The math proof — the foundational unit of academic credibility in mathematics — is losing its signaling power. AI systems, particularly those from OpenAI, can now generate proofs at scale, and the downstream effect on hiring is immediate: a publication record no longer reliably indicates that a human did the thinking. Frank Calegari, a mathematician at the University of Chicago, has laid this out plainly in a post on Terence Tao's blog. He argues that hiring committees can no longer rely on written output alone. A 30-minute conversation, he says, will separate candidates who used AI as a crutch from those who possess genuine mathematical vision. He wants to know where a candidate thinks mathematics should go — and whether they can explain the technical path to get there. Calegari's critique targets what he calls "lazy committees" — hiring bodies that rubber-stamp candidates based on a single paper or publication list. In a world where AI can produce those papers, this practice becomes an open door for credential inflation. The traditional signal is broken, and committees that don't adapt are importing noise into their departments. Cornell professor Steven Strogatz has framed it differently but arrives at the same conclusion: mathematicians with "rizz" — the ability to communicate, contextualize AI-generated results, and translate them into new research questions — are newly valuable. The technical term is communication competence; the practical reality is that interpersonal skill now functions as a filtering mechanism that AI cannot fake. The broader context is OpenAI's aggressive entry into mathematics, which has flooded the field with hundreds of proposed solutions to complex problems. This has left researchers unsettled, with some professors rejecting AI entirely and others scrambling to integrate it into their workflow. There is no institutional consensus. Students are caught in the middle. An MIT survey surfaces the complaint directly: "There is a lack of high-level leadership on all sides of academia." Students want clear guidance on whether and how to use AI, and they are not getting it. The rules of the game changed, but nobody updated the rulebook. What's emerging is a structural shift in how mathematical talent is identified and validated. The old system — publish, get cited, get hired — rewarded a specific kind of output. The new system rewards a specific kind of person: one who can think in public, communicate under pressure, and demonstrate vision that a machine cannot simulate. Whether hiring committees actually make this transition, or merely talk about it, will determine whether math departments import a generation of AI-assisted credentialists or genuine researchers.
