If you believe Elon Musk, Steve Wozniak, and Professor Stephen Hawking, The Singularity – the point at which artificial intelligence outgrows the human race – will trigger the apocalypse, Terminator-style. Google-owned DeepMind, recognising the fear of creating a real-life Skynet, has announced plans to create an “off switch” for AIs. Dubbed the “big red button,” the AI control system is outlined in Safely Interruptible Agents [PDF], a new paper on automation published by the Machine Intelligence Research Institute (MIRI).
DeepMind is the AI research outfit responsible for building AlphaGo, the artificial intelligence that beat Go grandmaster Lee Se-Dol over five matches. But the company sees the future of its AI systems extending beyond games and into automation of tasked traditionally handled by humans.
“Reinforcement learning agents interacting with a complex environment like the real world are unlikely to behave optimally all the time. If such an agent is operating in real-time under human supervision, now and then it may be necessary for a human operator to press the big red button to prevent the agent from continuing a harmful sequence of actions—harmful either for the agent or for the environment—and lead the agent into a safer situation.”
But what if the evil AI works out how to bypass the “big red button”? Don’t worry – DeepMind has got it covered.
“Safe interruptibility can be useful to take control of a robot that is misbehaving and may lead to irreversible consequences, or to take it out of a delicate situation, or even to temporarily use it to achieve a task it did not learn to perform or would not normally receive rewards for this,” the paper adds.
Your move, Skynet.
