Hassabis Predicts AGI Around 2030, Urges Society to Prepare

Last week at the Stanford Graduate School of Business, DeepMind CEO Demis Hassabis told an audience that artificial general intelligence could arrive around 2030, “plus or minus a year,” and that society does not have much time to prepare for its economic and social effects.

Hassabis defined AGI as systems capable of performing a broad range of intellectual tasks at or beyond human levels. He said recent advances in AI agents and tool-use have clarified the remaining steps toward systems that can generalize across many problems.

He pointed to 2026 as a turning point in how AI is applied to work and development, with agents becoming genuinely useful in practical tasks. At the event he predicted: “I believe that we’re only a few years away from that, maybe like 2030 plus or minus a year.” He added that, in retrospect, people might view the present as the start of a long-term change.

Hassabis warned that planning for the consequences of AGI cannot be left to technologists alone. He said, “Society needs to hear that because we don’t have long to prepare for what that means. It’s going to be enormously profound,” and urged wider public discussion about economic disruption, labor shifts and governance before more advanced systems become common.

Opinions within the industry vary. Some leaders have predicted AGI within a similar near-term window, while others have suggested current model capabilities are already approaching broad intelligence. Several founders and researchers argue that present frontier systems meet some definitions of general intelligence.

Other experts remain skeptical. Tests designed to measure adaptive learning in unfamiliar environments have found major models scoring well below human performance. One benchmark reported leading models scoring under 1 percent while human participants achieved perfect results, a gap cited by those who say more development is required before calling current systems AGI.

The lack of a single, shared definition of AGI contributes to the debate. Malo Bourgon, CEO of the Machine Intelligence Research Institute, noted that differing definitions make it hard to determine when a system should be classified as AGI.

Hassabis predicted substantial change over the next decade and called for broader societal engagement to consider how choices made in the coming years will shape the economic and daily impacts of more capable AI systems.