§ 01 · What actually happened

“AI didn’t replace the graphic designer. It replaced about thirty percent of what a designer used to bill for.”

Margot, there is a familiar story circulating among designers right now — that AI has replaced them overnight, that the profession is dying, that twenty-five years of experience now counts for nothing. How accurate is that picture?

Partially accurate. The dramatic version of the story circulates because it feels true to anyone who has been freelancing in the last eighteen months. The honest version is more complicated. What AI has done is replace the lower-value end of design work — the quick logo, the social tile, the simple menu update, the email banner, the trade-show poster mockup — with something the client can produce in twenty minutes for the price of a cup of coffee. That work used to pay a freelancer five hundred to fifteen hundred pounds a month if you stacked enough small clients. It doesn’t any more. That is the part of the story that is true.

And the part that is more complicated?

The higher-value end of design work has actually become more, not less, valuable. Brand identity that needs to be coherent across thirty touchpoints, editorial systems that have to hold together across years, print production for fine printing on unusual stocks, packaging that needs to survive being scanned at a till and photographed for Instagram — that work is the same work as it was five years ago, and it pays well. What has happened is that the middle has hollowed out. The bottom is now AI plus the client themselves. The top is still designers. There is no middle.

For a designer who has spent their career in that middle, what does this actually look like in 2026?

It looks like one of three things. They either retrained towards the top — got serious about brand strategy, art direction, complex production — or they retrained sideways into roles where design is one skill among several, like marketing manager or content lead. Or they left the profession. I know a fair number of people who took all three paths. The ones who took the third path are the ones writing the catastrophe pieces on LinkedIn. The other two are doing fine, or at least no worse than they were doing in 2021.

§ 02 · The expanding job description

“The job advert used to say ‘graphic designer.’ Now it asks for nine people.”

There’s a recurring complaint I see online — that small companies expect a graphic designer to also be a UI/UX designer, a video editor, an animator, a front-end developer, a copywriter, and a social media manager. Is this real or just venting?

It is entirely real. I have seen job postings in the last six months asking for proficiency in Adobe Creative Suite, Figma, After Effects, Premiere Pro, Canva, HTML and CSS, a working knowledge of WordPress, basic copywriting skills, social media management experience, and the ability to operate a TikTok strategy. The salary band is sometimes thirty thousand pounds. That is one person being asked to do the work that used to be done by five specialists, for the wage of a junior copywriter. It is not a sustainable expectation, but it is the expectation that employers have set.

Where did this come from?

Three things compounded. First, the tools genuinely got better. Figma made UI design accessible to print designers; Premiere Pro became smooth enough to be picked up alongside other Adobe products; Canva absorbed enough animation features that motion graphics felt less like a specialism. So employers reasoned that one well-rounded person should be able to handle it all. Second, budgets were tight, especially for small businesses and creative agencies through 2022 and 2023. Hiring five specialists was no longer feasible for most teams. So they hired one generalist and asked them to do the work of five. And third — the part nobody talks about openly — designers themselves contributed to the dynamic by accepting it. The market punished the ones who said no. The ones who said yes kept their jobs. And so the new normal got set.

And now AI sits on top of all of that.

Now AI sits on top of all of that. Which has two effects, both of them complicated. The first is that AI absorbs some of the work the generalist designer was already drowning in — quick social tiles, video subtitles, image upscaling, background removal, simple animation, draft copy. That should be good news for the designer. The second effect is that employers see this and assume the designer can therefore do even more, because they now have AI helping them. So the job description expands again. The designer who used to do five jobs is now doing seven, with AI doing the work of perhaps two of those seven invisibly in the background. The total workload, measured in things the designer is held accountable for, has gone up.

§ 03 · The logo and the menu

“The freelance logo at three hundred pounds is the work most directly replaced. And it’s the work that used to fund freelance careers.”

For the readers of this publication specifically — small business owners who design their own menus, certificates, invitations — what does the AI shift actually look like in practice?

