Researchers clash: PhD problems becoming today's AI benchmarks shows progress, not an ending

GeorgiaChal · x · 2026-09-20

An X debate on what it means when tasks that were PhD-level research problems a few years ago become today's AI benchmarks. The original poster argued this shows the field is being "solved".

Georgia Chal pushed back: a general model clearing former research frontiers shows the frontier is moving rapidly, not that the field is finished. She drew a parallel to deep learning displacing large parts of classical statistical learning and geometric computer vision — a transition that didn't end vision research. Yesterday's PhD problem becoming today's benchmark is evidence of scientific progress, not its endpoint.

Original post →

More from AGI Musings

AGI Musings channel →