Experts Correct Claims on AI MIMO Detection Result

A claim circulated that GPT-5.6 and Claude Fable produced a proof solving a MIMO detection problem first posed in 2009. Lenka Zdeborova posted a correction after consulting Florent Krzakala, stating that he and collaborators had already examined the approach. Krzakala shared an article describing a new finite-size theorem that applies approximate message passing at the maximum-likelihood threshold. Alex Dimakis noted the underlying algorithm is least squares plus greedy rounding and asked what AMP implies for it. Andrea Montanari replied that current research in the area is more advanced and that the communication risks long-term damage to science, while acknowledging the result itself may be new. Other replies reference prior AMP, state evolution, and replica formula work.