This presumes that LLMs can't reason, which appears to be untrue

Reply to this note

Please Login to reply.

Discussion

They can reason but they still have poor taste from what i’ve seen.

That may be true, but process can can improve it. Have them design architecture, have them write tests, have them profile, have them optimize, have them review, have them enforce standards. Have one context run this process, which is implemented by other contexts

can you reason?