A computer has finally passed the Turing Test. You may now commence the global panic.
Two years ago, Eugene Goostman, the artificial intelligence computer program that masquerades as a wise-cracking 13-year old boy, won the Turing Test contest, fooling the judges into believing the all-digital Eugene is a flesh and blood person 29% of the times they spoke to him.
In the latest contest, held this weekend, the Goostman program managed to accomplish the feat 33% of the time — not only winning the competition, but passing the official Turing Test threshold for the first time.
Back in 1950, famed computer scientist, mathematician and code-breaker Alan Turing posited in a now famous paper entitled Computing Machinery and Intelligence that by the year 2000 it would be possible for a computer to play what's known as "the imitation game" well enough that "an average interrogator will not have more than 70 per cent chance of making the right identification after five minutes of questioning."
In other words, Turing predicted that a computer could answer text-based questions in such a way that it could, roughly one third of the time, be mistaken for a human interlocutor.
Goostman notched that achievement during Turing Test 2014, held on Saturday at the Royal Society in London. It competed against four other supercomputers. All of them were required to hold a five-minute, keyboard-based conversation with the judges. According to the University of Reading, none of the programmers were given topics or questions in advance.
But Goostman is not designed to know the answer to every question. Developer Vladimir Veselov explained his programming philosophy: "He can claim that he knows anything, but his age also makes it perfectly reasonable that he doesn't know everything. We spent a lot of time developing a character with a believable personality."
The team worked on Goostman's "dialog controller" to achieve more conversational responses. "Going forward," said Veselov, "we plan to make Eugene smarter and continue working on improving what we refer to as 'conversation logic'."
The key to believability may be that fallibility. Since people tend to believe computers are unlikely to get anything wrong, programmers competing in Turning Test typically introduce errors. As Turing noted in 1950, "If it is rigged up to give answers to questions as in the imitation game, there will be some questions to which it will either give a wrong answer, or fail to give an answer at all however much time is allowed for a reply."
Not everyone is buying this as a tipping point for artificial intelligence. "It doesn’t make a dammed bit of difference to robotics or AI,” said Rodney Brooks Professor of Robotics (emeritus) at MIT and founder, Chairman and CTO of Rethink Robotics. Brooks told Mashablethat Turing was, in 1950, posing a rhetorical question.
In fact, the "imitate game" was originally designed to see if a man or women could fool an interviewer into inaccurately guessing their gender. "People spend so much energy on a rhetorical question ... [Turing] wasn’t saying that [the test] was the definition of intelligence and then we’d be off to the races."
Computers listening and responding to our questions in seemingly human ways is not new. In the 1960s a so-called "chatbot" named ELIZA fooled some of its interlocutors. You can check out a version of ELIZA here.
These days, many of us are already carrying around intelligent voice assistants in our pockets. Apple's Siri for example can answer a variety of natural language queries, like "How do I look?" with "I don't know. But I guess you look pretty good," and Windows Phone's Cortana can answer, "What are you doing today?" with this cheeky response, a dig at the iPhone: "multitasking."
Neither voice assistant is equipped to carry on five minutes of uninterrupted conversation, however. You'll notice that when Siri doesn't have the answer, it sends you to search.
And if you're frightened by the notion of a computer that can at least appear to think? Turing's advice in 1950 was the equivalent of "get over it."
We like to believe that Man is in some subtle way superior to the rest of creation. It is best if he can be shown to be necessarily superior, for then there is no danger of him losing his commanding position ... I do not think that this argument is sufficiently substantial to require refutation. Consolation would be more appropriate: perhaps this should be sought in the transmigration of souls.
You can try an online version of Eugene Goostman right here. Don’t be surprised if he fools you too.