Demis Hassabis, the Nobel Prize-winning AI researcher, has acknowledged that artificial intelligence may be far closer to surpassing human intelligence than he once expected.
Hassabis, the founder and chief executive of Google DeepMind, one of the world’s leading AI research organizations, received the Nobel Prize in Chemistry in 2024 for his groundbreaking work on protein structures.
A prominent figure in artificial intelligence and modern science, Hassabis is working to develop artificial general intelligence, or AGI—a hypothetical form of AI capable of matching human cognitive abilities.
Speaking to the Wall Street Journal, Hassabis said he now believes AGI—the stage at which computers could outperform humans across a broad range of tasks—may arrive within two to three years.
“It’s soon,” he said, adding that his estimate was “updating nearer and nearer.”
Hassabis had said earlier this year that AGI was likely three to five years away. In a 2025 interview with 60 Minutes, he predicted that the technology would not become part of everyday life until 2035.
His latest prediction comes as public anxiety over AI and AGI continues to grow. A recent poll by ABC News and the Washington Post found that 52 per cent of Americans believe AI is harmful to society, compared with 42 per cent who view it positively. Seventeen per cent described AI as extremely bad, while only two per cent called it extremely good.
Demonstrations against the rapid development of AI, along with demands for tighter regulation, have taken place around the world. In the United States, residents have also opposed proposals to build data centers in their communities to support the expanding technology.
Demis Hassabis, co-founder and chair of Google DeepMind, said in a recent interview that he believes AGI could be only a few years away
Hassabis received the 2024 Nobel Prize for his work on AlphaFold, an AI system that predicts protein structures. He is pictured above at a September meeting with King Charles III and Nvidia chief executive Jensen Huang
Hassabis’s warning comes amid global concern about AI and calls to slow its development. Above, protesters take part in the Nationwide Walkouts To Freeze AI in New York City
Hassabis defended AI in his interview with the Wall Street Journal, arguing that people would begin to see the technology’s benefits when those leading its development focused on tackling diseases such as cancer.
Asked whether AI could ultimately help cure cancer, Hassabis said: “I don’t see any insurmountable obstacles currently.”
IN CASE YOU MISSED IT: “Why My Son Has a Striking Resemblance with Former President Obasanjo” – Actor Ajibola Finally Opens Up – NewsFinale
“I see hard things, but not impossible things.”
He added that his team at Isomorphic Labs, a Google company, is working on AGI models intended to “revolutionize the drug discovery process” and help address multiple diseases.
“It’s slow, slow, slow—then suddenly all at once,” he said.
Hassabis’s pioneering research helped produce AlphaFold, an AI technology capable of predicting the structure of proteins.
The scientist said his 2024 Nobel Prize win for AlphaFold had reinforced his view that AGI could arrive sooner than previously anticipated.
Several former researchers from AI development companies have also warned about the risks of the race to advance the technology, leading industry figures to promise they will slow its progress.
Hassabis defended AI and AGI because of their potential to help cure diseases. Above, he stands with Princess Madeleine at the 2024 Nobel banquet in Stockholm’s City Hall, Sweden
Hassabis was featured alongside his fellow tech leaders for Time magazine’s 2025 Person of the Year issue. He said in a recent interview that he regularly speaks to all of the AI CEOs
Anthropic CEO Dario Amodei published an essay in September calling for slower AI development and stronger coordination among leading AI companies.
OpenAI CEO Sam Altman and SpaceX CEO Elon Musk posted on X that they agreed with Amodei on pacing AI advancement.
Hassabis has proposed an industry-funded body, under federal oversight, to monitor AI technology before releasing it to the public.
He argued that federal regulation would never keep pace with how fast AI develops, but noted that AI leaders regularly discuss monitoring technology.
When asked if all of the AI CEOs have a group chat, Hassabis coyly replied: ‘No. I mean, I speak to everybody. But some of them don’t get on with each other.’