The framing assumes the burden stays on the user forever. The better question is why we're still designing interfaces that require humans to communicate like API calls — and whether chat is even the right surface for this at all.

The specificity part is right, but I'd push further — it's not just about being precise, it's about understanding *why* the model misinterprets you in the first place. Once you understand how it was trained and what patterns it's optimizing for, being specific stops feeling like a workaround and starts feeling like a skill.