Vi Hart's Big Navier-Stokes Post

Commenters reacted strongly to Vi Hart’s critique of AI hype, with ivraatiems calling it an artful takedown that frames frontier AI labs as “thieves” who turned stolen human work into a flashy tool rather than prophets of utopia or doom. Gavinsyancey argued the core problem is meaning: large-scale enumeration or brute-force proof search is like a Library of Babel - without human context most outputs are vacuous, so AI successes indicate heavy reliance on human-created data and curation. Several contributors emphasized that formalized or algorithmic proofs are only another tool that can be folded into human-led practice rather than replace it. ComplexSystems pushed back at moral analogies equating AI collaboration with complicity in great crimes, calling such comparisons excessive.

Others argued that AI may already be doing things humans didn’t, and that future modalities and agentic systems will change the game. Visarga claimed training mixes have shifted toward mostly synthetic and RL-generated data, asserting the search process and discoveries can no longer be credited solely to humans and pointing to a “Move 37” moment in math. Starkshift and others forecast that world-mapping, vision, and embodied interaction will allow AI to invent beyond current human concepts. The divide is clear: some see AI as an oversold, derivative tool requiring human meaning, while others view recent methods and data shifts as evidence of a genuine, creditable leap.