ANKARA: Nobel laureate and artificial intelligence pioneer Geoffrey Hinton has said that putting the risk of AI causing human extinction within the next decade at around 10% is “not unreasonable,” as concerns over the rapid development of increasingly powerful AI systems continue to grow.
Hinton, widely known as the “Godfather of AI,” told BBC’s Newsnight on Wednesday that humanity is entering unfamiliar territory by developing systems that could eventually surpass human intelligence.
“We’ve never created beings that may soon be smarter than us. We don’t know what’s going to happen,” Hinton said.
Asked whether the probability of AI wiping out humanity within the next decade could be greater than 10%, Hinton said it would be “very foolish” to assume the risk was as low as 1%.
“A 10% chance seems not an unreasonable estimate to me. But nobody really knows how to give a sensible estimate,” he said.
Hinton also cautioned that highly advanced AI would not necessarily need direct control over physical infrastructure to cause widespread damage.
“The first thing to realize is it doesn’t even have to be able to act in the real world to cause devastation,” he said, adding that AI could generate “complete chaos just by talking to people.”
He further warned that sophisticated AI systems could potentially be used to develop dangerous biological or computer viruses or carry out highly damaging cyberattacks.
Hinton’s remarks come as other researchers working on advanced AI systems have voiced similar concerns.
Jacob Coxon, a 27-year-old researcher who previously worked on foundational AI models at OpenAI and Anthropic, resigned from Anthropic on Tuesday and accused both companies of moving too rapidly toward the development of self-improving superintelligence.
“Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives,” Coxon said.
He also claimed that some of those working on advanced AI privately fear that the technology could pose an existential threat before the end of the decade.
Evan Hubinger, who leads Alignment Science at Anthropic, publicly supported Coxon’s concerns on Tuesday, saying he personally believes there is a greater than 10% probability that AI could “kill all humans” within the next decade.
Hubinger said existing AI systems present little immediate danger, but argued that the greater concern lies in the possible development of superintelligence through recursive self-improvement — a process in which AI systems could autonomously develop increasingly capable successors.
Hinton said researchers must therefore find ways of ensuring that increasingly advanced AI systems do not develop objectives that conflict with human survival and well-being.
“We have to figure out how to design it so that it won’t want to,” he said.