Editorial illustration for Jeff Bezos funds hunt for brain's core algorithm; baby learns in 200K utterances
Jeff Bezos funds hunt for brain's core algorithm; baby...
Jeff Bezos funds hunt for brain's core algorithm; baby learns in 200K utterances
Rob Williams knows the Bezos playbook: write a press release as if the product already exists, send it, and wait for a thumbs‑up or down. He used that script countless times on Amazon’s S‑team, shepherding software like Alexa, until he left the company last fall. In December 2025 he turned the same tactic on its head, teaming with neuroscientist‑entrepreneur Thomas Reardon to pitch Bezos not as a boss but as a backer.
The note that landed on Bezos’s yacht read, in stark terms, that Flourish was tackling AI’s two toughest hurdles—power efficiency and continuous learning—by building Cortex AI, a synthetic system meant to mirror the brain’s compute capacity, learning speed and energy budget. A month later, the trio met over lunch in New York’s Flatiron district, where Reardon cut to the chase: today’s large language models guzzle power, consuming more than thirty times the 20 watts a human brain uses to think. The implication is clear—if AI is to keep advancing, it may need to emulate the brain’s efficiency, not just its scale.
The takeaway is that we want to do data collection across the nano, micro, and meso scales to support the discovery of the core algorithm,” says Sean Bittner, a computational neuroscientist who also worked with Reardon at Meta.
Why this matters
Can a well‑funded team actually decode the brain's so‑called core algorithm? Bezos’s backing gives Williams and Reardon resources that most labs lack, and the promise of matching a baby’s language acquisition—roughly two hundred thousand utterances—with artificial systems is alluring. Yet the article admits they have not yet figured out how to replicate that magic, and their confidence rests on assembling “expert, well‑resourced” researchers side by side.
For developers, the implication is simple: more data‑driven language models may still fall short of human‑like efficiency without deeper neuroscientific insight. Founders should note that even with Amazon‑level capital, the path from hypothesis to product remains unclear. Researchers are reminded that interdisciplinary collaboration is being touted as a shortcut, but the outcome is uncertain.
We remain cautious, acknowledging the ambition while recognizing that the core challenge—understanding the brain’s algorithm—has not been solved. Whether this venture will yield practical tools for our community is still an open question.
Further Reading
Related Reading
Business & Startups OpenAI, a Series F San Francisco startup founded in 2015 by eight pioneers
Business & Startups Terminal-Bench 2.0 launches with Harbor, testing any container-installable agent
Business & Startups Zuckerberg Unveils Meta Compute to Build Global AI Infrastructure
From Archives Microsoft and OpenAI split; both prepare for legal battle over AI training
From Archives