be conscious of being a machine. So if I tell it a joke, what it should do is come back to me and say, okay, I understand that you think that joke's funny, but it doesn't mean a thing to me. When it comes to emotions, I think things aren't quite that pessimistic, because there are theories about emotions in ourselves which are quite visceral. You know, we feel things inside our stomach, or the hairs on the back of our neck stand up. Now, many people feel that our more subtle emotions are reactions to that. Now, those visceral things can actually be reproduced in Magnus, and those are the things we're studying. This is an early memory of Verdi's Rigoletto. I used to be asked when my parents used to wind up their gramophone what I'd like to hear at about the age of three. And I'd always say rigoletto, I think because I liked the word rather than the music. But I then learned to sing Bell'Aphilia de la More, sciavocon de vetzitoye, which means beautiful daughter of love, I'm a slave to your wishes. And that used to send people in paroxysm of laughter. And I couldn't quite understand why, but I do understand it now.

Igor Aleksander9:47

Hello.

Presenter10:36

Ben Yamino Gigilli, with Amelita Gallicorci, Luis Homer, and Giuseppe DeLucca, singing part of the Quartette Bell'Aphilia della More from Act three of Verdi's Rigoletto, with the Metropolitan Opera Orchestra conducted by Walter Gohr, and that was recorded in nineteen twenty seven.

Presenter10:53

The um the first remarkable thing about your early years, Professor, is that you you didn't go to school, um which obviously didn't do you a lot of harm. No school, because you were a Yugoslav refugee on the run from the Nazis. How much do you remember about that?

Igor Aleksander11:07

I remember.

Igor Aleksander11:09

the more emotional occasions of that period. But by various routes we ran from Yugoslavia and we all met up outside that beautiful town of Perugia in Italy.

Presenter11:24

But you were in a a a form of concentration camp at some point, weren't you?

Igor Aleksander11:27

Uh very briefly um my mother tried to smuggle my sisters and myself out of Yugoslavia, and it didn't work. The false papers were discovered and we ended up in a holding camp. I wouldn't describe it as a concentration camp, but certainly that would have been the next step had we not been helped by old friends in Italy.

Presenter11:49

Italian royal family, I read.

Igor Aleksander11:51

They were people who had influence through the royal family rather than through Mussolini, which was enormously lucky.

Presenter12:00

What about your father?

Igor Aleksander12:01

My father was put into what was essentially a prisoner of war camp on an island off the coast of Yugoslavia. But being a a good swimmer he managed to escape. He he just swam off. I'm not terribly clear how this happened, but eventually he joined us in Italy.

Presenter12:20

So you you all met up again near Perugia, where where you live lovely countryside in of of Umbria and so on.

Igor Aleksander12:23

Yes, indeed.

Igor Aleksander12:27

Indeed. Yeah.

Presenter12:27

The next thing that happened was the British came and and bombed you.

Igor Aleksander12:31

Yes, my parents actually weren't with us because they had to keep on running and they went to hide on top of a mountain in the Abruzzi in Italy. But our house, where we were as children with my grandmother and some aunts, was bombed by the British. But in a very gentlemanly sort of fashion, it was artillery rather than aeroplane. And on the morning of the bombing, there were a couple of British soldiers who came to see the house and they said, Look, our governor has got it into his head that he wants to knock this house out because it's getting in the way of our advance. So can I have a look at your cellar? And he did, and he said, You'll be all right in there. Just go into the cellar and we'll knock your house out, but you should be all right.

Presenter13:16

That's what they did.

Igor Aleksander13:16

And that's what happened.

Presenter13:18

But your parents must have come back and found the house flattened.

Igor Aleksander13:21

Certainly my mother, when she came over the hill, they they joined the Eighth Army that was advancing and my mother was an interpreter. And she came over the hill and discovered that the house was flattened. This must have been a traumatic time for her.

Presenter13:35

It's not surprising, then, that your father thought he'd had enough of Europe and he'd take you all to South Africa. What what language were you speaking by this time?

Igor Aleksander13:43

I was fluent in Italian, I could speak Yugoslav, and I could vaguely understand German, because that's the language my parents spoke when they didn't want us to understand, but certainly no English.

Presenter13:56

English when you got to South Africa.

Igor Aleksander13:57

Yes, indeed.

Presenter13:59

Record number three.

Igor Aleksander14:01

This is the first record I ever bought. I started to take notice of music at the age of about twelve or thirteen, and it's Les Paul playing Little Rock Getaway.

Presenter14:42

Les Paul playing Little Rock Getaway, and that was uh re recorded, I think, in nineteen sixty seven. Um you apparently had a talent for magic in your youth, Professor. You joined the magic circle. What sort of tricks did you do?

Igor Aleksander14:54

Happy.

