27 November 2001 artificial intelligence

from a letter:

...I am reminded of some thoughts that come to me when confronted with the claims made for artificial intelligence:

It is not correct to say that the computer program Deep Blue equals or defeats the chess champion - the computer software is not magic, nor is it intelligence, nor consciousness. It enables a team of hidden chess players (the software writers) to play their games in advance, using resources not allowed during chess-playing. So the contest is not between one player and a machine but between one player and a hidden team who have the means to explore all possibilities out of 'real time'.

There is a rule of thumb in ergonomics (my former activity):

'in any seemingly automatic system there is always a "human operator" though he or she or they may be hidden by electronic shifts of time and place'. (Professor Tom Singleton said this at a conference)

For instance the effective operator of a train may not be the driver, who is blamed for a crash, but the people who designed that particular mix of signals, rail-network, and train schedules - in which a certain frequency of unconscious mistakes is inevitable however skilled and conscientious is the driver.

Similarly there is the Turing test for 'artificial intelligence'. This states that AI exists if a user cannot distinguish the response of the machine from that of people concealed in a 'black box'. But any 'intelligent machine' IS a group of people whose presence is concealed - who are acting, or did act, at other times and places...

If you wish to reply publicly to this please send your reply as email to:

jcj at publicwriting.net

(as your copyright but giving others freedom to communicate it for non-commercial purposes, and including the copyright statement on my homepage, version 2.2)

digital diary dates