Comment by lelanthran

> If you still believe LLMs are "autocomplete", your cache of understanding about them needs invalidating and regenerating.

Autocomplete in a feedback loop is still autocomplete, no?

Doesn't the process look like this:

    (context + prompt + "reason about this")
                       |
                       V
                Reasoning Output
                       |
                       V
    (everything + Reasoning Output + "Now do final output")
                       |
                       V
         (Final output seen by prompter)

???