Using GPT to solve a decade-old problem: Will humans only need to ask the right questions?
TimothyDuignan · x · 2026-08-25
The author shares how they used GPT to solve a math/algorithm problem that stumped them for ten years. They felt their role was reduced to teaching the AI to output in an understandable format, as the AI agents seemed to communicate perfectly in highly compressed language. They believe theoretical fields relying on public knowledge and logic will undergo profound transformation, potentially leaving humans to only ask the right questions, necessitating changes in university education.
More from AGI Musings
- India's AI Community Focuses on Real Problems, Not AGI — ravi_iitm · 2026-08-25
- AI Used to Argue, Sue, and Write Books Against Data Centers — kevinnbass · 2026-08-25
- Future AI agents may run on persistent cloud machines, turning laptops into interfaces — MatthewBerman · 2026-08-25
- a16z podcast: Is AI's math progress real reasoning or a new abstraction tool? — a16z · 2026-08-25
- Whether AI can be conscious: the question stems from a misconception — AnnaCiaunica · 2026-08-25
- MIT analysis: How automation differentially impacts wages and jobs — clarashih · 2026-08-25