Eugene, a supercomputer posing as a 13 year old boy, has become the first machine to pass the Turing test first posed in 1950
A computer programme masquerading as a 13 year old boy has become the first piece of software to trick people into thinking it is human.
The news has been hailed as an “historic milestone” for artificial intelligence, but what does such an achievement mean?
In 1950 renowned mathematician and codebreaker Alan Turing posed a simple question – can machines think? It has taken over 60 years for a computer to prove that, to an extent, they can.
The Turing test, as it has become known, requires a computer to trick people into thinking it is human. At the time Turing wrote that “an average interrogator will not have more than 70 per cent chance of making the right identification after five minutes of questioning”.
Rightly or wrongly this has been interpreted to mean a computer will pass if it is mistaken as a human more than 30 per cent of the time during a series of five minute text chats.
Turing’s original test was more more hypothetical but eager researchers have seen it as an opportunity to create a highly believable mimic. No computer has ever achieved this, until now.
Eugene convinced 33 per cent of human judges that it was a 13 year old boy, when in reality it was a computer programme.
The Turing test is a totally random conversation with no pre-determined topics or questions, meaning that the computer won’t necessarily be ‘prepared’ for what will be asked.
Conversations are spontaneous and carried out by normal people – in this case Red Dwarf’s Robert Llewellyn and Lord Sharkey, amongst others.
Eugene was one of five supercomputers competing for the Turing Test 2014 at an event organised by the University of Reading.
Turing himself admitted a flaw in his original question, later rephrasing it he pondered how well computers would do in “the imitation game”. In passing the Turing test for the first time, Eugene has shown the ability of a computer to be as dumb as a human.
Just as humans are intelligent, they are also very stupid. In order to be convincing a computer must make mistakes and it must be fallible. To pass the Turing test, critics have argued, a computer is simulating behaviour rather than exhibiting actual intelligence.
Eugene does not have a mind, it does not have a consciousness and it does not think. The Turing test, as its creator admitted, was designed to start a debate about artificial intelligence and not to prove its existence.
Turing reckoned his test would be passed by the year 2000, so we’re already fourteen years late. Breaking Turing’s test is rightly seen as a milestone but the philosophical debate of what constitutes intelligence will go on.
Speaking about Eugene becoming the first computer programme to pass the Turing test, professor Kevin Warwick from Coventry University said it still had implications for society today:
“Having a computer that can trick a human into thinking that someone, or even something, is a person we trust is a wake-up call to cybercrime,” he said.
“The Turing Test is a vital tool for combatting that threat. It is important to understand more fully how online, real-time communication of this type can influence an individual human in such a way that they are fooled into believing something is true when in fact it is not.”