A History of the World (89 page)

Was any of this surprising? Financial capitalism has always evolved through bubbles and crashes. Wherever wealthy companies or individuals are able to huddle together, improperly regulated, they will conspire against the public. Adam Smith told us that. The failures were to be located inside the same democratic-representative structures that were supposed to be the trump card of the West: politicians who spent too much money campaigning to really get tough on the bankers, and too much time worrying about geopolitics to attend to the health of their own economies. Their voters wanted cheap goods and easy credit, and it was comfortable to give them what they wanted. Finally, the philosophy of modern market capitalism, while greatly superior to that of state socialism, was narrow and unhistorical. It put consumerism on a pedestal while underestimating long-term human instincts such as spiritual questing, tribalism and fear. But they hadn’t gone away.

The Thinking Machine
⁵²

On 11 May 1997 an event happened in New York that deserves to be remembered far into the future. Someone who had a good claim to be the world’s sharpest man was defeated by a computer. Garry Kasparov is held by many to be the greatest chess-player of all time, a grandmaster so good he is in a category all of his own. It would be lazy to say that Kasparov played like a machine. He had an astonishing memory and a great sense of strategy, but he was also courageous and emotional. A reflective, pugnacious Jewish Armenian who had grown up in the high-pressure world of Soviet chess, he became the world’s number one player aged just twenty-two and kept the title for most of the following two decades until he retired in 2005.

Kasparov had taken on machines before. In Hamburg in 1985 he played simultaneous matches against thirty-two chess programs, and beat them all. Four years later at a match in New York, IBM challenged him with their ‘Deep Thought’ computer. Kasparov had warned that if he lost, it would be ‘unpleasant’ for the human race. He wanted ‘to be the man who saved human pride’. He won the game in two and a half hours. Seven years on, the IBM team had a new machine, ‘Deep Blue’. In Philadelphia in 1996 Kasparov lost a game to it, but had come back to win the match. Now came the much-hyped, much-touted rematch. At a tower block in Manhattan, the Equitable Center, the world’s media were watching for what both sides suggested would be an epic struggle between the human brain and the new power of computing. Posters around the city showed Kasparov staring intently into the middle distance, with the caption: ‘How do you make a computer blink?’
Newsweek
magazine carried a cover shouting: ‘The Brain’s Last Stand’.

Was all this mere commercial hyperbole? Not entirely. Ever since chess spread from its origins in India in the 500s, first through Persia and the Muslim world and then into Europe, it has been recognized as a special game that tests human memory and planning to the maximum. It is often, and naturally, compared to mathematics; the brains that are good at chess are also often good at maths. Yet it also requires a kind of genius that cannot be reduced to rules. It is a severe test of logic, but it has always had a mystique that neither
card games nor other board games, even chess’s Chinese rival, Go, possess.

Feng-Hsiung Hsu, one of the scientists behind Deep Blue, says that since the 1940s computer theorists had dreamed of a machine that could play chess. One of the pioneers of Artificial Intelligence said in the 1950s: ‘If one could devise a successful chess machine, one would seem to have penetrated to the core of human intellectual endeavor.’
⁵³
Kasparov agreed: but he was determined to prove that a machine could only play like a machine, and that at some important level it was stupid, emanating no creative energy.

The first game seemed to confirm Kasparov’s confidence – he won easily. The second game was the turning-point. For non-chess-players its importance can be hard to understand; but it
was
important. Kasparov decided to sacrifice a pawn to give himself a better position later. Computer chess programmes are designed to grab short-term advantage, and therefore Deep Blue ought to have told its non-human player to take Kasparov’s pawn. He assumed it would. Computers play chess not by intuition but by giving a score or number to all possible moves, then scoring the possible counter-moves of its opponent, then scoring its own next possible moves, and so on.

