Google's artificial intelligence expert hopes his machines won't take over the world

One of these days I really need to get around to ordering a DVD copy of one of my favorite science fiction films from my youth. Released in 1970, Colossus: The Forbin Project was a masterpiece of scientific misadventure for its era which gave me nightmares while still providing all manner of creepy thrills at the same time. In it, a brilliant scientist develops an artificially intelligent computer which is so much smarter and than humans and able to respond to threats so nimbly that they turn over the entire nation’s nuclear weapons systems to its care. It takes roughly five minutes for the machine to figure out that people are generally a bunch of pathetic losers and proceeds to team up with a similar computer in the USSR to take over the world.

Of course, that can’t happen in the real world… or at least not yet. Enter Demis Hassabis, founder of the company DeepMind, which is now part of the Google empire. He’s in charge of all their exciting artificial intelligence work, and he’s apparently now developed a program which can play the ancient board game of Go. Unlike checkers or even chess, Go is described as a game having “more possible moves than there are atoms in the universe.” It’s simply impossible to write an algorithm which can laboriously map out every possible move compared against every possible counter-move by the opponent and match the results against some optimum destination scenario. (Which is how Deep Blue works when it plays chess.) It’s a game which requires intuition and a “sense” of how the board looks. It should be beyond the abilities of a machine to master.

Hassabis has a machine which beat the European Go champion a few months ago in a secret tournament and will play the world champion in March.

The program is apparently smart, and not like Deep Blue. One might even say that it appears to be thinking. Given the number of major league eggheads who have already expressed reservations, if not fears, about AI turning into a disaster, how is Demis feeling about the possibility? (The Guardian)

Should a super-intelligence disaster loom, history is not exactly a reliable indicator that we’ll have had the foresight to withdraw from the AI arms race before it’s too late.“When you see something that is technically sweet,” Robert Oppenheimer once observed, famously, “you go ahead and do it, and you argue about what to do about it only after you have had your technical success.” “If there is a way of guaranteeing that superior artificial intellects will never harm human beings,” Bostrom noted, decades later, “then such intellects will be created. If there is no way to have such a guarantee, then they will probably be created nevertheless.” “Success in creating AI,” Hawking neatly summarised most recently, would be “the biggest event in human history. Unfortunately, it might also be the last.”

Well, I hope not,” Hassabis deadpans. In his view, public alarmism over AGI obscures the great potential near-term benefits and is fundamentally misplaced, not least because of the timescale. “We’re still decades away from anything like human-level general intelligence,” he reminds me. “We’re on the first rung of the ladder. We’re playing games.” He accepts there are “legitimate risks that we should be thinking about now”, but is adamant these are not the dystopian scenarios of science fiction in which super-smart machines ruthlessly dispense of their human creators.

We’re just playing games. Suddenly I’m thinking less about The Forbin Project and more along the lines of War Games, with Matthew Broderick.

To be fair, I’m still notas worried about a program that is strictly handling data in some geek’s laboratory nearly as much as I would be if it was tied into major manufacturing facilities with robots and tools and… hey. Wait a minute. Didn’t I hear something recently about The Internet of Things? There’s already a network out there where all manner of machines are quietly talking away to each other… refrigerators, microwave ovens, garage door openers, welding robots, conveyors. Um… they can already talk to each other and they’re hooked up to the web? And now this guy has a machine that thinks? And it’s hooked into GOOGLE?

Hopefully they can’t write political opinion columns. But in any event, I, for one, welcome our new supermachine overlords. Hopefully I’ll get a comfortable room in the reeducation camps.