AI is making output cheaper. It's not making design cheaper.

I don't think the change in UX right now is "we'll all become prompt engineers". It's that judgement, verification, and accountability are becoming a bigger part of what we do.

I also don't fully agree with "AI is like a junior designer" — despite having used this term previously. It's less like a junior designer, and more like a high-speed intern with no accountability. It can generate options fast. It cannot reliably explain trade-offs, hold context, or carry responsibility for outcomes.

That shift has implications that need thinking about.

1. The real skill is verification, not generation

AI can draft flows, copy, UI variants, and edge cases in seconds. The differentiator is what you do with that output.

• What evidence are we using?

• What assumptions are we making?

• What did we validate with real users, not "synthetic" users (an entirely different rabbit hole…)?

• Where are the failure states, safety rails, and recovery paths?

Generation is the easy part now. Critical evaluation — and the ability to stand behind a decision — is where the real design work lives.

2. "Lazy design" is an operating model problem

AI will amplify whatever culture already exists. If you reward speed and deliverables over outcomes, you'll get faster activity without impact. More screens, faster. More copy, faster. None of it connected to whether it actually works for users.

This isn't an AI problem. It's a leadership and incentives problem that AI is about to make very visible.

3. The junior pathway gets squeezed unless teams build it on purpose

Junior growth is built on repeated delivery: shipping smaller pieces of work, getting critique, learning constraints, and earning trust. If AI absorbs more of the "first draft" delivery, that entry route cannot just vanish.

Teams will need a more deliberate apprenticeship model: structured critique, research exposure, QA of AI-assisted outputs, and progressively supervised decision-making. It won't happen by accident.

4. Seniors won't just be "more productive" — they'll be more accountable

There's a push towards being more T-shaped, and it's something I've tried to build throughout my career. Not in a "learn everything" way — more: strong product judgement and enough literacy across data, systems, and delivery to set constraints and standards.

If you can't explain why a design decision is right, AI will happily produce something that looks right. We've all seen examples of this. And ultimately, it's users who pay the price when it fails.

The skill isn't generating more. It's knowing when something isn't good enough, and being able to articulate exactly why.