• Wion
  • /Trending
  • /‘Humans will be permanently unemployable when…:’ Anthropic researcher Tao Lin has a date for 'Superhuman AI' arrival

‘Humans will be permanently unemployable when…:’ Anthropic researcher Tao Lin has a date for 'Superhuman AI' arrival

‘Humans will be permanently unemployable when…:’ Anthropic researcher Tao Lin has a date for 'Superhuman AI' arrival

AI chatbots like Claude will look like child play when Superhuman AI arrives, as predicted by Anthropic's Tao Lin Photograph: (Others)

Story highlights

Anthropic researcher Tao Lin has predicted the arrival of “broadly superhuman AI" by the end of 2027. Eventually, all humans will become permanently unemployable. What's delaying this is the slow robot production, in his view.

Humans will soon be permanently unemployable, and the delay in it is only due to how slowly robots can replace them, according to an AI researcher with Anthropic. Tao Lin predicted that “broadly superhuman AI”, which matches or beats the best human experts in almost every field, will arrive in about 15 months from now. Speaking in an interview this week with Palisade Research, he gave several warnings about the future and nature of work and human relevance. In his view, while superintelligence could arrive by the end of 2027, human obsolescence will be a phased, rather than overnight, transition.

The late-2027 timeline for Superhuman AI

In the next 1 year and 3 months, AI will surpass human experts across nearly every domain, according to Lin. "I have timelines of something like 1.3 years to getting broadly superhuman AI... Or like end of 2027," he said.

This will be the milestone when AI is "as good as or better than the best human expert at almost every field", including technical and creative arenas like programming, making movies and even politics. "The strategic landscape will be just vastly different in even one year," he said.

When humans will become permanently unemployable

Trending Stories

Lin does not think mass unemployment will hit within the next 15 months when the kind of AI he described arrives. He gave it as a sequence of events. There will be a significant lag between AI gaining the capability and taking over physical labour, he said.

Immediate total unemployment will not happen because there are not enough robots deployed yet. Companies will first need to learn to build good robots and produce them at scale, he said.

The phased displacement of human workers

In Lins' view, workers will first lose their current roles to automation, temporarily pivot into alternative jobs, and "sometime after that", they'll become completely unemployable.

After this phased transition, Lin said, it is very likely that "pretty much all humans on Earth will become permanently unemployable", leaving people strictly reliant on government welfare to survive.

AI might replace humans in future, but what will happen to safety?

Lin, who works on safeguards for Anthropic's Claude models, said his late-2027 timeline is his personal forecast rather than an official position from Anthropic.

He expressed deep concern over the rapid pace of AI development, and the concentration of economic power in countries like the US and China.

He put the risk of AI eventually removing humanity entirely at roughly 20 per cent, though he more likely sees humans as comfortable yet powerless.

He also urged the industry to slow research to about a quarter of today's pace, while taking personal accountability for his role.

"People should be angry at me and everyone else developing AI," he said, while warning that society has very little time left to prepare for a post-work world.

About the Author

Share on twitter

Vinod Janardhanan

Vinod Janardhanan, PhD writes on international affairs, defence, Indian news, entertainment and technology and business with special focus on artificial intelligence. He is the de...Read More