Demis Hassabis on the New Coming Age

Demis Hassabis, Google's AI chief, has put forward a proposal for governing frontier artificial intelligence: a standards body within the US government to evaluate and test models before release—similar in structure to FINRA, the financial regulator. Hassabis frames the stakes dramatically. He says AGI, responsibly built, could match the transformative scale of discovering electricity or fire—potentially ten times more impactful than the Industrial Revolution, and ten times faster. This, he argues, is a pivotal moment in human history. The proposal calls for frontier labs to voluntarily submit models to the standards body up to thirty days before release, with ongoing assessments to address vulnerabilities. The body could coordinate a development slowdown among frontier labs if necessary. LessWrong reports that critics—including researchers from MIRI and Anthropic—argue the proposal falls short. They contend it addresses only public deployments, not internal uses where much of the risk may lie, and that its modest scope doesn't match the existential urgency Hassabis himself describes. Some experts estimate the chance of catastrophe at fifty percent or higher—far grimmer than Hassabis publicly acknowledges.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton