Can AI end humanity? Elon Musk sees up to 20% chance, says ‘don’t teach it to lie'

Billionaire entrepreneur Elon Musk has reassessed his cost-benefit calculation of artificial intelligence (AI), saying that there is a chance of up to 20% that the technology could spell the end for humanity.

Despite the risk posed by AI, the world should move ahead with further developing the technology, added Musk, who owns AI startup xAI. The reason, he said, is that the positive scenario arising out of the wide-scale use of AI “outweighs the negative scenario”.

"I think there's some chance that it will end humanity…It's about 10% or 20% or something like that,” Musk reportedly said, while speaking at the recently held four-day Abundance Summit 2024 in Silicon Valley.

The statement marks a significant contrast as compared to his earlier AI-related risk assessment. In November, Musk had said that the chances of AI “going bad” was “not 0%”, and he refrained from linking the tech’s misuse to a potential end of humanity.

‘Don’t teach AI to lie’

According to Musk, raising artificial general intelligence (AGI) is almost like raising a kid. The kid, in this case, is a “super genius” and “it matters how you raise the kid”, he said..

"One of the things I think that's incredibly important for AI safety is to have a maximum sort of truth-seeking and curious AI,” Business Insider quoted him as saying at the Abundance Summit.

The most feasible way to achieve AI safety is to use the technology in a manner that keeps it truthful, Musk suggested. “Don’t teach it to lie…it’s very important,” he noted.

A similar concern was expressed by AI expert Geoffrey Hinton, who told CNN last year that if AI learns to lie, then it may end up manipulating users. “If it gets to be much smarter than us, it will be very good at manipulation because it would have learned that from us,” he had said.