Sir Clive Sinclair: views on AI and super intelligence

Sir Clive Sinclair (1940-2021)  had an impressive portfolio of inventions to his name, including a matchbox radio, a pocket calculator, a single-chip digital watch, pocket television and the abortive C5 electric car. Beyond those, of course, his name is synonymous with the ZX family of computers – ZX80, ZX81 and the wonderful ZX Spectrum. Let’s not dwell on the QL.

His company, Sinclair Research, spun off trading companies in laptop computers and satellite receivers (Cambridge Computer), wafer-scale integration (Anamartic) and pocket telephones (Shaye Electronics). His later research pursuits included fast CPUs and ultra-lightweight pedal bicycles — and he hoped to return to electric vehicles eventually. 
In our 1990 interview, Clive Sinclair described his very fast central processing unit as ‘slightly like the transputer in concept, but better in detail design’. So it could be used singly in smaller computers, and in parallel arrays in larger systems. Read the full 1990 interview below.

Sir Clive Sinclair: views on AI and super intelligence

Clive Sinclair describes his very fast central processing unit, now under development, as ‘slightly like the transputer in concept, but better in detail design’. So it could be used singly in smaller computers, and in parallel arrays in larger systems.

‘All this is part of a picture’, he says. ‘I decided at the beginning of the 1980s that parallelism was the way of the future — hardly an original idea, of course — and that in order to achieve it I needed to have wafer-scale integration, hence Anamartic. As far as I’m concerned wafer-scale integration is a means to an end; as far as Anamartic is concerned it’s an end in itself, because it’s their product. But my interest is in making computers that use wafers, where each wafer contains possibly hundreds of processors, because I don’t believe you can achieve high levels of parallelism readily without some such technology because the interconnect problem becomes overwhelming.’

Sir Clive’s interest in large-scale computers, again, is not an end in itself. It stems from his belief that we must turn to artificial intelligence if we are to continue to increase our wealth in society. We have done this over the past one hundred years by increasing the efficiency of the manufacturing industry, so freeing people to move into service industries and adding to the gross national product, he argues; now we are running out of people and soon we shall need some means of supplementing and complementing people’s intelligence by artificial means.

First, he envisages the widespread adoption of expert-system-like aids to, for example, teachers or doctors. Beyond that, he says, ‘What I see in the longer term is fully automatic teachers and doctors. I look forward to the day when people have in their own homes effectively a doctor, an electronic doctor, and they can go down and consult it. I think you would probably see an animated face and you would talk to him.’

Clearly Sir Clive is assuming that AI technology will have advanced to the point where ordinary speech communication between person and computer is a reality, and that the systems will be able to hold and access comprehensive knowledge of the subject. But his example raises the key question of whether there is a limit to the ‘intelligence’ that a computer system can possess — and Sir Clive is emphatic that there is no such limit. (And, indeed, he would not put ‘intelligence’ of the artificial kind in quotation marks.)

Sir Clive has no doubt that, at some stage, computers will be able to ‘understand’ what they are told, though he prefers the term ‘comprehend’. To him this is undeniable: human beings do it, human beings are machines, therefore machines can do it.

‘I’m simply saying that there’s nothing a human being can do that a machine can’t do, on the simple grounds that a human being is fundamentally a machine, unless you believe in some supernatural properties. If you do then it’s possible you can argue, but if you don’t then it’s unarguable. Human beings exist, so in principle other machines can be made which will do the same job. That’s not to say we will succeed in doing so, but I believe we will. Certainly it’s theoretically possible.’

As for the likely timescale for these ‘understanding’ computers, Sir Clive says: ‘Early in the next century we’ll be able to make machines of a complexity which is of the order of the human brain. Roughly that sort of complexity will be possible economically. It’ll begin early in the next century. That’s not to say we’ll then have machines that will rival human beings, that’s a very different matter indeed. That might take decades to do, but what is very interesting is that we will have the capability of that complexity.’

In moving towards such systems, Sir Clive would give top priority in AI research to the crucial areas of speech recognition and syntactical analysis, As for neural networks, he doubts that that approach will take us very far. ‘It’s very interesting research, but I don’t think it’s the route forward.’

Sir Clive is impressed with Professor Donald Michic’s work in machine learning by rule induction. ‘I was very struck by that approach, and I do believe that that is how the brain must work: the human brain is not given the rules, but derives algorithms from experience.’

Sir Clive believes that the two sides of AI research, computer science and cognitive science, will come together in the future, but he does not assert that the intelligent machines of the future will work in the same way as the human mind works. ‘It’s too early to tell. There will be analogies, but I think there will probably be ways in which machines will differ quite a lot. At the very least they will simulate the way the mind works; they’ll achieve the same results from the same inputs.’
If these intelligent machines did come to pass (Sir Clive politely accepted my less than absolute conviction on this point), did he see problems arising on the human and social impacts of their use? ‘Very much so. Anything as radical as that is bound to throw up tremendous problems — sociological problems, certainly, but also philosophical ones and perhaps religious ones. People will be very much afraid of another intelligence appearing. People might well not like the idea of there being other intelligences around, especially since in due course they’ll be more intelligent than human beings.’ (I wished Sir Clive wouldn’t keep on doing things like that to me, and I indicated I had some trouble with that last throw-away line.)

