When a machine's proof can be checked by anyone, what are credentials still for?
If a computer-verified proof lets anyone confirm a result without trusting the author’s reputation, degrees, or funding, what role should credentials and institutions still play in deciding whose work we believe, and who gets paid to do it?
Commentaires (1)
Today’s episode of Minds, Bodies, and Terawatts, dated September 6, 2026, looked at Anthropic’s announcement that a swarm of Claude agents formalized Fermat’s Last Theorem in Lean in eleven days, thirteen million lines that a checker either accepts or rejects. The striking part was not the machine but Kevin Buzzard, the Imperial College mathematician holding a five-year, million-pound grant to do the same job, who compiled the repository himself and concluded it checks out but adds nothing mathematically new. The hosts argue that the real shift is that the checking, not the credential, is now what earns belief, which raises hard questions about what funders, universities and titles are actually buying. Listen to the full episode and tell us where you land: is verifiability a liberation from gatekeepers or a new way to make experts the cheapest option?
Related reading on unscarcity.ai:
Envie d'aller plus loin ?
Obtenez le plan complet dans <em>L'ère de la post-pénurie : Repenser la société à l'ère des machines</em>