Igor Aleksander14:55

I cut a girlfriend in half in front of an audience of other school kids. With my mother, we mimicked what a group of people called the Piddingtons were doing. One would go up in an aeroplane and the other would be down at the bottom of the sea. And one would say, What am I holding in my hand? and the other would guess it. And my mother and I did that. Unfortunately, someone called a journalist in to look at this, and we got into the newspapers, which I'm still a bit ashamed about.

Presenter15:30

But obviously you had a code.

Igor Aleksander15:32

I developed a mathematical code because it was obviously a mathematical trick. The question was asked in a particular way and that could be could encode the answer.

Presenter15:32

Uh

Presenter15:43

What about apartheid in South Africa? It would have been in its infancy when you were in your youth. How did you react?

Igor Aleksander15:50

I only

Igor Aleksander15:52

I became aware of apartheid when I was at university in South Africa and the university of the Vitvatusrant where I was educated did not have an apartheid policy and it never did but that got it into a lot of trouble and I think that while I was not discouraged from going on marches it wouldn't have been too good for my future.

Presenter16:18

Did you in effect then have to choose between being politically active and pursuing your career?

Igor Aleksander16:25

Yes, um

Igor Aleksander16:27

The two came together in a way, because I wanted to pursue my my um career and the opportunity for doing that was was to go to the UK. And that in a way, in a cowardly way, resolved resolved the problem.

Presenter16:43

Now the other influence on you um academically was um a man called Colin Cherry. That was he was really the conduit, the reason you came out and came here, isn't he?

Igor Aleksander16:53

Yes, indeed. Colin Cherry was a professor at Imperial College and he came to South Africa when I was coming to the end of my degree. Four years of engineering before computers were invented, before even transistors were invented, wasn't all that exciting and I was wondering whether I really wanted to be an engineer all my life. But Colin Cherry came to South Africa and his main message was that as an engineer one can contribute to the life sciences and study things like brains and that was a wonderful, wonderful sort of idea and I've never never moved away from that idea.

Presenter17:35

Record number four.

Igor Aleksander17:37

During my university time I played the drums, and one of the things to do was to play just piano and drums in smoky nightclubs. And one thing that's going to remind me of this is the Oscar Peterson trio playing Georgia on My Mind.

Presenter18:26

The Oscar Peterson trio playing Georgia on my mind. So um back in the sixties, being bored with what other people were doing, which was ordinary computers, um you went to this so called lunatic fringe of computer science, artificial neural machines.

Presenter18:41

You obviously have to get grants for this kind of work. Did did industry like what you were doing?

Igor Aleksander18:46

I don't think anybody's ever liked what I what I was doing, um but um certainly I I wasn't too unsuccessful in getting grants. I I think there were grants given to me to humour me.

Presenter18:59

So you were humoured, were you, when you created silicon neurons?

Igor Aleksander19:03

Uh certainly, um people believed that um the computer as it was then in the very early sixties was the way it was going to stay and uh its memory was largely created out of um magnetic materials. What I was proposing was to use a silicon neuron which had memory in its own right and I went to the Plessy company and asked them to fund this. Uh they said yes, they would, but uh it's a crazy idea because of course when you switch the power off they lose their memory.

Igor Aleksander19:37

Of course this is what much later turned out to be the random access memory, and anyone who has ever owned a computer will know that you can rush out to the high street and buy megabytes of the stuff now. Buy some more RAM. Lots of RAM, without which your computer couldn't work.

Presenter19:49

Buy some more

Igor Aleksander19:54

And so you

Presenter19:54

So you were ahead of the game, effectively?

Igor Aleksander19:56

Uh

Presenter19:56

Uh

Igor Aleksander19:56

Uh we were ahead of the game, but

Igor Aleksander19:59

The developments in the United States of RAM had absolutely nothing to do with our work, but we we did make a machine out of the first RAM in existence.

Presenter20:10

Then eventually in the nineteen seventies, I think, when you were at Brunel, you designed and built Wizard, which I referred to earlier.

Presenter20:16

What what could it do that no machine had done before?

Igor Aleksander20:20

Well, in a sense, I was under pressure to do something useful instead of messing about with things like uh thinking and artificial intelligence. In order to prove the fact that making neural systems had some applications, we built Wizard, which is a pattern a mindless pattern recognition machine. But it turned out to be useful. It could recognize intruders in airfields, it could find bad banknotes amongst good ones, it could find whether a cherry had been put on top of an ice cream in a production line. So it did turn out to be a very, very useful machine, and some versions of it are still working in various parts of the world.

Presenter21:00

But it simply recognised patterns and therefore, again, it was limited.

Igor Aleksander21:05

It was limited to labeling patterns. It would receive a pattern and say this is an elephant or this is a cow, but it had no concept of what a cow or an elephant looked like.

