In an essay published in the 1950s, the mathematician and computer pioneer Alan Turing proposed the Turing test. It has become one of the most important milestones in the research and development of artificial intelligence.
The Turing test is the first step in determining whether a machine can detect human intelligence. A machine must prove “human intelligence” by engaging in conversation with a human. The test examines whether humans can tell if they are talking to a machine or a human.
Google demonstrated with artificial intelligence technology the first time a computer has entered into a natural conversation with a human.
At a demonstration at Reading University, the computer convinced a human judge that it was a 13-year-old Ukrainian boy, not a machine. He was the first computer to pass the famous Turing test and won over one third of the jury of humanity.
Organisers at Reading University said no computer had previously passed the Turing test, which involves persuading 30 human interrogators to make a series of five-minute keyboard conversations. But a computer program designed to simulate a 13-year-old boy convinced the 33 judges that it was a human.
The program, known as Eugene Goostman, is the first artificial intelligence to pass the test, originally developed by 20th-century mathematician Alan Turing. It was one of five supercomputers to beat the famous test, and was first developed in St Petersburg, Russia.
The Reading University test was similar in its original form to the Turing test: an interrogator essentially sent a text message to a human computer and received a reply. But artificial intelligence experts dispute the win, suggesting the competition has been weighted in favour of the chatbot.
For many researchers, the question of whether or not a computer can pass the Turing test has become irrelevant. Instead of focusing on how to convince someone to talk to a human, rather than the computer program, our real focus should be on how to make human-computer interaction as smooth as possible.
So, it looks like today’s AI algorithms can pass the Turing test, but the real question is, could you? Or better yet, did you?
For those who haven’t clocked on yet, the above blog post was entirely generated by a machine, being both researched and written by an AI algorithm.
Personally, I’d like to think it doesn’t quite match up to my usual standard of writing — or the general standards of the Wildfire blog — but as a fun experiment, it’s interesting to see what today’s AI is capable of.
Did you pass the test? Find out more about why these types of experiments matter in my previous post, The need for explainable AI