In a 1951 paper, Alan Turing proposed the Turing test called, “The Imitation Game”. The Imitation Game, as he saw it would consist of three rooms with three different people in it. One room would host a man, another a woman, and the third would be the judge. This judge would be tasked with deciding if the person speaking to him or her through a computer was the male contestant or the female.
Here’s the twist. Turing then proposed that this same test can be used to test for artificial intelligence. With the two contestants consisting of a human and a machine with the third judge tasked with deciding whether it is a human or a machine talking to them. The job of the judge would be to interrogate the human and machine with a series of questions in order to judge its reaction, tone, and how they reply to the questions being asked. Nowadays, the game has changed from two players to one with the interrogator’s job to determine if the contestant is a human or a machine.
The test can only be successfully passed if the machine if mistaken for a human more than 30% of the time during a 5-minute interrogation.
Testing AI’s Ability to Think
When creating the test, Turing raises the question of a computer’s ability to think. Or interact in a way that would make a computer “intelligent”. The machines’ way of “thinking” is measured by the questions asked and can ultimately lead the interrogator to determine that it is indeed a machine and not human. If however, the computer is able to humanize its responses appropriately then it could quite possibly fool the interrogator into thinking it’s human.
Currently, there are a few annual Turing Tests being held. Some in the United States others in the United Kingdom and across the globe. While there have been two well-known computer programs or chatbots, claiming to have passed the Turing Test, the reality is that no AI has been able to pass it since it was introduced.
Turing, himself, thought that by the year 2000 computer systems would be able to pass the test with flying colors. Looks like technology might be lagging a little.
A Close Call
Eugene Goostman, which simulates a 13-year-old Ukrainian boy, is said to have passed the test in June 2014. While Eugene may have fooled quite a few people, there are actually a few red flags that prove it didn’t pass. One in particular is that people were told the computer was a 13-year-old Ukrainian boy to excuse all of the odd answers it may have given. Secondly, the test judges were hand picked by those who created Eugene. This doesn’t add much credibility does it?
Cleverbot’s developers also claimed it passed the Turing test back in 2011. But, if you’ve ever had a chat with Cleverbot then you would know the chatbot really isn’t as bright as its developers think it is. Don’t believe us? Strike up a conversation here.
While no computer has passed the Turing Test that doesn’t mean we aren’t on our way to creating true artificial intelligence even if Ada Lovelace stated otherwise. While there are those who argue about when a computer will actually pass the Turing test, one thing they can all agree on is that it is very likely to happen in our lifetime.