June 10, 2014
Chatting about the Turing test
Eugene the Turing test-beating teenbot reveals more about humans than computers - I blog at The Conversation about the recent 'success' in beating the Turing test.
Actually running the Turing test is a bit like putting cats in boxes or threatening people with runaway trolleys in order to learn more about quantum mechanics or ethics. It was intended as a thought experiment, not as a test. Riva Tez pointed me at what Turing said: “[A] statistical survey such as a Gallup poll [would be] absurd [as a way to define or determine whether a machine can think]” (Turing 1950).
However, running experimental psychology trolley problems have demonstrated odd things about how we think about ethics. Running the Turing test has shown us the Eliza effect: we are very bad at *not* ascribing minds to entities we converse with, even if they are decidedly stupid. Making software that interacts well with humans in a social setting is useful and nontrivial. So running Turing tests is not per se a waste of time, just not a way of measuring AI progress.
Posted by Anders3 at June 10, 2014 12:02 PM