Presenter21:16

But people were very impressed by it at the time, weren't they?

Igor Aleksander21:18

It was useful and it it was then taken up by a commercial company who made a few. I don't think they made the fortune out of it, but it was the beginning of looking at neural networks as a as a way of making useful machinery. Next record. The next record is from a somewhat turbulent time in my life when I was listening to a lot of Mahler and it's the Adago from Mahler's Symphony number five.

Presenter22:49

Part of the Adagietto from Mahler's Symphony No. V, played by the New York Philharmonic conducted by Bruno Walter.

Presenter22:55

Magnus then, Igor Alexander, some eighteen years on from Wizard, uh you designed and and built that. It was a huge breakthrough um built at Imperial College.

Presenter23:07

It can learn, it can imagine,

Presenter23:10

Therefore it has an inner state.

Igor Aleksander23:13

That's quite right. That was the main difference between Magnus and Wizard, that because the neurons in Magnus talk to one another, they end up saying things to one another which Magnus doesn't actually present to the rest of the world. So they that's where the inner state comes from.

Presenter23:33

And when you switch it off or take away its inputs or whatever you do, do does it dream?

Igor Aleksander23:38

The talking can carry on and in a sense that is a bit a bit like dreaming, or uh if it doesn't carry on it's like dreamless sleep. But there are some very interesting models of what happens not only in dreaming but also in anesthesia, which is uh something we don't know a great deal about, and Magnus is quite uh quite clear about what's happening.

Presenter24:02

But when you switch it on, can you find out what it's been dreaming about?

Igor Aleksander24:06

That's the cute thing about Magnus. We can always tell what it's dreaming about, what it's thinking, because it appears on a screen, which is not something we can do with human beings during a brain operation.

Presenter24:18

But effectively y you you gain from it a model of of us to a small extent. You you can find out what is happening to us when we're asleep or when we're dreaming.

Igor Aleksander24:28

That's the major reason for having Magnus, is to check hypotheses about what happens to us when we're sleeping and dreaming. What is it that could be happening in our heads? And that's what we look to Magnus for. Certainly some structures of Magnus don't dream and don't sleep, and some can't even build up a visual imagination. And that's the interesting part, which do and which don't.

Presenter24:55

What else can you uh learn? What else are its applications? Presumably, if you can watch how it learns, you can learn how we learn.

Igor Aleksander25:03

That's its main application. Knowing how our brains create our consciousness is the biggest problem in science at the moment. And working with this problem in an artificial domain gives us insights that you just can't have by performing operations or just listening to human beings speak or even scanning their brains. So my greatest hope for Magnus is that it will tell us a lot more than we know already about our own brains.

Presenter25:35

It would tell us about uh the problems of our own brains, like dyslexia, for example.

Igor Aleksander25:40

Certainly the nature of dyslexia is uh something that we're interested in. What happens under anesthesia is something we're interested in. What happens with distortions of consciousness, lesions in the brain?

Presenter25:53

as in Alzheimer's for example.

Igor Aleksander25:54

Alzheimer's, Parkinson's and many other lesions that can happen in the brain as a result of physical occurrences.

Presenter26:04

Bye.

Igor Aleksander26:04

Straight hold.

Igor Aleksander26:06

comes from the fact that I love going to Greece and uh Helen and I go every year uh and so we were minded to learn Greek and we went to evening classes and met a lot of very interesting people, amongst whom uh a young singer who spent some time in the UK by the name of Savina Yanatu and she's now made it big in Greece, but I'd like to hear her sing uh something called Vradi evening.

Speaker 426:35

The bit like capable pastor.

Speaker 426:39

Staniksiatikovi.

Speaker 426:44

Nyo Hima Christmay.

Speaker 426:50

Tyruchi Polojamiton Rovon Tashi.

Speaker 426:56

She there is again.

Presenter27:02

VRADI evening sung by Savina Yanatu and composed by Karyotakis. You obviously, um, Professor, despite your very rational Scientific view of the world, have a great sense of and feeling for its beauty. It's an important is that part of the joy of your consciousness?

Igor Aleksander27:20

It it certainly is. Um sometimes even the most advanced research feels like hard work and I like to g get away into the countryside, whether it be in England or Greece or Italy or France, you know, on the sides of the the shores of the Mediterranean with a nice sunset. Uh one one does have better thoughts, I think.

Speaker 427:35

Hmm.

Presenter27:44

And and when you have some of those thoughts and you're contemplating it all, do you ever have any fears? The sort of fears you know that a lot of people do give voice to, that the earth one day, that all of this beauty we're talking about could be destroyed by by some intelligence greater than our own that you may be sowing the seeds of?

