2 August 2009
Scientists fear a revolt by killer robots
Advances in artificial intelligence are bringing the sci-fi fantasy dangerously closer to fact John Arlidge
A ROBOT that makes a morning cuppa, a fridge that orders the weekly shop, a car that parks itself.
Advances in artificial intelligence promise many benefits, but scientists are privately so worried they may be creating machines which end up outsmarting — and perhaps even endangering — humans that they held a secret meeting to discuss limiting their research.
At the conference, held behind closed doors in Monterey Bay, California, leading researchers warned that mankind might lose control over computer-based systems that carry out a growing share of society’s workload, from waging war to chatting on the phone, and have already reached a level of indestructibility comparable with a cockroach.
“These are powerful technologies that could be used in good ways or scary ways,” warned Eric Horvitz, principal researcher at Microsoft who organised the conference on behalf of the Association for the Advancement of Artificial Intelligence.
According to Alan Winfield, a professor at the University of the West of England, scientists are spending too much time developing artificial intelligence and too little on robot safety.
“We’re rapidly approaching the time when new robots should undergo tests, similar to ethical and clinical trials for new drugs, before they can be introduced,” he said.
The scientists who presented their findings at the International Joint Conference for Artificial Intelligence in Pasadena, California, last month fear that nightmare scenarios, which have until now been limited to science fiction films, such as the Terminator series, The Matrix, 2001: A Space Odyssey and Minority Report, could come true.
Robotic unmanned predator drones, for example, which can seek out and kill human targets, have already moved out of the movie theatres and into the theatre of war in Afghanistan and Iraq. While at present controlled by human operators, they are moving towards more autonomous control.
They could also soon be found on the streets. Samsung, the South Korean electronics company, has developed autonomous sentry robots to serve as armed border guards. They have “shoot-to-kill” capability.
Noel Sharkey, professor of artificial intelligence and robotics at Sheffield University, warned that such robots could soon be used for policing, for example during riots such as those seen in London at the recent G20 summit. “Is this a good thing?” he asked.
Scientists are particularly worried about the way the latest, highly sophisticated artificially intelligent products perform human-like functions.
Japanese consumers can already buy robots that “learn” their owner’s behaviour, can open the front door and even find electrical outlets and recharge themselves so they never stop working.
One high-tech US firm is working on robotic nurses, dubbed “nursebots”, that interact with patients to simulate empathy. Critics told the conference that, at best, this could be dehumanising; at worst, something could go wrong with the programming.
The scientists dismissed as fanciful fears about “singularity” — the term used to describe the point where robots have become so intelligent they are able to build ever more capable versions of themselves without further input from mankind.
The conference was nevertheless told that new artificial intelligence viruses are helping criminals to steal people’s identities. Criminals are working on viruses that are planted in mobile phones and “copy” users’ voices. After stealing the voice, criminals can masquerade as a victim on the phone or circumvent speech recognition security systems.
Another kind of smartphone virus silently monitors text messages, e-mail, voice, diary and bank details. The virus then uses the information to impersonate people online, with little or no external guidance from the thieves. The researchers warned that many of the new viruses defy extermination, reaching what one speaker called “the cockroach stage”.
Some speakers called for researchers to adopt the “three laws” of robotics created by Isaac Asimov, the science fiction author, that are designed to protect humanity from machines with their own agenda. Each robot, Asimov said, must be programmed never to kill or injure a human or, through inaction, allow a human to suffer. A robot must obey human orders, unless this contravenes the first law. A robot must protect itself, unless it contravenes either of the first two laws.
While many scientists fear artificial intelligence could run amok, some argue that ultrasmart machines will instead offer huge advances in life extension and wealth creation.
Some pointed out that artificial intelligence
was already helping us in complex, sometimes
life-and-death situations. Poseidon Technologies,
the French firm, sells artificial intelligence
systems that help lifeguards identify when a
person is drowning in a swimming pool.
Microsoft’s Clearflow system helps drivers to
pick the best route by analysing traffic
behaviour; and artificial intelligence systems
are making cars safer, reducing road accidents.