Scott Aaronson eulogizes strange loops; davidad pushes back on his complexity-theory lens

burny_tech · x · 2026-09-04

Andrew Curran relayed Scott Aaronson's argument delivering a eulogy for the idea that strange loops and self-referentiality were the secret of intelligence. davidad responded with respect but argued Aaronson is trapped in a complexity-theory worldview — assuming a stack of self-attention layers obviously has nothing to do with self-referentiality just because it has a finite depth limit. The exchange touches the long-running debate over whether Transformers can host self-referential structure.

Related event: Scott Aaronson Explores Self-Referentiality in LLMs(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →