The design world has a new vocabulary problem. AI-generated visuals are converging — same typefaces, same gradient palettes, same three-column grid layouts — and commentators are reaching for “taste” to explain what’s missing. AI doesn’t have taste, they say. Taste is what separates human designers from the machine.
They’re noticing something real. The explanation is off.
What You’ll Learn
- Why AI design is converging on a recognizable aesthetic, and what’s actually causing it
- Why “taste” is the wrong word for what’s missing
- What discernment means — and why it’s the more precise, more accountable concept
- Why this matters for brands specifically, and where the real risk sits
- Why critics reach for “taste” instead of the sharper answer
What Is AI Design Slop — and What’s Actually Causing It?
AI design tools produce outputs that are locally correct but globally incoherent. The type pairing is fine. The color palette is balanced. The layout is clean. But set it next to ten other AI-generated brands and you’ll find the same Inter font, the same purple gradients, the same three boxes with icons in a row.
This has a name: distributional convergence. LLMs are trained on the aggregate of everything that’s been made. They gravitate toward the statistical median — the most common choices across every Tailwind tutorial, design portfolio, and startup landing page scraped from the web. The result is aesthetics that feel technically sound and culturally familiar without meaning anything specific. It is the same flattening earlier design movements rose against: when mechanical reproduction strips intention out of visual culture, the response—from Art Nouveau in the 1890s onward—has been to put it back.
Claude Design, Anthropic’s recently launched design tool, is the latest flashpoint in this conversation. Reviewers noted it could produce a launch-ready brand in thirty minutes. Developers noted it kept arriving at the same visual shorthand. Both things are true. That’s the problem — not that the outputs fail on their own terms, but that the terms are too narrow.
The most reliable rule of thumb here: AI makes locally correct decisions very well. What it doesn’t make are decisions that mean something across a whole system.
Why “Taste” Is the Wrong Word for What’s Missing
When critics say AI lacks taste, they’re describing a real gap with the wrong tool.
Taste implies preference. It’s subjective. It’s personal. It suggests the missing ingredient is some kind of innate aesthetic intuition — a feeling for what looks right — that humans have and machines don’t.
That framing does two things, both of them unhelpful. First, it individualizes the problem. If the gap is about personal flair, then the solution is to find humans with better flair, and the conversation stalls in vibes. Second, it makes the problem impossible to interrogate. You can’t test taste. You can’t build a system around it. You can challenge it with logic and it shrugs: “I just feel like it works.”
Marc Mueller, a UX consultant who wrote about this in April 2026, noted the risk directly: reducing quality to taste “puts us back to discussions over whether a button should be blue because an important stakeholder simply does not like blue.” He called for discernment instead — judgment that’s contextual, not aesthetic.
The critics reaching for “taste” are doing the same thing, just at a higher level of abstraction. They can feel the problem. They haven’t named it yet.
What Discernment Actually Means
Discernment is the ability to distinguish — not just what looks good, but what belongs.
It’s the capacity to evaluate a decision against a system. Not: does this typeface look good? But: does this typeface reinforce what this brand is trying to be understood as? Not: is this color palette appealing? But: does this palette signal the same things the language signals?
Where taste evaluates the part, discernment evaluates the whole. It asks whether a decision is appropriate to its context, not merely attractive. It asks whether a choice reinforces meaning or introduces ambiguity. It asks whether this output coheres with what came before it — and what will come after.
That last question is where the real separation lives. Discernment requires understanding how decisions accumulate over time. A single output doesn’t tell you much. A hundred outputs across two years tells you everything. Discernment is what governs whether those outputs build toward something specific or scatter into pleasant noise.
As a general rule: taste selects what looks good. Discernment selects what means something.
AI can approximate taste because it’s pattern-based. It can’t approximate discernment because discernment requires context, constraint, and continuity — none of which exist inside a single generation.
Why This Is a Brand Problem, Not Just a Design Problem
A brand doesn’t hold together over time because of taste. It holds together because of repeated, intentional decisions that reinforce a consistent meaning across every surface it touches.
That’s discernment at work. It answers: Does this campaign headline align with what the visual system already says? Does this product launch build on the identity that’s been established, or interrupt it? Are we clarifying what we mean, or adding another signal that points somewhere slightly different?
Taste might produce a logo that looks elegant in isolation. Discernment evaluates whether that logo belongs — whether it fits the system, whether it amplifies the right associations, whether it makes the next design decision easier or harder.
What design slop is actually producing isn’t just visual similarity. It’s systemic incoherence dressed up as polish. Brands that build on AI-averaged aesthetics don’t have a style problem. They have a coherence problem. The signals don’t reinforce anything specific. The brand looks like a brand. It doesn’t function like one.
The real risk isn’t that things look similar. It’s that signals stop reinforcing anything specific — and when that happens, people can’t form a stable understanding of what something is. You don’t just lose distinctiveness. You lose the conditions under which trust can accumulate. That’s a systems failure, not a stylistic one.
Why Critics Keep Reaching for “Taste”
The reach for “taste” isn’t accidental. It’s a defensive move.
People are noticing — correctly — that something is eroding. Distinctiveness. Authorship. Intentionality. The sense that someone made a decision for a reason. They want to name what’s being lost, and “taste” arrives quickly because it sounds like it fits.
But it’s a placeholder. It stands in for: “There’s a difference here, and I can’t fully explain it.”
Then a second-order problem appears. JargonSlop trying to explain design slop. The conversation fills with talk of “creative intuition,” “the human touch,” and “aesthetic sensibility” — real ideas all, but deployed at a level of abstraction that prevents anyone from doing anything useful with them. No criteria. No test. No system to build toward.
This is what makes “discernment” the more honest and more useful word. It’s accountable. You can ask: discernment of what, exactly? And the answer has to be specific — discernment of whether this decision reinforces meaning across a system, of whether this signal agrees with the others, of whether this output serves the brand’s job, which is to be understood clearly and consistently over time.
That gives you something to work with.
What This Changes
The design slop conversation will keep going. The tools will improve. The aesthetic tells will fade. And the question will remain: does this output mean something, or does it just look like it does?
Taste won’t answer that. You can’t interrogate it, test it, or build a system around it. You can’t use it to evaluate whether a brand decision coheres with the fifty decisions that came before it, or the fifty that will follow.
Discernment can. It’s structural. It’s accountable. It names the thing that actually separates a brand that accumulates meaning from one that produces credible-looking noise.
The word matters. Use the right one.