Igor Aleksander28:04

My fears have more to do with uh the intelligences that we have around at the moment that seem to be doing quite a good job of destroying uh large chunks of the world and uh

Igor Aleksander28:15

taking away some some of the beauty.

Presenter28:17

No, it's not totally incredible, is it? Uh the idea we we know f for example at the moment that viruses get into our computers and uh communicate with each other, as they well communicate from one to the other.

Presenter28:29

It's not totally beyond belief that if you do create some kind of artificial intelligence and it began to link up and create its own web, that it could.

Presenter28:39

Take over.

Igor Aleksander28:40

I don't think that Magnus will ever have a progeny that'll get up and want to take over the world. I think that's a misunderstanding. People do that. The sort of real evil in the world, I do believe, has to do with living consciousness rather than artificial consciousness.

Presenter28:55

But a sophisticated version of what you're creating could sh and there are serious minded people who do genuinely fear this, that, for example, the computer that controls the store of biological weapons could be communicated with and told to release those weapons, for example.

Igor Aleksander29:13

I think that's true, but that happens as a result of artificial stupidity in those computers. Let's say you have a robot driving a car. I would say that making it conscious would make it safer rather than if it weren't conscious. You know, would you like an unconscious computer to drive your car? W would you like it to have some consciousness? So I actually see the introduction of consciousness into machinery as being something that's going to improve the way that we can interact and control machinery rather than it rushing off on its own.

Igor Aleksander29:50

More music.

Speaker 430:26

Oh your stay.

Speaker 430:33

With Lord's Happy Horn Blood is left with

Presenter32:03

Elizabeth Schwarzkopf and Janet Baker with the Philharmonia Chorus singing part of the Annus Dei from Verdi's Requiem, conducted by Carlo Maria Giulini.

Presenter32:12

So, um, Igor Alexander, as a man who likes the unknown, you could be quite amused on a desert island.

Igor Aleksander32:19

I think I'd be quite practical on a on a desert island, but I think I'd miss people like crazy, particularly the nearest and dearest.

Presenter32:27

So you would be but you could fend for yourself, you could build the shelter.

Igor Aleksander32:30

I I think so, yes. I think one of the things I try and build is a cafe,'cause uh, you know, being a Central European at heart, the one thing I like doing is going into cafes to do my work, so I'd have to build a an espresso machine out of bamboo or something.

Presenter32:47

You might find the odd coffee bean hanging around. You never know.

Igor Aleksander32:49

Uh yes, or anything. I don't mind the coffee all that much, as long as I have cardboard cutouts there to create the right atmosphere.

Presenter32:58

So you're happy to live in a virtual world?

Presenter33:01

Ugh.

Igor Aleksander33:03

I could probably persuade myself that it's the next best thing to start with.

Presenter33:06

But could you save yourself from going mad? I mean, after all, what we're talking about, what your whole subject is, is how to control the mind. Could you control your own in this situation?

Igor Aleksander33:14

It wouldn't be my technical knowledge of the mine that would lead to that. I think it would have something to do with my personality, and I'm not sure, quite honestly. I think I might quietly go mad.

Igor Aleksander33:27

Last record.

Igor Aleksander33:28

A couple of years ago my college celebrated jointly with the École Polytechnique in Paris their two hundredth centenary. And it was a whole day of celebrations and I gave a lecture in the afternoon. And in the evening there was a concert where Rostropovich conducted the Shostakovich first violin concerto and Wengerov played it. The interesting thing about that was that at a dinner afterwards, because the organizers thought I was Russian, they sat me next to the performance and I had a a wonderful discussion with Rostropovich about how he thinks the mind works and how he makes contact with everyone in his audience when he's playing the cello.

Presenter35:03

Maxim Wengerov playing part of Shostakovich's first violin concerto with the London Symphony Orchestra conducted by Missitslav Rostropovich. If you could only take one of those records, Professor?

Igor Aleksander35:15

It would have to be the Verdi Requiem. There are more things to be discovered in that than I've had time for so far. What about your book? I'd like Richard Gregory's Companion to the Mind. That's a book that's been an absolute Bible to me. But again, I haven't had time to look at its more interesting corners.

Presenter35:37

Annual luxury.

Igor Aleksander35:39

My luxury I'm afraid I'm going to be a bit nerdy about this and ask for a virtual reality version of the London Symphony Orchestra, which would respond to my conducting movement. I'm a sort of conductor manque, and uh I'd love to be able to conduct the London Symphony Orchestra.

Presenter35:58

Igor Alexander, thank you very much indeed for letting us hear your desert island discs.

Igor Aleksander36:02

It was a pleasure, thank you.

Speaker 336:12

You've been listening to a podcast from the Desert Island Discs archive. For more podcasts, please visit bbc.co. uk slash radio four.