Why AI can tackle Millennium Prize Problems but still gets everyday questions wrong
davidpattersonx · x · 2026-09-10
The author argues AI's uneven performance comes down to three fixable issues:
- Missing data: companies and governments don't publish comprehensive info about their products—especially the problems with them—so models can't train on it; AI agents that email, call and chat with support could gather it.
- Scale: low-parameter models can't retain everything; the largest models now exceed a trillion parameters and keep growing.
- Laziness: models skip research or reasoning; more retrieval and longer reasoning fixes it, at a cost that will fall over time.
All are data-and-scale problems the author expects solved within a few years.
More from AGI Musings
- From answering to acting: agentic AI's real challenge is governance — AryHHAry · 2026-09-10
- Lab insiders reportedly quote ~10% AI doom publicly but believe 30-90% privately — jeremiecharris · 2026-09-10
- Most non-tech users just want ChatGPT as search, health advice and email drafts — bendee983 · 2026-09-10
- Paul Christiano joins OpenAI board as AI risk hits its 'March 2020' moment — NathanpmYoung · 2026-09-10
- Beff Jezos: 90% sure Anthropic will 'synthesize bioweapons' to spur regulatory capture — AIFlow_ML · 2026-09-10
- A dung beetle and an ant score alike on benchmarks, yet differ vastly in potential — sebkrier · 2026-09-10