Karl Friston

Karl Friston (born 1959) is a British neuroscientist at University College London whose inventions — statistical parametric mapping, voxel-based morphometry, dynamic causal modelling — became the standard tools of brain imaging, making him one of the most-cited scientists in any field. He is best known for the free-energy principle, a sweeping mathematical theory of how brains (and perhaps all living systems) perceive and act by minimizing surprise.

From theory of brains to builder of AI

In December 2022 Friston joined the Los Angeles company VERSES AI as chief scientist, betting that “active inference” — the engineering arm of his free-energy principle — can produce machine intelligence that is more efficient, more curious, and better at quantifying its own uncertainty than deep learning. He renewed the arrangement in December 2024, saying he eagerly anticipated shaping “the future of intelligent ecosystems.”

Friston is a friendly critic rather than an opponent of the current AI wave: he argues large language models are fundamentally reactive — they cannot know what they don’t know or ask their own questions — and that scaling alone won’t reach genuine agency. That position set up a public sparring match with Meta’s Yann LeCun at Davos in 2024, and aligns him loosely with fellow deep-learning skeptic Gary Marcus. Unlike most critics of the mainstream, though, his answer is to build the alternative.