It looks like the small café owner I’ve mentioned before, who designed her own menus on Canva on a Sunday afternoon. She had access to AI-generated background images for the headers, AI-generated icons for the menu categories, and AI-generated colour-palette suggestions that matched her interior. None of that existed in usable form in 2020. All of it existed by 2024. By 2026, it’s built into Canva natively. She made a menu that would have cost her four hundred to seven hundred pounds at a small design studio five years ago, for the cost of her Canva Pro subscription. The work was good. The freelance designer who used to do that work for her, however, has lost that client and another twenty like her.

And logos specifically?

Logos are the most directly affected. A small business that needed a logo in 2020 had three paths. Pay a designer eight hundred to three thousand pounds for a proper identity. Use a marketplace like 99designs and crowdsource it for three hundred. Or use a logo-builder tool and produce something acceptable for fifty. In 2026, the fourth path has effectively eaten the second and third paths and is starting to eat the bottom of the first. AI generates a logo concept in thirty seconds. Tools like LogoAI, Looka, Brandmark, and increasingly Canva’s own brand kit feature, will iterate on it until the small business owner has something they’re satisfied with. The total cost is the monthly subscription. The freelance market for small-business logos has, frankly, collapsed in the last eighteen months.

Are the AI logos any good?

This is the question that designers and clients answer differently. Designers will tell you, accurately, that AI logos lack typographic discipline, that the symbol marks are generic, that the colour systems don’t scale across applications, and that an AI logo will look dated within eighteen months. All of that is true. Clients will tell you, also accurately, that their AI logo looks fine on the menu, on the business card, on the Instagram bio, and on the awning. And that they spent twenty pounds on it rather than eighteen hundred. The market is making its preference clear, and the market is not in the designer’s favour.

§ 04 · The cost backside

“A designer running ten client projects on GPT-image-2 at retail rates is now spending three hundred pounds a month on what used to be free Photoshop time.”

There’s a part of this story that isn’t often told — the cost of the AI tools themselves. Walk us through what that looks like for a working freelance designer in 2026.

It’s a real and growing expense, and it gets very little attention compared to the “AI is free” narrative. A freelance designer running ten to fifteen small client projects per month, generating perhaps fifty images per project across drafts and final outputs, is now spending somewhere between one hundred and four hundred pounds a month on AI image generation alone — before the cost of Midjourney, of Canva Pro, of Adobe Creative Cloud, of any video tool, or of any subscription stack. The cost is variable. It depends on which model they’re using, how often they iterate, and whether they generate at high resolution. But it adds up faster than people expect.

What are the main models people are running?

Midjourney for general illustration and stylised imagery; DALL-E and its successor models for things that need to be more literal; Stable Diffusion for designers who want local control; and increasingly OpenAI’s GPT-image-2 for designers working inside ChatGPT or via the API, especially for product mockups and lifestyle imagery. The pricing varies considerably. GPT-image-2 at the retail API rate is, for example, ten cents per high-quality image. That sounds nothing. Multiplied by five hundred generations a month, it’s fifty pounds in raw inference. Multiplied by an agency-scale workflow with three thousand generations a month, it’s three hundred pounds. That money used to be entirely in the designer’s margin.

And there’s a response to this happening on the procurement side?

There is, and it’s the part of the conversation that working designers actually talk about — not whether AI will replace them, but how to keep their margin intact now that AI consumes a meaningful chunk of it. Most freelance designers I know are spending a bit of time figuring out how to buy cheap gpt-image-2 credits and Midjourney credits in bulk, rather than paying the retail subscription rate. The marketplaces that have emerged in the last twelve months aggregate enterprise-bought credits and resell them to small users at margins of fifteen to forty percent below list price. For a working designer that’s the difference between making a 20 percent margin on a client project and a 28 percent margin. Multiplied across a year of project work, it’s the difference between a frustrating year and a viable one.

Is this becoming standard practice?

Among the designers paying attention, yes. Among the designers still on the fence about whether to engage with AI at all, no. The first group is treating their AI costs the way agencies have always treated their software stack — as a recurring expense that needs to be procured intelligently. The second group is still paying retail rates because they haven’t had a serious conversation with themselves about what the tools actually cost. The gap between those two groups, measured in annual margin, is now somewhere between five and ten thousand pounds. It’s the difference between feeling like you’re running a business and feeling like you’re working for the tool providers.