Easy? The number of possible combinations of moves in a game of chess is greater than the number of atoms in the universe. Just to look a few stages ahead requires extraordinary computing power; human chess-players use their sense of patterns and psychology instead. But now Deep Blue behaved like a (very good) human player. After some time in what its creators called ‘panic mode’, the fridge-sized metal box refused to take Kasparov’s pawn. It made another, long-term, tactical move instead. It was not playing like a machine.

Kasparov seemed to have been spooked by the computer’s apparent intuition. Shortly afterwards, he conceded the game and stalked off, shaking his head. In fact, the computer had made a mistake in a later move and Kasparov could still have managed a draw rather than losing – something he was shocked by when told later. He has argued that, on this basis, he did not really lose. In the following games he drew three and lost the last, playing so badly that it was hardly considered a contest at all.

In the post-match press conference, Kasparov was asked whether the IBM team had cheated – whether there had been ‘some kind of
human intervention during this game’. He replied that it reminded him of the goal that the Argentinian footballer Maradona had scored against England in 1986, when he had knocked the ball into the net with his hand but had not been caught by the referee, and had claimed: ‘It was the hand of God.’ The IBM team of scientists, lined up beside him on the platform, were furious at the slur. From their point of view, many years of hard work were being denigrated by a bad loser. The controversies about this episode will never subside. Kasparov had repeatedly asked for printouts of what the computer was doing, and never got them, because the IBM team thought that would have given him an unfair advantage, including in any future match. After the match was over, Deep Blue was disassembled and put into storage; it has never been used since.

There is another point about the kind of contest that was going on here. Was it really man against machine? Kasparov suffered from exhaustion, worry, anger and suspicion, which the computer, using vastly more computational power than any Kasparov had encountered before, did not. So to that extent, it was. He had an ego. It had none. Yet Deep Blue was itself the creation of human brains, who described its chess struggle with fatherly concern. Feng-Hsiung Hsu has written that the contest ‘was really between men in two different roles: man as a performer and man as a toolmaker . . . Deep Blue is not intelligent. It is only a finely crafted tool that exhibits intelligent behaviour in a limited domain.’ Though Kasparov lost the match, he added with a feline twist, only he had real intelligence: ‘Deep Blue would never have been able to come up with the imaginative accusations.’
⁵⁴

Later Kasparov himself came close to agreeing: Deep Blue was a great achievement, he said, but it was ‘a
human
achievement by the members of the IBM team . . . Deep Blue was only intelligent the way your programmable alarm clock is intelligent. Not that losing to a $10 million alarm clock made me feel any better.’
⁵⁵
Alongside the computer scientists, no fewer than six chess grandmasters had been working on the programmes: Kasparov was taking on not simply a corporation (and IBM did massively well from the publicity, gaining kudos it had been losing to the new kids of Silicon Valley), but also a lot of accumulated human knowledge and preparation.

All that said, the possibility of machines matching and outpacing human intelligence in many fields is clearly a real one, taking place all
around us today. The interweaving of billions of people’s imaginative lives through the Internet is the most obvious way technology has changed our lives recently; but Artificial Intelligence, or AI, may soon prove itself much more significant. Major advances in enabling machines to ‘see’ (one of the hardest problems) and respond to natural human language are occurring now. New insights into the way the chemical-biological human brain processes information, and how this can be mimicked by later generations of computer, are lively subjects in the universities and labs. So Kasparov’s look of astonishment when Deep Blue made its crucial move ought to be remembered as a special moment in human history.

The dream of machines able to match human intelligence is an ancient one, but it only became a serious scientific subject in the 1950s, thanks to advances in computer science and, to a lesser extent, in the understanding of the brain. Alan Turing, the brilliant scientist and pioneer in computing who was critical to Britain’s wartime organization at Bletchley Park (which broke secret German codes), became fascinated with it. He had worked before the war on computer theory; in 1936 he had proposed what became known as a ‘Turing machine’, which would read symbols on a long tape, to make mathematical calculations. At this time, punched cards and vacuum tubes were the best technology available, but war tends to accelerate invention and the ‘Colossus’ machines at Bletchley Park used to break Nazi codes are generally regarded as the world’s first proper computers, in the sense of being programmable and digital, as well as electrical.

