What Happened
The Guardian reported on 7 September 2026 that industry bodies expect generative AI to enhance the work of designers rather than replace them, with firms more likely to use the technology as “the intern in the office” than as a substitute for skilled staff. According to the report, professional designers should not feel “threatened” by the rapid growth of generative AI, despite fears over job losses in the sector.
The Guardian noted that design and film production companies and manufacturers are all adopting AI at an accelerating pace. The claim from industry leaders is not that adoption is slowing. It is that the technology changes which parts of the job a person does, rather than removing the job.
That pattern has precedent. Desktop publishing in the 1980s and template site builders in the 2000s both put production tasks within reach of non-specialists, and both increased the total volume of design work being commissioned rather than ending it. The scarce skill moved up the stack, towards judgement, brand consistency and knowing which of 40 generated options is worth putting in front of a client. Generative tools sit in the same place. They produce output quickly, but they do not decide what the output is for.
For a UK business, the useful question is not whether AI removes your design team. It is what happens to quality control when the cost of producing a first draft falls close to zero.
Where AI actually fits in a design workflow
Separate generation from decision. AI software is fast at producing variations: 30 hero images, 12 headline options, a dozen layout directions. It is weak at knowing which one matches your brand, your audience and the claim your legal team will sign off. Teams that get value from it treat the output as raw material a designer edits, not as finished work to publish.
Look at your brief before you look at the tools. Most disappointing AI design output traces back to a vague brief rather than a weak model. If your in-house team or agency cannot state the audience, the single action you want and the constraints in writing, no AI software will fill that gap. It is the part of our AI services work that surprises clients most often, because the setup takes longer than the generation.
Then decide who signs off. If AI drafts the first version of imagery, copy and layout, someone still has to own accessibility, colour contrast, alt text and brand consistency across templates. Assign that person by name. Without it, you get slow drift, and a site that started coherent ends up as a set of mismatched landing pages that nobody remembers approving.
The skill that matters is editing, not prompting
Prompt writing is learnable in an afternoon. Judging output against a brand, a market and a commercial goal is not, and that is exactly the gap The Guardian’s report points at when industry bodies describe the technology as an intern rather than a replacement.
What this means for your design and AI budget
Most UK businesses we speak to are not choosing between a designer and AI software. They are deciding how to split the same budget: fewer hours on production, more on research, messaging and testing. We set that split out at the start of a project so you know which deliverables are AI-assisted, who reviews them and what the review actually checks.
We are also straightforward about pace. Rebuilding a design system with AI assistance still takes weeks of review cycles, and measurable movement in organic performance after a redesign usually appears at 4 to 6 months, not 4 to 6 weeks. Anyone promising faster than that is selling you the generation step and ignoring the approval step.
Questions about AI in design work
Will AI reduce what we pay for design?
It can reduce production hours, but review, strategy and brand governance still cost time. Most budgets shift rather than shrink.
Is AI-generated imagery safe to use commercially?
Licensing terms differ by tool, so check the commercial-use terms and any indemnity offered before publishing. If the terms are unclear, use it internally for concepts only.
Do we need to tell customers when design work is AI-assisted?
There is no blanket UK requirement to label AI-assisted design, but if your sector makes trust claims, being open about it is easier than being found out later.
Thinking about where AI fits in your design process
If you are working out which parts of your design and content process should be AI-assisted and which should stay entirely human, that is a conversation worth having before you buy tools rather than after. You can get in touch and we will talk through your current workflow and where the sensible starting point is.
Written by the London Marketing Company team.
