The 'AI in Design' Reports Keep Forgetting Who Actually Does Design
Every "State of Design" report published in 2026 tells you the same reassuring story: AI adoption is up, efficiency is up, designers are thriving. Figma's flagship survey put AI usage at 91% and reported that most respondents feel faster, not replaced. Read that number as good news for design and you've missed who was in the room when it was collected — because almost everyone in that room works in-house at a tech company on a product design team, and that population is having a completely different 2026 than the graphic designer doing brand identity work for local businesses, or the freelancer who used to make a living off logo commissions.
The gap between those two groups isn't a footnote. It's the actual story, and it's mostly not getting written.
The surveys measure the tier that isn't in trouble
"The AI in Design 2026 Report Forgot Graphic Designers," a critique published this year by House of AI, makes the specific charge directly: the dominant industry surveys — Figma's State of the Designer among them — draw their samples overwhelmingly from product and UX designers employed at software companies. That's a coherent, real population, and their experience of 2026 (AI tools embedded into Figma itself, faster prototyping, expanded scope) is genuinely well-documented and genuinely positive on average. It is also not representative of design as a profession, which includes a much larger and less-surveyed population doing commodity execution work: packaging, brand collateral, marketing graphics, social templates, the unglamorous bulk of what "graphic design" actually means as a job.
That second population is the one absorbing the disruptive side of AI adoption, and it's largely invisible in the data because it's harder to survey — freelancers and small-studio designers don't show up in a Figma in-product survey the way a product designer with a company Slack does. The report samples the group least likely to be hurt and calls the resulting number "the state of the designer."
What's actually happening to commodity work
Set the survey methodology aside and look at what's observably happening in the market for execution-only design — work where the brief is well-specified and the output is judged on competence rather than distinctiveness. That category of work is collapsing toward free, fast, and it's not subtle. A logo that used to be a $500–2,000 freelance commission is now a first draft a founder generates themselves in a tool with a monthly subscription cheaper than a single hour of a designer's time. Template-based social graphics, basic marketing one-pagers, straightforward packaging mockups — all of it is now something a non-designer can produce a passable version of without hiring anyone.
This isn't designers being "replaced" in the dramatic sense the discourse likes to reach for. It's the price of commodity execution converging on the marginal cost of generating it, which is close to zero. That's a predictable economic outcome whenever a previously-scarce skill becomes cheaply reproducible, and it's exactly what happened to commodity stock photography once generative image tools matured, and to basic copywriting once LLMs got competent at brand voice. Design is not special in this respect. It's just later.
Meanwhile, the premium end of the market is behaving exactly as predicted — the same 2026 data shows the opposite trajectory at the other end: taste-led, strategic design work — the kind that requires judgment about what a brand should mean, not just what it should look like — is commanding a premium, because AI tools are good at generating options and bad at knowing which option is actually right for a specific audience, moment, and business context. The designers doing that work aren't seeing prices compress. They're seeing demand concentrate toward them, because the AI tools every client now has access to make the difference between generic-competent and genuinely-considered more visible, not less.
A profession splitting instead of transforming
The framing that's missing from most 2026 coverage is that "AI's impact on design" isn't one trend — it's two trends running in opposite directions inside the same job title. Design is stratifying into a premium tier where AI is a leverage multiplier for people who already have taste and strategic judgment, and a commodity tier where AI is a direct substitute for the exact skills that tier's income depended on. Averaging those two trajectories together, the way an aggregate adoption survey does, produces a number that describes neither group accurately. It flattens a story about winners and losers into a story about a rising tide.
This matters beyond the accuracy of one report, because it shapes what advice gets handed to designers entering or currently working in the profession. "Learn to use AI tools, you'll be more productive" is true and useless for someone whose entire value proposition was fast, competent execution of straightforward briefs — the exact work AI tools now do for the client directly, without a designer in the loop at all. The advice that would actually help that person is harder to give and less flattering to hear: the execution-only lane is narrowing permanently, and the sustainable path runs through developing the judgment-heavy, context-heavy skills that sit at the other end of the split — brand strategy, art direction, the ability to say why one AI-generated option is right and four others are wrong for reasons a client couldn't articulate themselves.
What the industry data should actually track
If the design industry wants a survey that means something, it needs a sampling frame that includes the freelancers and small-studio designers who are living the commodity side of this split — not because their story is more dramatic, but because it's the leading indicator for where the rest of the profession is headed. Tech-company product designers are, structurally, the most insulated segment of the design labor market: internal headcount, embedded tooling, work that's inherently collaborative and judgment-heavy rather than transactional. Measuring "design" primarily through that segment is like measuring "how retail is doing" by only surveying employees at the one store chain immune to online competition.
The honest version of the 2026 design story has two charts, not one: a line going up for strategic, judgment-led work, and a line going down for commodity execution, with a widening gap between them that the industry's own flagship reports have mostly declined to draw. Ninety-one percent AI adoption is a real number. It's just not the number that tells you whether design, as a way to make a living, is getting better or worse — because the answer to that question depends entirely on which half of the split you're standing in.