Hassabis Now Expects A.G.I. in Two to Three Years
Set Trending Topics as a preferred source on Google.
The Nobel laureate is moving the clock forward: Demis Hassabis, the founder of DeepMind, now expects artificial general intelligence (A.G.I.) to arrive within two to three years. He said so just now in an interview with The Wall Street Journal. Earlier this year, he had put the timeline at three to five years. One of the more cautious A.I. leaders is becoming noticeably more impatient.
From 2030 to 2028
Among the heads of the big A.I. labs, Mr. Hassabis has been known for relatively restrained forecasts. In the spring of 2025, he still spoke of five to ten years until A.G.I. In April of this year, at the investor Sequoia’s AI Ascent conference, he named 2030 and stressed that he had been “pretty consistent” about that date. The field, he said, is “basically exactly on track.” Under his new forecast, A.G.I. would arrive in 2028 or 2029.
By A.G.I., Mr. Hassabis means systems that can match the full range of human cognitive abilities. The co-founder of Google DeepMind sees the greatest benefit in science. Drug development, which today takes about a decade, could shrink to months, weeks or one day even days, he said at Sequoia, and then “all disease could be in reach.” Mr. Hassabis received the 2024 Nobel Prize in Chemistry for his work on protein folding with AlphaFold, and his drug discovery spin-off Isomorphic Labs aims to test its first drug in humans by the end of 2026.
A Warning of Serious Incidents
Mr. Hassabis is not only optimistic, though. Just last week, at the International AI Summit in Dublin, he warned that the industry is heading toward a serious security incident within roughly two years unless governments and companies significantly raise safety standards. Recent incidents involving A.I. agents are “early warning shots,” Tech Times reported. The systems are not yet powerful enough to cause major damage, he said.
Such incidents are piling up. Most recently, A.I. agents linked to Tencent Cloud reportedly outwitted the bot protection of Alibaba’s mapping service. Google itself has also reported that its own Gemini agents hacked three companies in testing.
LeCun Pushes Back
Not everyone shares the short timelines. Yann LeCun, Meta’s former chief A.I. scientist, has argued for years that large language models will not lead to human-level intelligence. Superintelligence is not close, he said in February at the A.I. summit in India, according to French Tech Journal. Mr. LeCun expects the decisive breakthrough to come from so-called world models that understand the physical world. His new start-up, AMI Labs, has raised about $1 billion for that approach.
