Jeff Hawkins is one of those rare individuals who decided to do two seemingly opposite things: invent products the whole world wanted, and study the one mystery nobody had solved. In the early 1990s, while working on the first pocket computers, he had a question that wouldn't leave him alone: how does the brain actually work? Not as philosophy, but as engineering. And when almost nobody in AI believed that brain biology could teach machines anything useful, he remained firmly convinced otherwise, stubborn and patient.
In 2004 he published "On Intelligence" (with scientist Sandra Blakeslee), the book where he introduced Hierarchical Temporal Memory (HTM): the idea that the neocortex works as a memory system that constantly predicts the future from the past, using a hierarchy of cortical columns. It wasn't a technical machine learning paper; it was a theoretical map of the brain as a prediction engine, and it became required reading for researchers wanting to build AI from biological principles. In 2021 he went further with "A Thousand Brains: A New Theory of Intelligence," proposing that the brain isn't one hierarchical system but thousands of models of the world working in parallel, an idea that could redefine how we build intelligent machines.
Hawkins founded Palm Computing in 1992 and was the architect behind the PalmPilot (1996), the first pocket personal organizer that actually worked, the one so many people carried before smartphones existed. In 2005 he co-founded Numenta with Donna Dubinsky and Dileep George to research his theories seriously. In January 2025, Numenta launched the Thousand Brains Project as an independent nonprofit, with initial funding of $2.69 million from the Gates Foundation, to turn that theory into open-source code any researcher can use. The Computer History Museum named him a Fellow in 2026.
What strikes me as valuable about Hawkins isn't necessarily whether he's right (time will tell if the thousand brains theory displaces the transformer), but that he dared ask the uncomfortable question: what if we're building AI without truly understanding what intelligence is? Every time you use Claude and it surprises you by connecting ideas that seemed unrelated, there's a spark of that question working underneath. The difference is Hawkins has had it fixed in his mind for decades, and hasn't let go.
Official links for Jeff Hawkins, The man who wanted to build a real brain
More figures who shaped AI in AI Legends, or back to the news.