In 1950, Turing proposed his famous ‘Turing test’. This posited that if a judge was having a conversation with a human being and a computer (the identities of each being disguised through a keyboard) and was not able to tell which was which, the computer had passed the test. This, he suggested, was the sensible, measurable answer to the question about whether machines could ever think, or achieve consciousness. He did not live to see the advances that would follow. Turing was gay, and in 1952 was convicted for ‘gross indecency’ with another man and obliged to accept chemical castration as part of his punishment, as well as losing his clearance to work on government projects. He died of cyanide poisoning, probably an act of suicide, in 1954.

Two years later a conference took place at Dartmouth College,
New Hampshire, where Marvin Minsky, one of the fathers of AI, and John McCarthy, the computer scientist who coined the term, led discussions on natural language, computer programming and mathematical logic. It was a breakthrough moment for the new discipline. Back then, the optimism of people like Minsky and McCarthy ran far ahead of what was possible. Urged on by fiction writers like Arthur C. Clarke, predictions were made in the late 1950s and 60s that artificial intelligence would have arrived by the 1970s or 80s. Turing himself had focused on chess as a useful test-system for AI because of its complex logic and patterning; in 1958 two key scientists at Carnegie Mellon University, Pittsburgh, had predicted that by 1968 a digital computer would be the world’s chess champion.
⁵⁶
All that had held them back was lack of computing power – the physical slowness of the machines available.

But they were working on it. After transistors replaced the old vacuum tubes, the problem was packing together and powering enough of them. Transistors are basic semiconductors switching electronic signals, and thus the essential components of digital computing. The first generation used copper wires and were relatively slow. Many people worked on the problem, but it was an employee of Texas Instruments called Jack Kilby who was generally credited with this advance. In 1958 he etched transistor elements onto slices of germanium, a carbon-based semi-metal, and connected them with fine gold wires to an oscillator and amplifier. Silicon would soon prove to work better, but the ‘chip’ had been born – in essence, pieces of cooked, sliced sand that are engraved using ultraviolet light and gas to turn them into electrical switches. In 1965 Gordon Moore, a co-founder of Intel Corporation, said mankind would now see a repeated annual doubling of the number of transistors that could be fitted onto a circuit, and though widely criticized for this explosive exponential prediction, he has been proved largely correct. By the late 1970s, entire microprocessors were being put on a single chip. For the IBM team this was essential to their chess-playing machine, which began with a circuit board of six thousand transistors.

What can we expect next? The enthusiasts rest heavily on the idea of acceleration, or exponential growth – the notion that technological progress multiplies by a constant figure, rather than simply adding a constant (as in linear growth). The difference is between a very slowly
rising stable line and one that starts slowly and then suddenly erupts upwards to a near-vertical line of ‘take-off ’. A graph of human population increase during the timespan described in this book shows something like this. Moore’s law on computing power does the same. More generally and unscientifically, much of the underlying shape of the story told here is of exponential growth – the millennia of hunter-gathering, followed by the relative speed of the farming revolution, then the ever-faster hurtle through towns, cities, empires and industrial technology.

The scientist and writer Ray Kurzweil has popularized the phrase ‘the Singularity’ – dignified, like God, with a capital letter – which he defines as the time when the pace of change is so rapid and profound that human life is transformed. The idea came from a mathematician and science-fiction writer called Vernor Vinge, who boldly plumped for the year 2030 when ‘computer super-intelligence’ would give birth to the Singularity, leading to a time when large computer networks might wake up as a superhuman intelligence. The language is close to religious, and may yet provoke a new religion or cult. Kurzweil proclaims ‘a transforming event looming in the first half of the twenty-first century’. Like a black hole changing the patterns of matter and energy, ‘this impending Singularity in our future is increasingly transforming every institution and aspect of human life, from sexuality to spirituality’.