When AI Solves Major Open Problems: What Does It Mean for the Future of Mathematics—and Beyond?

Summary At the MoMath 2026 Gala (“To Infinity and Beyond”), a panel of mathematicians and computer scientists discusses the immediate fallout and philosophical implications of frontier AI models autonomously solving hundreds of major open mathematics problems. The panel features Alex Kontorovich (Rutgers), Stephen Wolfram (Wolfram Research), and Ankur Moitra (MIT / IAS), moderated by a host discussing recent mass releases of AI mathematical proofs (including proofs claimed for the Quasi-Riemann Hypothesis and the Unique Games Conjecture).

What is shown

• [00:00] Title slide: “MoMath 2026 Gala: To Infinity and Beyond”.
• [00:05] Host introduces panelists Alex Kontorovich, Stephen Wolfram, and Ankur Moitra in front of the gala dinner audience.
• [01:27] Alex Kontorovich discusses leafing through the hundreds of newly announced AI solutions, detailing the claimed proof of a Quasi-Riemann Hypothesis [01:37].
• [04:10] Stephen Wolfram analyzes AI synthesis across the corpus of ~4 million published math papers versus human conceptual frameworks [04:12–06:38].
• [06:46] Ankur Moitra details findings in theoretical computer science, focusing on the Unique Games Conjecture and how the flood of automated results impacts mentoring and research aesthetics [07:07–10:07].
• [13:16] Kontorovich recaps the evolving community reaction across 2026 to AI paper drops—from the unit distance conjecture to non-sofic groups to the latest 200-page proofs [13:28–15:35].
• [16:33] Moitra and Wolfram debate computational irreducibility and whether AI requires new paradigms for mathematical explanation [16:33–19:26].

Claims & numbers

• Kontorovich states there were roughly 400 problems and solutions announced in the release [01:30], noting the Riemann Hypothesis was introduced roughly 160 years ago [01:46] and the prime number theorem about 120 years ago [02:59].
• Wolfram states there are about 4 million papers published in the mathematics literature [04:13], and mentions an automated theorem proving result he was involved with 26 years prior where the proof remains incomprehensible to humans [06:12].
• Moitra mentions jokingly asking Subhash Khot how much it would cost for his time to verify the claimed AI proof of the Unique Games Conjecture, citing “$500k” [08:32].
• Kontorovich notes that earlier in the year (May), AI proved the unit distance conjecture [13:33], followed by releases addressing non-sofic groups [14:06], and that one of the new proofs is 200 pages long [14:48].
• Wolfram states that AI can easily generate a “trillion theorems” from axioms, but determining which higher conceptual structures matter to humans is the core challenge [11:28].

Notable quotes

• [04:53] Stephen Wolfram: “Mathematics is not solving particular problems. Mathematics… the great things of mathematics have mostly been constructing new conceptual frameworks.”
• [08:31] Ankur Moitra: “Subhash’s number is 500k, so that’s the going threshold just FYI.”
• [15:20] Alex Kontorovich: “Anytime I’ve tried to sit down and articulate very carefully what it is that I’m capable of that LLMs or AI systems more generally will never be able to do, I failed at that exercise.”

Assessment This is a live panel discussion and interview recorded at a gala event. No software interfaces or live benchmarks are directly operated on screen; the discussion reflects the participants’ immediate reactions and perspectives on recent mass-scale AI mathematical proof releases.

Described by gemini-3.8-flash on 2026-10-08 from the video’s audio and frames.