rife comments on Recursive Self-Modeling as a Plausible Mechanism for Real-time Introspection in Current Language Models

[rife]

check it out @Nathan Helm-Burger . It appears to be possible vindication on the ‘signal to noise’ part of the hypothesis:
Anthropic Post