Is Conversational AI Here?

The idea of Conversational AI and the hopes and fears that are associated with its rise are fairly prevalent in our common subconscious. Whether we imagine Judgement Day at the hands of Skynet or egalitarian totalitarianism at the hands of V.I.K.I. And her army of robots – the results are the same – the equivocal displacement of humans as the dominant life forms on the planet.

Some might call it the fears of a technophobic mind, others a tame prophecy. And if the recent findings at the University of Reading (U.K.) are any indication. We may have already begun fulfilling said prophecy. In early June 2014 a historic achievement was supposedly achieved. The passing of the eternal Turing Test by a computer programme. Being hailed and derided the world over as being either the birth of artificial intelligence or a clever trickster-bot that only proved technical skill respectively, the programme known as Eugene Goostman may soon become a name embedded in history.

The programme or Eugene (to his friends) was originally created in 2001 by Vladimir Veselov from Russia and Eugene Demchenko from Ukraine. Since then it has been developed to simulate the personality and conversational patterns of a 13 year old boy and was competing against four other programmes to come out victorious. The Turing Test was held at the world famous Royal Society in London and is considered the most comprehensively designed tests ever. The requirements for a computer programme to pass the Turing Test are simple yet difficult – the ability to convince a human being that the entity that they are conversing with is another human being at least 30 percent of the time.

The result in London garnered Eugene a 33 percent success rating making it the first programme to pass the Turing Test. The test in itself was more challenging because it engaged 300 conversations, with 30 judges or human subjects, against 5 other computer programmes in simultaneous conversations between humans and machines, over five parallel tests. Across all the instances only Eugene was able to convince 33 percent of the human judges that it was a human boy. Built with algorithms that support “conversational logic” and openended topics. Eugene opened up a whole new reality of intelligent machines capable of fooling humans.

Leave a Reply

Your email address will not be published. Required fields are marked *