FULL STORY

LLMs Crack Conjectures, Sparking a Math Crisis Debate

LLMs produced counterexamples to long-open conjectures, unsettling mathematicians. The community debated whether AI math is sampling rather than understanding, while theorists weighed their future role.

2026-08-24 ~ 2026-08-27 · 3 episodes · 8 posts

Episode 1 · Mathematicians in Crisis as LLMs Crack Long-Standing Conjectures (2026-08-24, 3 posts)

LLMs have produced counterexamples to several long-standing conjectures, shaking the math community. Blogger Kirwin Hampshire describes a 'dark night of the soul' identity crisis in a widely discussed essay, echoed by the 'Leiden Declaration' on AI and mathematics.

Episode 2 · 'Sampling Is Not Understanding': AI Computes but Doesn't Grasp Math (2026-08-26, 2 posts)

Gerard Sans argues that frontier AI doing math is like AlphaGo's move 37: it computes and samples, producing verifiable proofs without genuine understanding, and thus cannot replace mathematicians.

Episode 3 · Mathematicians Debate Whether AI Makes Theorists Obsolete (2026-08-26, 3 posts)

A debate flared over whether AI-generated proofs make theoretical mathematicians obsolete. Mathematician Alex Kontorovich argued that work providing no value to others will disappear, urging researchers to adapt and create value people want.