Even Grigori Perelman took years to have his proof of Poincaré conjecture accepted, and he had a Ph.D. and a Berkeley fellowship.

Expecting a human author to put substantially more effort into a proof than needed to verify it is oversimplifying. It is more a matter of credentialing and collegiality in mathematics whereby someone’s reputation and work-product demonstrates that a purported proof put forward for review is likely to be true or at least a valuable or interesting contribution even if imperfect.

AI makes this a much bigger challenge.