Taste.
The case is made with conviction across conferences, podcasts, LinkedIn threads, and design leadership forums: AI can generate but it cannot discern. It can produce volume but it cannot exercise judgment. It can simulate the output of design without understanding why one decision is better than another. Designers are safe — not because AI is weak, but because taste is irreducibly human.
I've heard this argument dozens of times in the last two years. And I agree with part of it.
Taste is real. Earned taste — built from years of putting work into the world, watching it succeed and fail in the market, sitting in rooms where good design got killed by organizational politics and mediocre design somehow made it through anyway, learning through repetition and failure what actually works versus what looks like it should work — that kind of judgment is genuinely hard to replicate. The machine pattern matches. It doesn't have scar tissue. It hasn't been humbled by a product that failed in market or had to defend a design decision to a skeptical CFO at 9am on a Monday. That experience-based discernment is real and it matters.
But taste without execution speed is just opinion.
And the execution gap — the distance between a design concept and a realized artifact — is collapsing faster than the design community is willing to acknowledge. That collapse is what changes everything.
Before pushing back, it's worth being precise about what the taste argument is actually defending.
The designers making this case are not wrong that judgment matters. They are not wrong that aesthetics without discernment produces work that is generic, derivative, and ultimately forgettable. They are not wrong that the ability to recognize quality — to feel when something is off even before you can articulate why — is a real skill developed over time through deep practice.
They are also not wrong that AI, as it currently exists, does not have this. A model trained on the internet's design output will regress to the mean of that output. It will produce competent work. It will rarely produce surprising, category-defining, genuinely original work without significant human direction. The ceiling on AI-generated design, unguided, is mediocrity.
Taste is the thing that pushes above that ceiling. That's a legitimate and important point.
But the argument tends to stop there. And stopping there is where it becomes a comfort blanket rather than a strategy.
The taste argument assumes that taste is currently being used at full capacity in most design organizations. It isn't.
A significant portion of what designers do every day is not taste-driven. It is execution-driven. It is the production of variations, the generation of assets, the building of component iterations, the documentation of specifications, the preparation of presentations, the creation of deliverables that demonstrate what has already been decided. Taste informs this work, but it does not primarily drive it. It is craft production.
AI is coming for that layer first. Not the judgment layer. The production layer.
And when the production layer is handled — when the first hundred variations are generated before the designer sits down in the morning, when the component library updates itself, when the spec documentation writes itself from the design file — what remains for the designer is a fundamentally different job.
The remaining job looks like this: constrain the machine before it runs, curate its output when it returns, and present the result to the humans who need to act on it. Curation-orchestration-curation-presentation. That's the new shape of the work.
Which means taste shifts from being a creative skill to being an editorial one.
This is not a trivial distinction. Creating and editing are related but different cognitive modes. The designer who is excellent at making things from scratch — who thrives on the blank canvas, who does their best work in the generative phase — may struggle in a workflow where the canvas is never blank, where the machine has already filled it with plausible options, and where the job is to navigate, filter, and improve rather than originate.
The designers who are honest about this are already adapting. They are learning to work with and through AI tools not as novelties but as the primary medium of their practice. They are developing the editorial judgment that curation at scale requires. They are figuring out how to be excellent at the new shape of the job, not just excellent at the old one.
The designers hiding behind the taste argument are, in many cases, not doing this. They are waiting for the threat to resolve itself, banking on the irreplaceability of their judgment while the execution layer that surrounds and supports that judgment gets automated away underneath them.
The shift from creative to editorial is worth dwelling on because it has significant implications for how designers need to develop, how design teams need to be structured, and how design leaders need to hire.
Editing is not easier than creating. In some ways it is harder. The editor has to hold a vision clearly enough to recognize when the generated output serves it and when it doesn't — often across hundreds of options, under time pressure, without the luxury of starting fresh. The editor has to know not just what looks right but why, and has to be able to communicate that judgment quickly and precisely enough to direct the next iteration.
Great editors are rarer than great creators. Not because editing is a higher art, but because it requires a different and complementary set of skills that most people haven't deliberately developed.
For designers, the editorial turn means developing skills that design education and design culture have historically undervalued: the ability to make fast, confident decisions under volume and pressure; the ability to articulate judgment in terms that can direct a system rather than just inform a human collaborator; the ability to distinguish between options that are merely different and options that are genuinely better; and the ability to do all of this at a pace that matches the speed at which AI can generate.
This is a real skill gap. And it is not going to close on its own.
The designers who will thrive in the AI-augmented environment are not the ones with the best taste in the abstract. They are the ones with taste that is precise, communicable, and fast.
Precise: They know not just that something is wrong but specifically what is wrong and what would make it right. Vague aesthetic discomfort is not actionable in an AI workflow. "This feels off" doesn't direct a model. "The hierarchy is competing with itself at the tertiary level and the type scale needs to compress by two steps" does.
Communicable: They can translate their judgment into constraints, prompts, and direction that shapes AI output before it is generated, not just critiques it after. The ability to front-load taste — to encode it into the inputs rather than simply apply it to the outputs — is a new form of design skill that matters enormously and is currently rare.
Fast: The volume of AI-generated output is going to continue to increase. The designers who add value are the ones who can process that volume at speed, making good decisions quickly rather than excellent decisions slowly. Deliberation is valuable. Paralysis in the face of abundance is not.
Earned taste — built from experience, failure, and genuine engagement with the hard problems of design — is the foundation of all three of these qualities. But it has to be activated in the new context. It doesn't transfer automatically.
The taste argument is not wrong. It is incomplete.
Taste matters. It will continue to matter. The designers with deep, earned, experiential judgment will have real advantages in the AI-augmented environment — advantages that designers who have relied primarily on craft production will not have.
But taste alone is not a career strategy. It is not an organizational strategy. And it is certainly not a reason to slow down the work of repositioning for what's coming.
The design community is at its best when it is honest about hard things. The hard thing right now is that the nature of design work is changing faster than most designers are acknowledging, and the comfort of the taste argument is making it easier to avoid that reckoning than to engage with it.
The next 24 months will make this undeniable. The designers and design leaders who are already moving will look prescient. The ones who waited for the threat to become obvious will find that by the time it was, the window to reposition had already closed.
Taste got us here. It won't be enough to get us to what's next.
Part 2: The shape of the valuable designer is changing faster than hiring practices are adapting.