AI’s real path was less theory and more data, compute, and gradient descent
burny_tech · x · 2026-07-28
A repost of a quip about how people imagined AI would emerge from deep theories of cognition, while the actual path was more like “throw the internet into a giant blender” plus enough compute for gradient descent to work.
It frames AI history as a funny mismatch between grand theories and the brutally practical scaling path that actually won.
More from AGI Musings
- A long-form AI safety debate weighs open models against biological risk — bookwormengr · 2026-07-28
- Miles Brundage says AI discourse has matured, but open-weight debates have not — Miles_Brundage · 2026-07-28
- Poster says no one serious is still working on deep learning or reinforcement learning — fkasummer · 2026-07-28
- LLM mania could still delay AI diffusion by rattling the economy — fkasummer · 2026-07-28
- AI-for-math has hit short-horizon wins, but the next leap may need deeper theory — burny_tech · 2026-07-28
- AI should not be steered by a handful of companies, the post argues — ctjlewis · 2026-07-28