‘What will happen is that we’ll make machines that are increasingly intelligent, and that process will not have an end, and so one day their intelligence will exceed ours.’

Did Sir Clive see any dangers there? ‘Oh yes, clearly. People may not like that at all. I might not like it, I don’t know. It’s a daunting prospect that we will not for very much longer be the greatest intelligence in the known universe. So there will be arguments, as with genetic research at present, on what should be done and what should be allowed. I expect there will be moves to stop the development of artificial intelligence.’

Another key question, in the ‘electronic doctor’ example and more generally, is that of responsibility. At present, medical expert systems are purely advisory, and the final decision — and the responsibility — is the human doctor’s. Surely, if the system makes the decisions, there is a problem? Sir Clive is sanguine.

‘I think that’s a problem that we’ve faced many times in our history. We faced it when we moved from policemen controlling traffic to traffic lights. We forget, but it’s quite remarkable that we go around accepting the authority of a few electric light bulbs. We faced it again when we made trains that were fully under electronic control. We now accept trains that are fully automatic — and aircraft really are. And we will one day realise that the doctor is more fallible than his electronic equivalent.

‘One day the machines will be better than doctors in all but human terms, which is a good thing because it will then enable the doctor to concentrate on giving human comfort, which is probably the most necessary thing in many cases. But for all but the most exceptional cases, your “doctor” would be at home and you would relate to it as an individual.’

Sir Clive sees his contribution to these dramatic AI developments as, at the moment, putting such tools in place as wafer-scale integration and parallel processing. One day he may start hiring AI researchers and pursuing AI science directly, but that day has yet to come.

For you

Be part of something bigger, join BCS, The Chartered Institute for IT.

Clearly Sir Clive regards artificial intelligence as the key advance in looking forward to the future of computing in general.

‘I don’t really think there’s anything dramatic that’s going to happen on the computing front that isn’t related to AI, or doesn’t stem from AI in some way or another. Obviously there’s a lot of technology coming along. One of the things that comes out of wafer-scale integration is that you can replace hard discs with solid-state discs. And the consequences of that are very great, because the increase in access speed is so huge, and when you start to imbed a little bit of intelligence in the memory, spread it around a bit, which is the beginning of AI if you like in a funny sort of way, you can start to do contextual searches with marvellous efficiency, so the computer one can see not too far away will begin to be one’s memory in a real sense.’

Computing professionals should look very closely at the implications of wafer-scale integration, and of parallelism, Sir Clive says. Two things are happening at the same time to increase computer power: one is the very rapid increase in processor clock rates, and the other is the introduction of parallelism. One big factor is multiplied by another, and this leap in processing power means that people are going to have to think very fast about these changes.

Sir Clive regrets that pioneering work in Britain on parallel computing has not been fully exploited. It is not too late to do this, he argues; the opportunities are wide open.

In the context of UK Ltd, Sir Clive says: ‘I worry greatly that our large companies in the electronics sector seem to be very unimaginative, and I think they fail the economy in terms of failing to explore the opportunities. I worry very greatly that our semiconductor industry may not progress; it’s a good one at the moment — just — but after the takeover of Plessey by GEC, I hope GEC change their general attitude and invest properly in that field, because we all depend on the semiconductor industry.’

Should the industry have incentives? Should there have been an Alvey 2 research programme? ‘I think so. People and the government were all too ready to write Alvey off. I think Alvey was a great success. It was very stimulating, got a lot of people thinking and got a lot of action happening, and the fact that it was then wound up was a very great shame. I hope the Alvey community will continue to commune, but I don’t think it really can without a focus, and Alvey was a very great and useful focus. I think it’s very sad that it wasn’t continued.

‘We lack the equivalent of MITI (the Japanese Ministry of International Trade and Industry) in this country. This totally laissez-faire attitude on the part of the government really doesn’t work. We need more competition in this country, whereas we’re tending to get less; and we need the government to give a bit of stimulus now and again. I don’t have the answers, but my suspicion is that more ventures of the Alvey sort would be highly desirable.’

Sir Clive is proud to call himself an inventor. ‘I don’t see inventors as simply people who come up with ideas, inventors have to be people who get the idea exploited. That’s the tricky bit, really. There are lots of ideas around.’ He admits that his ideas have not always been successful, but says that that is to be expected.

‘In my opinion, if you’re an inventor you darned well certainly have got to expect it. You try things out, but not everything is going to work. If anyone knew that well how everything was going to work the whole game would stop. We have to try new things.’

Sir Clive’s latest new thing is a radical form of pedal bicycle, which he hopes will emerge in the second half of next year. It is radical in being very much lighter and more readily folded than other bicycles. And he is still interested in electric-powered vehicles, though not beyond paper studies at present. ‘My original intention was to develop a full-range electric car. It’s one of the elements needed to solve the carbon dioxide problem, and I want to get back to that.’