Top Mathematicians Call LLMs Strong Calculators but Poor Creative Thinkers
The Decoder · rss · 2026-08-16
Renowned mathematicians Timothy Gowers and Peter Sarnak evaluated LLMs' capabilities, stating that while they are good at combining known methods for calculation, they lack the intuition for genuinely new mathematical ideas.
More from AGI Musings
- Researchers Fear Leaking Hypotheses, But Future AIs Can Spell — TheZvi · 2026-08-17
- User cites Jensen Huang: AI-nuclear analogy spreads fear, need action — arieljalali · 2026-08-17
- Simon's ant parable applied to LLM vs harness debate — rao2z · 2026-08-17
- Model vs harness: Simon's parable urges focus on interplay — rao2z · 2026-08-17
- Simon's Ant: Is LLM intelligence in the model or the harness? — rao2z · 2026-08-17
- User counters LeCun: AI and nuclear have dual-use potential — arieljalali · 2026-08-17