Researcher Notes Shifted Views on LLM Intelligence
Post recalls early disbelief that language models could be intelligent.
TLDR
Janus posted that a few years ago the idea LLMs were actually intelligent seemed insane to almost everyone. The message contrasts that past reaction with current events where the models solve open math problems. It frames the change as evidence that earlier doubts have been overtaken. The post appears among replies on X and attributes the observation directly to the author without additional claims or sources.
Combined views
7.8K
2 Sources, first seen 26d ago