Google DeepMind CEO predicts AGI breakthrough by 2029
28 May 2026

Demis Hassabis, the CEO of Google DeepMind, has predicted that Artificial General Intelligence (AGI) could be here as early as 2029.

The prediction was made speaking to Axios after his Google I/O appearance.

Hassabis said we are at the "foothills of the singularity," and society may have only a few years to prepare for this major technological shift.

Need for societal preparation
Urgency emphasized

Hassabis said he expects AGI around 2030, but now sees 2029 as a real possibility.

He stressed the need for governments, economists, and society at large to prepare faster for this impending change.

The AI industry is still divided on what exactly qualifies as AGI. Sam Altman of OpenAI describes it as a system that can solve increasingly complex problems at a human level across many fields.

Self-improvement and coding agents
Self-improvement potential

One major milestone among researchers is that AGI-level systems may eventually be able to improve themselves at an accelerating rate.

However, Hassabis noted that current systems aren't fully autonomous in their improvement yet.

He described today's coding agents as a form of "soft self-improvement," already making engineers significantly more productive.

Despite the lack of a clear definition, every major AI company seems to agree on one thing: AGI is coming sooner than we think.

How to identify real AGI
AGI test

Hassabis also shared his own benchmark for determining whether AI has truly reached AGI.

According to him, a real AGI system should be capable of independently discovering entirely new scientific ideas, much like humans once did.

He said, "The kind of test I would be looking for is training an AI system with a knowledge cutoff of, say, 1911 and then seeing if it could come up with general relativity like Einstein did in 1915."