This story is over 5 years old

You Shouldn’t Be Alarmed That a Computer Passed the Turing Test

Alan Turing’s test set the bar for intelligence at convincing 30 percent of human judges that they are communicating with one of their own.

by Natasha Lennard
Jun 9 2014, 10:05pm

Photo by Elliott Brown

Did you hear about Eugene Goostman? He’s the 13-year-old Ukrainian boy who wowed the computer science community over the weekend. There is something special about Eugene. He’s not quite what he seems. Or, more precisely, the most special thing about Eugene is how he seems.

Eugene is, in fact, a “chatterbot” computer program designed to seem convincingly like a young teen from Odessa. A demonstration over the weekend at the Royal Society in London found that Eugene is largely indistinguishable in text communication from a human. The event's organizers claim that “he” is the first computer program to pass the Turing test. First described by Alan Turing in 1950, the test set the bar for “intelligence” at convincing 30 percent of human judges that they are communicating with a member of their own species.

By that measure, Eugene is a thinking, intelligent computer. The program convinced 33 percent of the 30 participating judges that they were interacting with a human. (That said, Eugene only passed the test one time — and a real Turing test should be repeated often, for consistency).

Eugene is impressive, but he isn’t scary. He is neither the prototype nor the grandfather of our oft-hypothesized robot overlords. The computer program’s success in passing the Turing test owed significantly to his creators’ decision to make him appear like a child using English as a second language, such that gaps in his knowledge or eloquence could be read by judges as human mistakes. (Have you met a 13-year-old boy recently? One wonders, after all, how many human 13-year-old boys would fail a Turing test.)

Furthermore, Eugene is really a conversation simulator rather than a thinking machine displaying artificial general intelligence (computer-based reasoning capable of performing intellectual tasks that a human mind can carry out). Eugene’s performance is no harbinger of the robot revolution.

As Alex Hern wrote in the Guardian of Eugene’s success, “chatbots are a fairly limited application of the sort of artificial intelligence which science-fiction authors have been imagining for decades. By having to pretend to be human, they are prevented from being more than human.”

Science fiction and futurism readily lead us astray when considering milestones such as a computer program passing the Turing test, which was never intended as predictor of machine superpowers. Instead, it served as a philosophical comment on what constitutes thinking and intelligence. Turing proposed that a computer could be said to be thinking if, and only if, it seemed as such to enough other humans.

However, in order to convincingly display human intelligence a computer program must also contain flaws and foibles enmeshed in the human mind. Tellingly, a cognitive science professor at the University of California, Irvine, is working to develop neurotic and obsessive-compulsive robots to better replicate humans.

“The brain has incredible flexibility and adaptability,” Professor Jeff Krichmar told Discovery News. “If you look at any artificial system, it’s far more brittle than biology.”

Humans are not brittle; we are softer, messier, more afraid.

In this way, the development of more human-like machines stands at odds with the idea of technological singularity — a hypothetical moment posited by some futurists when super-intelligent machines will bypass human abilities and change civilization beyond what our piddling meat brains could ever imagine.

A program like Eugene, though relatively crude, theoretically brings us closer to the reality imagined in films like Spike Jonze’s Her — of human-like digital counterparts that can speak unbroken English, pass as human adults, and understand the conversations they are having — rather than an emergent “singularity” determined by the whims of super-human computer thought. Eugene and future programs that expand his ability to comprehend information and mimic communication are really no scarier than the proliferation of human thinking that already exists.

If there is a variety of artificial intelligence to fear, we shouldn’t regard Eugene as its root. At their very worst and their very best, Eugene’s successors will be all-too-human.

Follow Natasha Lennard on Twitter: @natashalennard

Image via Flickr