Human brains runs on Potatoes
summary
Claim: human brains run on 20 watts yet perform general intelligence from poetry to plumbing, offering evolutionary priors that models lack. Proposal: build state‑of‑the‑art foundation models in India using constrained compute rather than waiting for access to 100,000 H100s, leveraging small teams (he cites 600 talented people available nationally and 100-core teams) and innovations from papers like "deeps" to trade scale for algorithmic and data efficiency.
AI-written summary · about
related talks
27mComputers for Super-Human AI
01h 16mRethinking Computing with Neuro-Inspired Learning: Algorithms to Hardware
58mStrong Generalization from Small Brains and No Training Data
21mOrganoid Computing – Can Nature Save AI?
52mHuman Brain vs AI Brain: Can We Design a Truly Human-like AI?
01h 14mTissue vs silicone musings on the future of deep learning hardware and software
51mComputing like the Brain: the Path to Machine Intelligence
43mKeeping Moore's law alive: Neuromorphic computing