§ 05 · Who wins, who loses

“The interesting thing is not that AI killed graphic design. It’s that AI redistributed who gets to do it.”

If we zoom out from the individual freelance experience and look at the profession as a whole, who has gained from this shift?

Three groups, broadly. First, small business owners and the self-employed who previously couldn’t afford a designer at all. Their access to acceptable design has expanded dramatically. The local hairdresser who couldn’t justify a four-hundred-pound logo can now have one for nothing. The community theatre that used to print posters from a Word template now produces something genuinely attractive. The neighbourhood coffee shop, the wedding photographer’s own marketing, the small accountancy practice — all of them now look better than they did in 2018, and at no incremental cost. That’s a real welfare gain, even if it comes at the cost of someone else’s livelihood.

Second group?

Senior designers who specialised hard. Brand strategists, art directors at established agencies, designers running their own studios with reputation and clientele built over fifteen or twenty years. These people are doing well. Their work is now harder to commodify, more clearly differentiated from the AI alternative, and more valuable than ever to clients who care about brand-level outcomes rather than asset-level outputs. The number of these designers is small relative to the profession, but their hourly rates are up, their backlog is full, and their work is more interesting than it was five years ago.

And the third?

The designers who pivoted into adjacent roles. Marketing managers, content leads, product managers at small software companies, in-house creative directors. These are roles where design is one of several core skills, and they reward generalists with strong visual judgement. Many of the displaced freelancers landed here, and most of them are reporting that they’re happier than they were as freelancers — more stable income, less feast-or-famine billing, more interesting cross-functional work. It’s a real path that doesn’t get enough credit in the “designers are dying” narrative.

Who has lost?

The middle. Specifically: mid-career freelance generalists working in the small-business segment. Ages roughly thirty-five to fifty-five, ten to twenty years of experience, billing the bulk of their work at three to fifteen hundred pounds per project. Their work has been most directly replaced. Their costs have gone up because of the AI subscriptions. They are too senior to compete with junior generalists on price and not specialised enough to compete with senior designers on craft. That group is genuinely struggling. The story of design in 2026 is mostly about what’s happening to that group, and the broader narrative gets it mostly wrong by attributing the cause to AI alone rather than to a market structure that was already shifting against them.

§ 06 · What to do with all this

“Stop fighting the tools. Start procuring them properly.”

If a small business owner is reading this and trying to design their own menu or invitation, what should they take away?

That you have access to better tools than any small business has ever had, and that the tools are improving faster than you can read about them. Use them. The cafe owner with the menu was right. The wedding planner customising her own template was right. The school deputy head designing certificates on a sixty-pound budget is right. You don’t need to feel guilty about not hiring a designer for work that AI can handle adequately. You do need to be honest about when the work has outgrown the tools — brand identity that has to last a decade, packaging that has to survive at retail, anything print-production complex — and at that point hire a designer worth their fee.

And to working designers?

Two things. Procurement and positioning. On procurement: stop paying retail for your AI credits. If you are running a freelance practice and not buying through a marketplace, you are leaving fifteen to forty percent of your AI margin on the table for no reason. On positioning: identify what part of your work AI is actually replacing, and stop selling that part. Pivot upward into the work that genuinely requires you. The designers I know who are thriving in 2026 are doing brand strategy, complex production, art direction, and editorial systems — not the work AI can now produce in thirty seconds. That pivot is uncomfortable but it’s also the only one that’s working.

Last question. Will AI replace designers entirely in ten years?

No. AI will continue to absorb the lower end of the work, and there will be displaced freelancers throughout, which is genuinely sad. But the highest-value design work — the work that requires sustained taste, real judgement about culture and audience, and the ability to hold complex systems coherent over years — will still be done by humans, and the humans doing it will be paid more, not less. The profession will be smaller, more specialised, and harder to enter, but it will exist. The same way watchmaking still exists. The same way bespoke tailoring still exists. The same way fine printers still exist. The work moves up the value chain. It doesn’t disappear.

Margot, thank you.

Thank you. Write in with your own experiences — I read everything, and the stories from working designers are the most useful primary source we have on what this transition actually looks like from the inside.