The prediction was that AI would hollow out web design. The data so far says otherwise. New figures reported by The Guardian show UK design employment grew 15% between 2020 and 2025, reaching 2.27 million jobs. The wider UK design economy grew 40% between 2019 and 2023. Industry leaders describe AI as the intern in the office: fast, useful, occasionally impressive, and not the person who should be making the calls.
I'd be careful with headline numbers, though. Employment totals for a category as broad as design can hide a lot of movement underneath, with some roles growing while others shrink. The figures as reported here don't break that down, so we can't say which parts of the field are expanding. What they do show is that a sector many people expected to contract has, at least so far, grown. That's the part worth sitting with.
Where the intern analogy holds
The comparison fits web work unusually well. AI can produce a first-draft landing page, write placeholder copy, resize images, and generate some working code. Tools like this tend to be strongest where the brief is clear and the output is easy to check, which describes a lot of routine web production. It will also cheerfully produce fourteen variations of a hero section that nobody requested. That's intern work, done at superhuman speed.
The speed is real, and it changes the economics of the bottom tier. If your pitch is "I can build you a five-page site," you're now competing with a tool that can draft one before you've finished the proposal. Clients who used to wait a week for a template-based site may now expect a draft in minutes, and they'll anchor their budgets to that expectation. Mediocre web design is becoming close to free, and that's the actual threat. It isn't that the machine suddenly has taste. It's that the market no longer pays much for competent-but-generic output, because competent-but-generic output is suddenly everywhere.
What the intern can't do
AI doesn't reliably tell you why people abandon the checkout. A tool can generate a checkout flow, but it has no direct view of the customer who gave up at the shipping step, or of the support ticket that explains why. It won't notice that the brand feels off, or that the navigation makes sense only to the people who built it. It also can't explain why a technically flawless homepage looks like every SaaS site launched since 2023. Those are diagnostic questions. They require someone who understands users, business goals, and the messy history of a particular company.
Look at the list of things that matter most in this work: UX, conversion, accessibility, interaction, and strategy. None of them is a deliverable you can hand over in a folder. Each one is a set of decisions about trade-offs, and those decisions depend on context the model usually doesn't have. Here's the thing: the more output AI produces, the more valuable the person who decides which output is worth shipping becomes. Quantity stops being the bottleneck. Judgment is.
A new niche: cleaning up the intern's work
A strange new category of freelance work is taking shape. Designers are being hired to clean up bad AI-generated creative work. The sequence runs like this: AI generates the asset, and a human fixes it. That's not a flattering arrangement for the tool, but it's a workable business for the people doing the fixing. Anyone who has inherited a project from a previous vendor will recognize the pattern, since rescue work has always paid reasonably well.
My read is that this pattern won't stay confined to cleanup jobs. Clients who once paid for production will increasingly pay for review, direction, and accountability for outcomes. The job description shifts from "make this" to "make this work." That's a bigger change than it sounds, because "make this work" requires measurement, a point of view, and the willingness to tell a client that their favorite design is hurting sign-ups.
The intern does have one clear advantage. It works around the clock, never complains, and doesn't need coffee. The designers who will do well are the ones who learn to direct it, and who can tell the company when its best-looking output is quietly failing its users. (And this is the part most people overlook: the pressure to ship fast makes that conversation harder, not easier.)
What this means for the field
The honest reading of the current numbers is cautious optimism, not a victory lap. Growth in employment and in the wider design economy suggests demand hasn't disappeared, but it doesn't tell us whether the average designer is better off. Those figures could coexist with a shrinking pool of entry-level work and a rising bar for everyone else. Until the breakdowns show up, I'd treat the headline as a starting point rather than a verdict.
If you're working in this space, the practical question is which part of your job the tool can't easily absorb. Production is getting cheaper. Diagnosis, prioritization, and accountability for results are not. What this means for you is that your portfolio should show decisions and outcomes, not just finished screens. A case study that explains why a layout changed will carry more weight than a gallery of polished mockups.
Not a crisis. Not yet, anyway. The intern is useful, and it's getting better. But the people who decide what ships still have the job, and that's the part to protect.
