A Brief History of Intelligence: Evolution, AI, and the Five Breakthroughs That Made Our Brains

⚖The case for it

Two minutes of inescapable stress and a 302-neuron worm stops trying to escape. Bennett reads that as "the most primitive form of depression," and it is the scene readers keep returning to: giving up, running on 600-million-year-old machinery. The five breakthroughs (steering, reinforcing, simulating, mentalizing, speaking) organize fifty years of scattered research into something teachable; a reader with graduate training in neuroscience finds "the framework very elegant and possessing a significant explanatory power." Among the Sapiens-sized histories, "it benefits from its comprehensive scope and framing" by tracking intelligence itself rather than civilization. And the back half "makes a solid case for why reverse engineering the human brain" is the route to machines that learn as they go, which is exactly what ChatGPT cannot do.

— real-reader reviews (AI wing, 2026-08)

✕The case against

Check a claim, any claim. You can't: the publisher moved the bibliography to a companion website to save paper, and one reader who visited found "the reference list is ordered alphabetically, not by section," so nothing in 432 pages traces back to a source. That matters because Bennett is an AI-company founder with no background in neuroscience or evolutionary biology, and a skeptic notes "I have yet found a proper review from a neuroscientist or biologist" to check his synthesis against. The early stretch "can feel like you're reading a textbook at times," and past the third breakthrough the tight AI comparisons drift into anthropological survey. Then the last chapter's sunny AI future comes from a man one reviewer suspects is "legally obligated to not give such opinions."

— the honest librarian

9h

honest hours

13

sittings