If computers can beat humans at chess, does that make them smarter than us? Julia Grey reports
MACHINE has finally conquered man: Garry Kasparov, world champion chess player, succumbed this week to IBM’s supercomputer, Deep Blue, in their six-game rematch. The outlandish vision of the science fiction world, seen in 2001: A Space Odyssey, where an intelligent computer dictates to the world of humans, is about to become real.
Or is it? The underlying idea of Artificial Intelligence (AI) – that anything besides humans can think – has intrigued humanity for thousands of years. The mechanical devices, or automatons, which earlier ages dreamed up were all dismal failures – and then came computers. Indeed, the computer boffs of the 1950s had great expectations: a conference in the mid-50s optimistically predicted that a computer would be able to beat a human at chess within 10 years.
But some experts in the intriguing realm of AI do not see Deep Blue’s victory as a sign that humans are finally being outdone by the genius of their own inventions. After all, intelligence isn’t simply the capability to multiply 6-trillion by 354 in milliseconds. As Roger Layton, a South African computer consultant with years of experience in the field explains, intelligence involves a myriad of facets that are ultimately linked to being alive.
Exactly how humans are able to understand concepts and to build a world of memory is hard to pinpoint, and therefore difficult to replicate. Other mental attributes, such as the capacity for perception and imagination, are equally difficult to simulate mechanically – mainly because the questions around exactly what intelligence is must still be adequately answered.
A working definition of intelligence provided by Layton is “the ability to do something new with data you’ve already got” – basically the capacity to create new ideas. Compared to computers, humans are particularly strong in this area. Our minds are endowed with what Layton calls “a massive parallelism” – the ability to look, through inference and association, at millions of things at the same time.
This mental quality is what gives brains like Kasparov even the slightest chance against AIs like Deep Blue. In the game of chess, Deep Blue’s main advantage is its ability to make calculations far quicker than humans: the computer can make 200- million calculations a second, while the human brain works at about 100 calculations a second.
This, argues Layton, amounts to Deep Blue’s victory being “more of a success for high- speed computing than it is for AI”. He offers this challenge: “If we can get a computer limited to 1 000 cycles per second, and still challenge brains like Kasparov – then I will call it intelligent”.
Added to this calculating capacity, computers are also programmed (with ideas and information extracted from humans) to have intuition – or “heuristic” knowledge, as it is known in AI-speak. What this intuition amounts to is that the computer has advance knowledge of the implications of positions, and is therefore able “to look at situations in the future and determine probable outcomes”.
The fortune-telling capacity of computers like Deep Blue is where the immediate future of AI lies – particularly in areas like the stock market and weather forecasting.
Other potentials of AI – like emotion-based responses – are, says Layton, decades away. The science is presently “in the middle of an AI winter” – developments have been characterised by “a lack of true success”, as it becomes clear that the simple problems are simple to do, while the fundamental problems – like identifying and defining exactly the processes of human thought and understanding – remain out of reach.
But the bright sparks of the human race continue in their quest to illuminate the unknown. An International Joint Conference on Artificial Intelligence is scheduled for later this year, and will feature some novel activities. The first computer versus computer contest of the national Japanese game, Go (reputedly even more complex than chess), will be staged; and, most curiously, robots are going to play a game of soccer.
These steps may eventually lead the human race to situations similar to those envisioned by science fiction-writers – where we must bow down to the awesome superiority of The Machine, humbly surrendering to utter perfection in the light of all our incurable human fallibilities.
But there is one reassuring fact: the digits of our hands are perfectly suited to pulling the plug.