AI is changing the designer's workflow, not replacing itAI is changing the designer's workflow, not replacing it

The debate is usually framed as replacement, which does not match how the tools are being used. Industry surveys through 2026 put weekly AI use among designers above ninety percent, and daily use at roughly three in four — but the reported use cases cluster around ideation, production and iteration rather than final craft.

That matches my own week. Generated imagery is the part I use least: sports work depends on real photographs of real players, and an identity cannot be prompted into existence because its constraints belong to a client nobody else has. What has changed is the stretch between an approved direction and delivery — the same idea at eleven sizes, the twentieth variation of a lockup.

It is also useful early, as a way of being wrong faster. Twenty rough directions in an afternoon, nineteen discarded, is a reasonable way to find out what you actually think. That is not the same as using any of them.

Where it does not help is judgement. Choosing between two lockups for a particular client selling to particular people is the job, and no prompt has carried that context yet. The designers getting the most out of these tools are not the ones with the best prompts — they are the ones who already had standards clear enough to direct a machine against.

The honest summary: it moved the tedious middle, left the hard parts alone, and made it much easier to produce a large volume of mediocre work quickly. The last of those is the one worth watching.