Stephen Hawking: artificial intelligence could kill off mankind
Professor Hawking fears that AI 'would take off on its own, and redesign itself at an ever-increasing rate
Artificial intelligence is a threat to human existence, Stephen Hawking, one of Britain’s best known scientists, has warned.
“The development of full artificial intelligence could spell the end of the human race,” he said in an interview yesterday.
His warning came in response to a question about a revamp of the technology he uses to communicate, which incorporates a basic form of AI.
The theoretical physicist, who has the motor neurone disease amyotrophic lateral sclerosis, or ALS, was using a new system developed by Intel to speak.
Machine learning experts from the British company SwiftKey were also involved in its creation. Their technology, already employed as a smartphone keyboard app, learns patterns that the physicist tends to use and suggests which words he might want to use next.
Professor Hawking told the BBC that “smart” algorithms designed for specific tasks were proving useful, but that he feared the consequences of creating “general” artificial intelligence, which could potentially surpass human abilities on a broad range of tasks.
“It would take off on its own, and redesign itself at an ever-increasing rate,” he said. “Humans, who are limited by slow biological evolution, couldn’t compete, and would be superseded.”
His fears echo claims by the technology entrepreneur Elon Musk that AI is “our biggest existential threat” — despite Mr Musk having been an early investor in DeepMind, an AI company bought by Google earlier this year.
Professor Hawking said he was looking forward to being able to write much faster with his new system. However, he insisted that he did not want his robotic voice upgraded to sound more natural.
“It has become my trademark, and I wouldn’t change it for a more natural voice with a British accent,” he said. “I’m told that children who need a computer voice want one like mine.”
His comments came as scientists suggested that a new benchmark was needed to measure machine intelligence.
The current test, which was set down by Alan Turing in the 1950s, requires a computer to fool a human judge into thinking they are talking to another human. A group of scientists have proposed an alternative, which challenges computers to understand the meaning of sentences using specially designed comprehension questions.
http://www.thetimes.co.uk/tto/science/a ... 285351.ece
