Letter: Turing and the artificial mind

Sir: Michael Lockwood ("Man v Machine", 13 May), like so many others, has overrated the Turing test. The Turing test is not some goal or benchmark artificial intelligence (AI) researchers are (or should be) shooting for. It is a thought experiment designed to get us each to ask ourselves, "How do I know that anything other than myself experiences consciousness?" If we give each other the benefit of the doubt, then why not extend that to something that passes the Turing test?

Beyond that role the test is not important. If and when we first implement an algorithm on silicon that experiences consciousness, it is very unlikely that it would pass the Turing test. (After all, we wouldn't expect some species out of Star Trek to pass the test either.)

The more important lesson from Alan Turing is the separation of algorithm from hardware. While the brain is very different from integrated circuits, nobody (pace Roger Penrose) has been able to argue that it is different enough given the substrate neutrality of algorithms. The arguments for strong AI are compelling. Saying that there is a lot about consciousness and the brain that we don't understand yet is no counter argument - it is mere wishful thinking.

JEFF GOLDBERG

Cranfield Computer Centre

Cranfield University