In what is both ground-breaking and unsettling news, a Russian chatterbot dubbed “Eugene Goostman” has become the first to pass the Turing Test. This intelligence assessment, started in 1950 by Alan Turing, the mathematician and logician, is considered a benchmark for artificial intelligence in human interaction.
The bot managed to convince an average of 1 in 3 judges that it was a live, non-native English-speaking Ukranian boy.
In a competition with four other computerized contenders (and human controls) at the Royal Society of London’s Turing Test 2014 Competition. The five-minute text-based conversations with panel judges had the contenders attempting to convince at least 1/3 of them (30%) that they are human. Eugene managed to convince 33% of them (1 in 3) that it was, despite being nothing more than a small computer program in the “chatterbox” category. Eugene had previously taken first place, though not beaten the Turing test, in 2013.
The competition was held on the 60th anniversary of Turing’s death by suicide, which he committed after being convicted of “gross indecency” and sentenced to be chemically castrated for being homosexual.
The chilling part of this story is that this chatbot is something that could (and probably is) be used in various scams and questionable marketing circles. While the way that Eugene beat the system at the Turing contest was less than innovative, it still implies that the bot is believable to a substantial number of people – all of whom were expecting that many of their interactions were with machines to begin with.
That is more telling.