I keep a note of design reviews that went in circles. Looking back, almost none of them were about pixels. They were about which problem we thought we were solving — and nobody had made that explicit.
Framing is invisible leverage. If the brief says “redesign the settings page,” you will redesign a settings page. If the brief says “users don't trust what happens to their data,” you might end up deleting half the settings page and shipping a receipt email instead.
My practice now is to write the problem as a sentence with a stakeholder in it before opening any design tool: who is stuck, what are they stuck on, and why does it matter to the business. If I can't write that sentence, I run research — not “explore layouts.”
As AI makes execution cheaper, this is where a designer's value concentrates. Generation is abundant; judgment about what deserves to exist is not.