Boffins conclude machines still not quite people
Sorry Dave, give us a hand with this Turing Test?
Politicians are not quite ready to pass themselves off as human: but machines are almost there. This was the shock conclusion from a Reading University competition run at the weekend designed to sort out machines from people.
Five computer programmers from across the globe competed for the $100,000 Loebner prize for Artificial Intelligence. The competition began in 1990 when philanthropist and academic Hugh Loebner agreed to underwrite a contest to investigate machine "intelligence" by applying the Turing Test – named, not unsurprisingly, after British mathematician Alan Turing.
He was one of the first to ask not only "Can a Machine Think?" but also "If a computer could think, how could we tell?" His answer was that if the responses from the computer were indistinguishable from that of a human, the computer could be said to be thinking.
To pocket Loebner's pledged grand prize and gold medal, a computer would have to fool a human at least 30 per cent of the time. The silver medal would go to machines passing a slightly longer Turing Test.
Each year an annual prize of $2000 and a bronze medal is awarded to the "best in show" - that is, the one that outperformed the others, even if it didn’t quite convince subjects of its human credentials.
To win the Gold Medal, a computer must take part in a spoken conversation with human subjects. Silver and bronze Medals depend only on a simple on-screen text exchange. The winner of the annual contest (bronze) is the best entry relative to other entries that year, irrespective of how good it is in an absolute sense: to date neither gold nor silver medals have been awarded.
This year’s winning machine, known as Elbot, only achieved a 25 per cent success rate.
Organiser of this year’s contest, Professor Kevin Warwick from the University of Reading's School of Systems Engineering, said: "This has been a very exciting day with two of the machines getting very close to passing the Turing Test for the first time. In hosting the competition here, we wanted to raise the bar in Artificial Intelligence and although the machines aren't yet good enough to fool all of the people all of the time, they are certainly at the stage of fooling some of the people some of the time.
"Today's results actually show a more complex story than a straight pass or fail by one machine. Where the machines were identified correctly by the human interrogators as machines, the conversational abilities of each machine was scored at 80 and 90 per cent. This demonstrates how close machines are getting to reaching the milestone of communicating with us in a way in which we are comfortable.
"That eventual day will herald a new phase in our relationship with machines, bringing closer the time in which robots start to play an active role in our daily lives."
Dr Warwick is an enthusiast when it comes to the rise of the machine. His personal website – I, Cyborg – explains in detail how he became the world’s first Cyborg, when he agreed to have a silicon chip transponder implanted in his forearm. Subsequently he was able to operate doors, lights, heaters and other computers without lifting a finger.
As for determining the difference between machines and politicians? Those responsible for organising the weekend event have not yet done any in-depth analysis of what factors made some people more likely to believe they were dealing with a machine – or a human. That will come later.
However, one of the organisers did suggest that the obvious machines didn’t really answer questions - they were quite vague or definitely evasive, obviously picking up on key words, and building an answer around those, rather than what they had been asked.
It is a technique with which Jeremy Paxman has had a great deal of experience. Maybe one day he'll be able to rely on a robot co-host to extract the truth, but he probably shouldn't hold his breath. ®