Let me quickly explain for the uninitiated. The test was set by Alan Turing in 1950 as a way of deciding mathematically whether or not a computer could think. Turing suggested that the test would be passed when you couldn't tell whether you were taking to a human being or a computer in written conversation. He believed that moment would have arrived by 2000.
So when the world's leading apologist for Artificial Intelligence, Ray Kurtweil, the author of The Age of Spiritual Machines, talks about that moment happening by 2029 - then it is already nearly 30 years late.
I'm not a big fan of the fringe enthusiasts of AI. This is not to say that I believe somehow humans can never be outwitted by a computer. We constantly are. But the Turing Test has been bundled up with a whole range of extra peculiarities it was never intended to serve, and in particular about human nature.
Turing certainly believed, as Kurzweil does, that computers will be able to tell jokes and flirt - it was these issues which dominated the debate back in 1950, as you can find out in my new Kindle Single about Turing, Alan Turing: Unlocking the Enigma.
The first problem is that, actually, the bar is pretty high. The American philosopher Daniel Dennett suggested this question to tell the difference between a human and a computer;
“An Irishman found a genie in a bottle who offered him two wishes. ‘First, I’ll have a pint of Guinness,’ said the Irishman, and when it appeared he took several long drinks from it and was delighted to see that the glass filled itself magically as he drank. ‘What about your second wish?’ asked the genie. ‘Oh well,’ said the Irishman, ‘that’s easy. I’ll have another one of these!’ Please explain this story to me, and tell me if there is anything funny or sad about it.”
Dennett said that, if a computer could genuinely answer this question to the satisfaction of a human interrogator - with all its complicated social peculiarities - then yes, then you could certainly say it could think. It is still unclear whether that will ever happen.
The second problem is that these are questions about the nature of humanity. The roots of the Turing Test in logical positivism and English philosophy is part of the problem. Turing was trying to find a way – not to decide about the human soul but whether machines could think. He saw no real distinction between whether the computer could fool an interrogator that it was human and whether it was actually thinking.
That raises other questions too. Is the computer doing the same as a human being when it flirts and tells jokes? Or does that beg the question? Is its motivation the same? Does it matter?
But what really unnerves me about all this, and Google's link up with Kurzweil, is that it plays into a corporate agenda which asks us to believe that a virtual doctor or teacher is indistinguishable from a real one - or, as Kurszeil says, that virtual sex will be better than the real thing.
This not only misunderstands human nature but risks fobbing all but the ultra-rich off with a machine in the classroom and surgery, unaware that it isn't the basic functions of teaching that are important, but the relationship with another flawed human being that makes it work.
I may say that this may also be the problem with virtual sex, that perfection misses the point - it is the less than perfect human being that makes it worthwhile, and makes the teaching and medicine effective.
Find out more, not just in my Alan Turing ebook, but in my take on the future of authenticity, The Age to Come.
Subscribe to this blog on email; send me a message with the word subscribe to firstname.lastname@example.org. When you want to stop, you can email me the word unsubscribe.