intrust

“The imitation game” is a 1950s thought experiment formulated by Alan Turing, one of the great pioneers in computing. Today we call it the Turing Test. “Can machines think?” was then – still is – a pressing issue. But there is a problem involved in asking this question: Since we don’t really know how humans think, or exactly what thinking is for that matter, how can we reasonably ask if machines do it? If we could deductively determine how a mind works and map it all out and know what and where to measure then we would not need a Turing test.

Either we take the solipsistic view that we don’t really know whether anyone other than ourselves actually does think – or exist for that matter, or we grant that things are as they seem: Other humans do exist and think because they appear to do so. If machines also appear to do so, then Turing was willing to grant them the ability to think.

Turing’s proposal was to test machines in the same way we would test another human. If, to any rational observer, machines seemed to think in the same way that humans seem to think; if machines were indistinguishable from humans in their responses to question that we asked them – then we would have to join Turing in granting them the ability to think just as we grant it to fellow humans.

In Turing’s test an examiner would submit questions in writing to two unseen and unheard examinees, one human – the other a machine. If the examiner could not distinguish between the two by their written answers then she would have to concede that machines can think.

// posted by Greg FitzPatrick @ 14:05