AI collapsed the production layer of design. A competent model now returns a plausible screen, a serviceable component, or a boilerplate flow in minutes. What used to signal skill is now a commodity, and that changes everything about where designer value actually lives.
John Maeda's argument in 'Practicing CRAFT: Keeping Ahead of the Machine' is the spine of this piece: craft relocated upstream, from making to judgment. The efficient answer and the right answer are not the same answer. When a tool generates a thousand plausible screens before lunch, the scarce skill is knowing which one deserves to exist. Teresa Torres adds the operational counterweight: continuous discovery, weekly customer contact by the team building the product, is now the cheap insurance against a machine that is wrong in fluent, well-formatted prose. The article walks through exactly how to audit your own deliverables, separate the parts a machine could regenerate from the parts where you owned the decision, and defend your calendar for the second kind.
Read the full piece for the action items, which are blunt and specific. Write two sentences defending one direction before anyone asks. State the problem in one sentence before opening any tool. The argument is not that polish stops mattering. It is that polish is table stakes, and the differentiator is the judgment that points cheap production at a problem worth solving in the first place.
[READ ORIGINAL →]