AI proof of ζ(5)(3) irrationality sparks debate over what counts as research

An AI-generated proof of the irrationality of ζ(5)(3) sparked a multi-round exchange on X between researcher RexDouglass and mathematicians @littmath and @dsavitt, expanding from the value of the result itself to whether invoking AI counts as research and the professional-versus-amateur gatekeeping problem. Against the backdrop of rumors that a Clay millennium problem had been solved by AI, several bloggers and researchers issued more pessimistic judgments from institutional-critique and personal-anxiety perspectives. No consensus emerged, but the debate clearly exposed the lack of agreement in mathematics and the broader AI community over who does research and what counts as a contribution.

Confirmed

• Douglass opened by questioning how mathematicians treat AI-generated results, arguing it is the old trope that "things made by certain people don't count," and hinting mathematicians themselves have produced "flawless hallucinated work" while refusing to define contribution attribution meaningfully (m6, m10).
• Douglass repeatedly criticized "moving goalposts": mathematicians wanted solutions for decades, yet now that AI is close, they claim these don't count as "real mathematics"—a god-of-the-gaps argument where math is whatever AI cannot currently do; he also flagged historical revisionism (m4, m5, m13, m19).
• On whether AI systems "contain insight," Douglass defended AlphaProof-style systems, arguing systematic solutions to long-unsolved problems inevitably move toward computational formalization and general algorithmic search (m8).
• littmath clarified he absolutely considers some AI-generated math valuable, and criticized Douglass for rebutting positions no one held (m9).
• dsavitt added context: Frank long knew zeta5(3) was achievable with existing methods but preferred spending time on methods not obviously viable (m11).
• Douglass counterattacked that some practitioners are artificially creating a professional-vs-amateur market divide, dismissing ordinary people building with AI as losers when they are winning an entirely different game (m12).
• Douglass advised mathematicians to market themselves as reviewers and interpreters of the coming flood of AI discoveries rather than complain, criticizing those who amplify AI failures for differentiation; he described the field's response as that of "qualitative romantics" rather than scientists or engineers (m14, m17, m18).
• Douglass also noted the awkward state of AI research progress metrics: progress consists of unlocked small problems that nobody actually tracks (m15).
• On the day the millennium-problem rumor spread, Igor Carron criticized CNRS-type institutions for fixating on damaged peer review while underestimating the real shock, asking whether math and engineering research will survive (m3); @XMihura asserted math academia is already collapsing and AI will kill most knowledge work (m2).
• @rao2z framed the anxiety historically: every technological revolution destroyed some people's life meaning, as weavers experienced; the difference now is the victim class—intellectuals like mathematicians are experiencing it for the first time (m1).
• Separately, @vxnuaj announced leaving mathematics to "money maxxing," half-jokingly reflecting anxiety among researchers as model capabilities rise (m20).

Unconfirmed

• Some posts identify littmath as Terence Tao, inconsistent with other posts; this cannot be verified from the materials (m16).
• Douglass's paraphrase that the other side views amateur AI-assisted research as "wasted time and compute" is disputed; littmath publicly denied holding that position (m9, m16).
• The claim that a millennium problem was solved by AI appears only as a rumor, with no first-hand confirmation in this cluster.

Why it matters

• The core dispute—whether pressing a button to obtain a proof counts as research—shows mathematics lacks consensus on contribution attribution (m11, m16).
• The fight also reflects structural tension between professionals and outsiders over recognition as AI tools spread (m12).
• Douglass's "become the quality gatekeeper" proposal offers a pragmatic capital-facing strategy; XMihura's and Igor Carron's institutional critiques, @rao2z's historical analogy, and @vxnuaj's exit together illustrate the multi-layered impact of this wave on mathematics (m1, m2, m3, m14, m17, m20).

2026-09-07 ~ 2026-09-09 · 20 related posts

• Academic infighting over AI threats is devaluing the whole field, argues researcher — RexDouglass · 2026-09-07
• Mathematicians respond to AI like romantics, not scientists, argues commenter — RexDouglass · 2026-09-07
• RexDouglass clashes with Terence Tao over AI-assisted amateur math being dismissed as slop — RexDouglass · 2026-09-07
• [source] Debate over AI-generated proof of ζ(5)(3) irrationality splits mathematicians on X — littmath · 2026-09-07
• [source] Mathematician clarifies: some AI-generated math results do have value — littmath · 2026-09-07
• [source] Math community debates whether AI-assisted proofs of known-solvable problems deserve credit — dsavitt · 2026-09-07
• RexDouglass pushes back on AI coding gatekeeping: amateurs are winning a different game — RexDouglass · 2026-09-07
• Math community clashes over whether AI-generated results count — RexDouglass · 2026-09-08
• Math community mishandling AI, argues researcher: pitch should be verification in months not centuries — RexDouglass · 2026-09-08
• Researcher: AI progress comes from unlocked small problems nobody tracks—until it becomes algorithmic search — RexDouglass · 2026-09-08
• Researcher pushes back: solving problems humans couldn't crack does contain insights — RexDouglass · 2026-09-08
• Researchers slam AGI goalpost-moving: 'we only care what AI can't do this minute' — RexDouglass · 2026-09-08
• 'God of the gaps' argument: whatever AI can't do is suddenly defined as math — RexDouglass · 2026-09-08
• Math community mocked for shifting goalposts: "If AI can solve it, it wasn't math" — RexDouglass · 2026-09-08
• 'The math academic community is already collapsing': AI will kill knowledge work — yacineMTB · 2026-09-08
• 12-Post Thread Argues Academia Is Gatekeeping AI Math Breakthroughs With Unscientific Tactics — RexDouglass · 2026-09-09
• 'I'm Quitting Math': Researcher Says AI Progress Makes Everything 'Solved Anyway' — vxnuaj · 2026-09-09
• As AI claims a Millennium Prize problem, researchers ask whether math and engineering research will survive — IgorCarron · 2026-09-09
• AI and the Purpose of Life: This Time the Angsty Class Can Write — rao2z · 2026-09-09

1 near-duplicate retellings: RexDouglass