Universities should be bolder: LLMs are consensus machines, not discovery engines

_akpiper · x · 2026-11-21

Hollis Robbins argues universities must be bolder in the AI era. Her core claim: LLMs are optimized for safety and consensus, not discovery—they handle known unknowns, always demand evidence, and stall or reframe when none exists, making them better called consensus machines. Ask an LLM to find something nobody has found and you get either small recategorizations of the known or unverifiable conjecture. Since the AI enterprise was largely built outside academia, and LLMs are now safe and useful for undergraduates, she argues this is the moment to invest in originality and discovery.

Original post →

More from AGI Musings

AGI Musings channel →