Turing Tests

As an individual, we’re never certain whether others have thoughts and inner workings just like our own, simply because we can never look into someone else’s mind. The most we can do is to figure that if others have normal human interactions, they’re probably conscious like ourselves.

Alan Turing proposed that we test computers the same way; through interacting. If computers are capable of human interactions, would that mean they’re conscious? In his 1950 paper, Turing wanted to answer the big question, ‘Can machines think?’, and devised a test which is now known as the Turing Test.

To pass the Turing Test, a computer has to correspond with a panel of judges through text messages and convince them into thinking they’re chatting with a real person. If enough judges believe they do, the computer passes the test.

In the 1960’s, the program ELIZA tried to pass by acting like a psychologist, encouraging the correspondents to talk more and reflecting their own questions back at them. 1972’s software PARRY simulated a person with paranoid schizophrenia, steering the conversation towards its own preprogrammed obsessions. The chatbot ‘Eugene Goostman’ fooled judges by enacting a young Ukrainian boy writing in English, thus using the apparent language and cultural barrier to explain its awkward use of grammar.

These particular programs did well because the makers realised that easiest way to pass the test is to fool the judges into thinking the program is intelligent, rather than actually being intelligent. In other words; they realised that to pass the test, being intelligent is totally irrelevant.

Turing himself had his own Turing Test in his life, as a homosexual in 1950’s United Kingdom. In that time and place, homosexuality was considered a criminal offence, but reality it didn’t matter whether Turing was attracted to men. It only mattered that he didn’t show it, especially in public.

Whether it’s a good or a bad thing, our lives are riddled with such Turing Tests. Just think about how in many relationships, especially in business, it doesn’t matter whether people actually like each other, it only matters that they leave the impression they do. And we think chocolate tastes better when it’s sold in a golden Ferrero Rocher’s wrappings, and therefor it actually does (whether it actually does in a blind test is irrelevant). It’s not important for placebo medicine to really work, it only has to fool its judge in order to work.

And so what if virtual reality becomes so good that we can’t distinguish it from reality? Could we could stop flying around the world to experience a tropical beach? We could save a lot of money and CO2 that way. And when substitute meat passes the Turing Test of meat, we could stop intensive animal farming.

Turing’s actual name for the test is ‘the Imitation Game’, and it’s something we play every day. The Imitation Game’s point not to be real, but to pretend to be real. And if we can’t distinguish between those two, why does it matter?