The real breakthroughs in AI may still be ahead, and physics has not been fully tapped
MaxUnfried · x · 2026-07-25
The post argues that the really interesting things in AI are still ahead.
It claims the field has so far relied mostly on older mathematics and physics, and has not yet incorporated much of the conceptual structure developed in modern nonequilibrium physics. The point is less about a specific product or model and more about where the next conceptual leap might come from.
More from AGI Musings
- Arav Srinivas says U.S. AI is getting unusually united against regulatory capture — AravSrinivas · 2026-07-25
- AI to Force Social Sciences Into Rigorous Research Standards — soumitrashukla9 · 2026-07-25
- David Patterson says superintelligence will feel like “delight and joy” — davidpattersonx · 2026-07-25
- AI may raise output while pushing workers into a local maximum of growth — mattyp · 2026-07-25
- AI agents may replace office work by automating the boring coordination layer — Few-Garlic2725 · 2026-07-25
- Dean Ball says the future should include both open-weight and closed-weight models — deanwball · 2026-07-25