Yeah this looks like another version of prompting for chain of thought which is very effective: https://
docs.anthropic.com/en/docs/build-
with-claude/prompt-engineering/chain-of-thought
… It has been "trained in" to some degree in newer models, 3.5 sonnet will usually contemplate a bit before giving an answer to hard questions you ask it but it

→ View original post on X — @alexalbert__,