Hassabis leaves DeepMind as essay explores seeing 90% of future

Demis Hassabis, co-founder and CEO of Google DeepMind, is reportedly stepping away from the lab he built into an AI powerhouse. In response, LessWrong user Mikewins published an essay titled "The Infinity Machine," now available on Substack, that uses Hassabis's rise and presumed fall as a lens to explore what happens when an individual can foresee 90% of the future. The piece is framed as a book review, but it functions as a speculative meditation on power, prediction, and hubris.

The LessWrong post itself is brief—just a one-minute read with a link to the longer essay—but it has already gained traction, earning 10 upvotes and a spot on the AI Frontpage. The timing is significant: Hassabis has been central to breakthroughs like AlphaGo and AlphaFold, and his departure would reshape DeepMind's trajectory. The essay's "90% of the future" conceit raises questions about whether superhuman forecasting is a blessing or a curse, both for individuals and for the AI industry as a whole. As the community digests the news, "The Infinity Machine" offers a timely, philosophical counterpoint to the usual technical discussions on LessWrong.