In half a century, computers will rule the world

The line between genius and lunacy is a fine one, it is often said, and for some, Reading University's Professor Kevin Warwick is the very embodiment of that view.

Talk to his supporters and you'll be convinced that Warwick, with his all-encompassing knowledge of cybernetics, is an intellectual solar system. Countless TV appearances testify to the fact that, if nothing else, he's good for soundbites.

But talk to others and you'll believe that Warwick, with his notions that computers are set to take over the world, in the best, or worst, traditions of an Isaac Asimov novel, is nothing short of barking.

Either way, Reading's celebrated boffin has been hailed as the prophet of the robot age. Unfortunately, his prophesies don't bode too well for mankind.

When Kasparov was recently beaten by IBM's Deep Blue chess robot, it was Warwick who could be seen on our TV screens. On the one hand he was endorsing the inevitability of computer triumph, on the other predicting a nightmare scenario where, if mankind doesn't stop such machines in their tracks, it is doomed.

In his book March of the Machines, Warwick lays it on with a trowel. 'In 2050, our lives are run by machines and we must do whatever they have scheduled us for. Many humans are kept as general labourers.' And for anyone who doesn't share this Jeremiah outlook, he stresses: 'This is not science fiction.'

The upshot, according to Warwick, is that by the year 2050, we might find ourselves on the run from real-life Terminator figures, while criminals and geriatrics are deemed synonymous by the state. Just make sure you oil the wheelchair castors generously.

'We must obey the machines' wishes and live only to serve all our lives - what there is of them - under their control,' opines Warwick bleakly. His theory runs thus. Right now, at Reading, there are machines with the computational power or 'intelligence' of insects. Not the cleverest of insects, admittedly, but with around 50 model neurones - twice as clever as your average garden slug. In five years time, the university reckons to parent robots with the brain power of cats. Arm them with gas canisters and watch them give dogs a hard time.

Fast-forward another 10 to 50 years and, according to Warwick, there will be computers more intelligent than humans.

Kasparov's downfall at the hands of Deep Blue was just the start. 'Humans are far cleverer than machines now and so control them. But if computers become more intelligent than man, then why should we assume they would still be subservient?' he argues.

'Won't robots just treat us as we treat other creatures - as curiosities in a zoo, or as slaves?'

Warwick's background is sober enough: a six-year career as a BT engineer preceded a university degree and a career in academia. He says he spent his youth messing about with motorbikes, rather than burying his nose in science fiction.

Meeting Warwick is a bit of a let-down. Instead of the archetypal mad professor stumbling out of some cubby-hole, soldering iron in hand, you are confronted with a neat figure with a broad West Midlands accent.

Even more disappointingly, he prefaces his greeting by apologising for his failure to sport a tie. For a man whose eyes appear trained on the future, he displays surprising old-world charm, leaping ahead to hold doors open and apologising profusely that he can't offer lunch - 'I've got to be at Sky TV'.

In his office, where he has been plotting the theory of mankind's demise, the only evidence of computer power is a PC hooked up to Reading University's Internet site. And instead of algorithms, charts and dire warnings, the walls are full of photos of his two children, Maddi and James. In a cabinet sits an unopened half-litre can of Trent Bitter.

'It doesn't look that threatening,' I say, nodding in the direction of the PC sitting mutely on a desk. 'And even if it was, couldn't it just be switched off?'

'But what if it had arms, legs, sensors and a machine-gun?' he counters. 'What if it wouldn't let you switch it off? What if it was programmed to shoot you if you tried? What if it was looking after missiles and military defence systems?'

But surely nobody would be mad enough to build such a computer? 'Surely nobody would want to build a bomb which when it exploded would wipe out a city?' is his entirely reasonable response.

Having pontificated on man's stupidity, Warwick turns to the intelligence capabilities of the robots which have been built at Reading. Nestling under his desk in a silver case are two of the university's 'seven dwarves'.

These are essentially programmed mini-robots built in the style of insects. They have a neural network modelled on, and roughly equal in 'intelligence' to, a gnat. They have six ultrasonic 'eyes' and are entirely self-sufficient, with a bundle of microprocessors attached to their backs, a rechargeable battery on their stomach and a built-in program of self-preservation.

They are coded differently to produce different behaviour. One, which Warwick cheerfully demonstrates on his office carpet, is programmed to protect itself by judging the distance between it and an object, and retaining that distance.

'If that was a dog, you'd call it intelligent,' says Warwick as he leaps about his office, the mini-robot's wheels trundling along a few inches from his heels. Another dwarf has been taught to shy away from objects. It scurries to a corner when a hand is extended in its direction, and swerves to avoid bags and chairs in its path.

Later versions are programmed to recognise when power is low and find an electricity point for recharging, to communicate with each other so they move as a pack, and to 'learn' how to move around a corral. There are more in the laboratories housed in the bowels of the cybernetics department. There's also Elma, an intelligent walking robot which has an artificial nervous system in the style of a localised operating network, complete with a series of standalone neurone-like silicon chips able to communicate with other neurone nodes.

Elma has ultrasonic transmitters and receivers on her 'face' which enable her to move through obstacles. She also walks by using the neural network base to 'learn' which combination of leg movements will coordinate her legs so she can move without falling over.

But although these feats are rather endearing, I can't see the seven dwarves kicking Tony Blair out of Number 10, or Elma strutting all the way to the White House.

Warwick then leads the way to a lab downstairs. Inside is a five-foot-tall half-finished Dalek. Wires trail out the back, its 'arm' and sole 'eye' hang limply. Like the dwarves it has a series of ultrasonic sensors and will back away rapidly if you try to touch it. Its metallic arm will shoot CO2 gas when it senses heat. Designed as a fireguard to seek and extinguish fires, it has another potential. 'Just imagine if that sought out body heat. It could shoot out gas from the front. It could automatically recharge, and you couldn't switch the thing off. It's the same principle as the dwarves, but with potential danger.'

Intriguing as these gizmos are, though, the suspicion remains that Warwick is more interested in soundbites than in winning recognition for his theories from his academic peers. Is he just playing to the pop, pseudo-science gallery?

Having your professor in the media eye the whole time can't be doing the college funds that much harm. Indeed, Sun Microsystems has donated chunks of cash to the centre - much to the fury of institutions across the pond. Warwick gloats: 'Stamford was up in arms. Great - long may that continue.' He guides the way through a series of rooms where industrial projects abound, which he says have brought enough funds into the college 'to fund the blue-sky stuff'.

Warwick is quite miffed that Bill Gates is pouring his millions into Cambridge rather than Reading, but has no real complaints. 'I wouldn't turn away extra funds, but I'm happy with what we've got,' he says.

Warwick is keen to point out that the pioneering work isn't all geared towards producing cute computer creatures. The university has produced an automated wheelchair platform, now used by disabled children in Reading to teach them motor skills and how to operate an electronic-powered wheelchair.

The university has also devised a phone system for the deaf, and an intelligent bath that is designed to predict an epileptic fit and drain the water quickly. Warwick's faculty is working on an intelligent wheelchair that responds to brainwaves.

So surely he'd hate it if computers did prove to be mankind's nemesis, so jeopardising his own livelihood?

No. What Warwick wants is an international anti-proliferation treaty on arms. 'I suppose I'm doing a bit of scaremongering,' he admits. But the scenario of computers in control is not unreasonable. 'Surprisingly, the people I've discussed it with, the philosophers, have been saying "yes" more than "no".'

Warwick insists on driving me to the station himself - almost colliding with a van on the way. At the station, the platform indicators appear stuck, the train lurches violently forward, throwing coffee over the leg of the student next to me. Back at the office, the Mac pleads a virus and adamantly refuses to work properly.

According to Warwick, we're only just out of the slug stage of computer evolution, yet some machines are already refusing to cooperate. Maybe it's not science fiction after all.