green_leaf comments on LLM chatbots have ~half of the kinds of “consciousness” that humans believe in. Humans should avoid going crazy about that.

[green_leaf]

Are you saying that after it has generated the tokens describing what the answer is, the previous thoughts persist, and it can then generate tokens describing them?

(I know that it can introspect on its thoughts during the single forward pass.)