Elon Musk has predicted that artificial intelligence could surpass the combined intelligence of all humans within the next five years, while warning that its rapid progress is now effectively impossible to halt.
Speaking to The Economist in an interview published on Thursday, the Tesla and SpaceX chief executive said AI would eventually outperform people in almost every discipline, with the possible exception of the human experience itself.
“I think AI may exceed the sum of human intelligence in around five years,” Musk said, adding that there would be little AI could not do “better than humans, apart from being human, perhaps”.
Despite repeatedly warning in the past about the dangers posed by artificial intelligence, Musk said he had come to accept that its advance could no longer be stopped. While he maintained that AI and robots still carried risks, he said there appeared to be “an inexorable progress” that neither he nor anyone else could prevent.
Asked whether he still believed AI carried a 10 to 20 per cent chance of causing human extinction, Musk said that possibility remained, although his outlook had shifted. “Let’s enjoy the ride,” he said, explaining that he now preferred to focus on the potential benefits rather than the threat.
Musk argued that the most likely outcome would be an era of extraordinary prosperity rather than catastrophe, with AI making it possible to satisfy almost any material need. He described the future as “an age of amazing abundance”, suggesting that people could eventually obtain almost anything they imagined.
He also said the speed of AI development had moved beyond the control of any individual or organisation, including himself, and urged leading AI companies to establish regular meetings on safety and security. Musk proposed weekly or fortnightly discussions in which rival firms could examine new AI models before they were released.
According to Musk, competitors were better placed than governments to assess whether a new AI system posed a security risk because “the competitors can keep each other honest”. He added that such arrangements should be introduced quickly, noting that AI breakthroughs were now announced at an extraordinary pace, “sometimes multiple per day”.