Cobbling up thought
Sir: Mr Matthew Huntbach (Letters, 28 May) is unlikely to become a successful computer scientist if he thinks the pros- pects for teaching machines to think are a 'load of cobblers'.
His pessimism is misplaced. He tells us that 'when it comes to actually discovering how to do things, machines have hardly progressed at all'. On the contrary, they have progressed mightily since 1961 when A. L. Samuel of IBM wrote a draughts- playing program that consistently defeated its own creator. Since then, not only has a computer defeated the world back- gammon champion, but machines can now routinely defeat all human chess players except grand masters. When they can overcome that last hurdle, three scientists wrote recently in the IBM Journal of Research and Development, 'they will have penetrated to the very core of human intellectual endeavour'.
Playing good chess is an excellent exam- ple of intelligence, for it requires genuine 'thinking'. In effect, all the machine is told when it is programmed are the rules of chess and its objectives. The method by which it defeats its human opponent is entirely its own creation.
As Hans Berliner, creator of the victo- rious backgammon program put it: 'Chess is a game of such complexity that a machine that mastered it could soon be adapted into one which could manage much of human affairs.'
Adrian Berry
Science correspondent,
Daily Telegraph, 181 Marsh Wall, London E14
Previous page
