DeepMind founder Hassabis says AGI could arrive in two to three years
Demis Hassabis, Google DeepMind founder and Nobel laureate in chemistry, now believes AGI could arrive within two to three years. Earlier this year he put the figure at three to five years. He told the Wall Street Journal: "It's going to happen soon. My estimate keeps moving closer to the present."
Hassabis defines AGI as a system matching the full cognitive range of the human brain. His litmus test is what The Decoder calls the "Einstein test": could a system trained only on pre-1901 knowledge reproduce Einstein's 1905 papers?
He has compared AGI's impact to "10 times the industrial revolution at 10 times the speed."
Since stepping down from Google DeepMind, Hassabis has focused on Isomorphic Labs, Google's drug discovery AI spin-off. On safety, he recently pitched an industry-funded oversight board under government control to vet models before release. Bad regulation, he warns, is worse than none.
The piece is a short report on the interview. It gives Hassabis's own statements and no technical evidence for the shorter timeline.
Key facts
- Hassabis now says AGI could arrive within two to three years; earlier this year his estimate was three to five years.
- He defines AGI as a system matching the full cognitive range of the human brain.
- His test: could a system trained only on pre-1901 knowledge reproduce Einstein's 1905 papers?
- He has compared AGI's impact to "10 times the industrial revolution at 10 times the speed."
- He recently pitched an industry-funded oversight board under government control to vet models before release, and warns that bad regulation is worse than none.
Why it matters
Hassabis leads from a position of authority: he is the Google DeepMind founder and a Nobel laureate in chemistry. When someone with that profile moves his own AGI estimate closer, from three to five years down to two to three, it feeds the wider debate about how soon advanced AI is coming. He frames the stakes as 10 times the industrial revolution at 10 times the speed. He also gives a concrete bar for AGI, the Einstein test, rather than leaving the term vague.
Who it affects
Anyone who plans around AI timelines: AI labs, companies and policymakers. The oversight board idea, industry-funded and under government control, is aimed at how models get vetted before release. Hassabis himself now works on Isomorphic Labs, Google's drug discovery AI spin-off.
How to use it
Treat it as a data point on how one prominent figure sees the timeline, not as a product or method to adopt. The Einstein test offers a simple yardstick for judging claims that a system is AGI: can it, trained only on pre-1901 knowledge, reproduce Einstein's 1905 papers?
How solid is it
This is a forecast, and the modality matters: Hassabis says AGI could arrive within two to three years, not that it will. The report is a brief secondhand account of what he told the Wall Street Journal. No technical basis or evidence for the shortened timeline is given beyond his own statement, and the source gives no start point from which the two-to-three-year window is counted.
Risks and caveats
The date of the Wall Street Journal interview is not given. The source does not say whether any system has passed or been tested on the Einstein test. No details of the oversight board proposal are given beyond "industry-funded" and "under government control". The source gives only one earlier estimate, from earlier this year, and Hassabis himself says his estimate keeps moving, so the figure should be read as a moving personal estimate.
“It's going to happen soon. My estimate keeps moving closer to the present”
— Demis Hassabis, to the Wall Street Journal, as reported by The Decoder