AI math wins aren't '10 years of progress in a month,' researcher argues against explosion analogy

joshua_saxe · x · 2026-11-19

Greg Burnham pushes back on framing recent AI math results as "10 years of math progress in a month," calling that analogy to future AI R&D acceleration strained.

His read: current models are "brilliantly optimizing current paradigms," which likely doesn't yield a strong intelligence explosion. That could change, he notes, if AI soon develops better mathematical taste. A measured middle position in the intelligence-explosion debate.

Original post →

More from AGI Musings

AGI Musings channel →