Generative AI shifts the product designer's job, it doesn't replace it
Generative AI can produce screens, variants, interface copy, at a speed that makes mockup production almost free. The easy conclusion would be that the product designer’s job shrinks. I believe the opposite. When producing becomes cheap, value moves toward what AI doesn’t do for you: framing, arbitrating, judging.
What AI makes cheap
The visible part: generating plausible interfaces, declining them, filling them in. That’s real, and useful, it shortens the trip from idea to something you can look at. But a plausible mockup isn’t a good decision. It doesn’t tell you whether you’re solving the right problem.
What gains value
Three things, exactly the ones I ground my work in:
- Framing the right problem. An AI answers the question you ask it well; it doesn’t choose the question. Yet most bad features come from bad framing, not bad execution.
- Arbitrating between constraints. Creative vision, technical constraint, real user need: the trade-off between the three remains a human judgment, situated, informed by the trade.
- Judging against reality. Measuring, confronting real users, knowing when the best decision is to change nothing. AI has no skin in the game.
The real risk
The danger isn’t that AI replaces designers. It’s that it makes producing plausible screens so easy that we skip the framing step, that we ship something correct fast rather than right. The job, more than ever, is about slowing down in the right place.
In short
- Producing gets cheap; deciding doesn’t. Value moves toward judgment.
- AI answers the question, not the choice of question. Framing stays human.
- Arbitrating constraints is situated. It needs trade knowledge AI doesn’t have.
- The risk is speed. Shipping the plausible instead of the right.