Melanie Mitchell backs claim that today's systems aren't really LLMs, citing original definition

MelMitchell1 · x · 2026-09-27

Santa Fe Institute professor Melanie Mitchell responded to Aidan Clark's claim that "what we have now are not LLMs," agreeing unless the term has been redefined. She argued the Stochastic Parrots paper used the original definition: a language model is a statistical model of language, not of all the extra content today's post-trained AI systems have absorbed. The exchange feeds into an ongoing debate about how far the LLM label has drifted from its technical meaning.

Related event: Stochastic Parrots Debate: Researchers Argue Whether Modern Models Still Count as LLMs(11 posts)→

Original post →

More from AGI Musings

AGI Musings channel →