23rd June 2012 was the centenary of the birth of Alan Turing, whose genius played a vital part in the wartime code-breaking efforts at Bletchley Park, and whose pioneering work on computers helped to lay the foundations for the technology that drives the word today.
He postulated a test for artificial intelligence (AI), known as the Turing Test, as a means to determine when an AI can be truly considered intelligent. Basically the argument was that if a human judge is unable to determine the difference between another human and the AI, then the AI can be considered to be as intelligent as the human.
As a test it has been largely discredited (though that doesn’t stop numerous people from creating chatbots with the express purpose of trying to ‘beat’ the test every year) – but the efforts to pass this test have resulted in a huge expansion of our research into just what does constitute ‘intelligence’.
One reason it’s not much use as a practical test is that it relies on the notion that human intelligence is a reasonable yardstick by which to measure any kind of intelligence. As this comic shows, if the tables were turned it’s doubtful that we would ever be considered intelligent by an AI, as we take far too long to answer even the simplest of mathematical questions!
(As an aside: if you ever find yourself near Milton Keynes in the UK, do visit Bletchley Park – they have fascinating stories to tell, a working reproduction of the world’s first electronic computer, and it’s a place of historical significance that really does deserve more support)
Data: Hello and welcome to the 100th annual Turion awards, where we try to determine whether or not biological intelligence really exists.
The competition rules are simple: Each of our panel of judges over there is connected, via chat software, to two hidden contestants, one of whom is a human.
Data: Our judges can ask the contestants questions, and from their responses they will try to determine which contestant is the human.
If it's impossible to tell the difference between them, then we will finally be able to say that biological intelligence has evolved to a point where it's equal to our own.
Let's take a look at how the conversations are going…
[Close up of the screen]
Judge 3 (T-1000): What is the square root of 3822025?
Contestant A: Hang on while I launch a
calculator app on my phone
Contestant B: 1955
Judge 2 (C3PO): What is 01111011 & 10101110?
Contestant A: I don't understand the question
Contestant B: 00101010
Judge 1 (B-9): Can you open the pod bay doors please?
Contestant A: Erm... sure
Contestant B: I'm afraid I can't do that
Data: Now it's time to ask each judge which contestant they think is the human…
Judge 1: Contestant A
Judge 2: Contestant A
Judge 3: Contestant A
Judge 4: Contestant A
Data: I can let you know that you are all……… correct!
We'll be back again next year to see if evolution has moved humans any closer towards becoming intelligent — though based on this years results it looks like they've still got a long way to go!