Researcher rebuts 'zero-expertise GRH' claim: her model can't even do minor novel ML theory

BlancheMinerva · x · 2026-11-22

BlancheMinerva pushed back on Boaz Barak's claim that anyone with internal models and compute could soon solve GRH: she can't even get the Astra model to produce minor novel work in ML theory, calling the prediction incomprehensible. Barak responded that anecdotal reports say many ML theory researchers were already productively using internal models before 6.2.

Original post →

More from Models

Models channel →