The Design Debt Crisis:
When AI Erases Your Taste

The taste you didn’t know you had

Design taste isn’t about being artistic. It’s about having standards — an internal reference point for what “good” feels like. Engineers develop this over years: from reading codebase after codebase, from seeing which abstractions age well and which become liabilities, from debugging interfaces where the bug was in the design, not the code.

Senior engineers tend to have better taste than juniors. Not because they studied design — because they’ve accumulated enough examples of what works to develop instincts. They can look at a new pattern and feel whether it’s going to create problems in six months. They can read a PR and sense that something about the naming feels wrong before they can articulate why. This is taste working — and it’s quietly one of the most valuable things a senior engineer possesses.

Now consider what happens when AI starts handling the implementation. You describe a component. AI generates it in 30 seconds. You scan it, it looks fine, you ship it. Where did the taste get exercised? It didn’t. You made no judgment call about flex alignment or color weight or spacing rhythm. Your taste wasn’t consulted — it wasn’t needed. And if it’s not needed for months at a time, it doesn’t just rest. It atrophies.

What AI takes from your design sense

Design taste operates through continuous micro-judgments. Every time you look at an interface and make a call — even a small one, like “that padding feels right” or “this hierarchy is confusing” — you’re maintaining the neural pathways that constitute your design sense. AI tooling removes these micro-judgments from your workday.

The degradation is subtle but measurable. Engineers who used to notice bad UX in their own work report that they stopped noticing. They review AI-generated designs and think “looks fine” — but can’t articulate why it might feel heavy, or clinical, or generic. The internal standard hasn’t disappeared; it’s just gone quiet, like a voice you’re not listening to.

The pattern is invisible because the loss is felt as平静 — you don’t notice you’re losing something when you’re not trying to use it. You just become someone who accepts things more readily. Which sounds fine until you need to make a design decision and find you have no basis for choosing between options.

Five symptoms of accumulating design debt

Design debt is different from technical debt in one crucial way: you don’t see it accumulating. Technical debt creates compile errors, slow tests, merge conflicts. Design debt is silent. Here are the signals it’s building.

Why this matters more than it used to

Design taste used to be a nice-to-have for engineers who worked near the UI layer. In an AI-accelerated world, it’s becoming the primary differentiator of senior engineering quality. Here’s why.

When AI handles implementation, the scarce resource becomes direction. Someone has to decide what to build, whether this approach feels right, whether this design will age well or create debt. That’s taste. The engineer who knows what good looks like — who can look at an AI-generated output and immediately see three problems and two directions to explore — that engineer is an AI multiplier. They get more value out of AI tools because they know what to ask for and what to reject.

The engineer whose taste has atrophied is an AI amplifier in the other direction: they accept everything AI produces, they can’t distinguish good from adequate, and their output converges toward generic. They’re faster than they used to be, but the ceiling has dropped. They’ve optimized for velocity and sacrificed the standard.

This is particularly acute for staff and principal engineers who are increasingly making architectural and product decisions where design judgment — the ability to evaluate quality across dimensions — is as important as technical knowledge. When you’re deciding whether an abstraction is worth the complexity it introduces, you’re using taste as much as logic. If that faculty has gone quiet, you’re making worse decisions without knowing it.

How far has your taste drifted?

The taste rebuilding protocol

Rebuilding design taste is a skill maintenance problem, not a learning problem. You built taste once; you can rebuild it. Here’s a practical protocol, built on deliberate practice principles from Bjork’s desirable difficulties framework.

The key insight from skill science: the struggle is the learning. When you find design decisions difficult — when you have to think hard about whether this spacing or that one feels right — you’re rebuilding the pathways that AI made dormant. Passive exposure won’t do it. You have to actively exercise the judgment.

The design debt compound effect

Design debt compounds in the same way that skill atrophy does: slowly, then all at once. For the first few months, you don’t notice anything. Then you realize you’ve lost something you can’t quite name, and you can’t rebuild it in a week.

The engineers who will thrive in an AI-augmented world aren’t the ones who use AI most skillfully — they’re the ones who maintain the judgment to know what AI should be used for. Design taste is judgment. It evaluates quality, detects problems before they’re coded, sets the standard that AI can then be pointed toward. Without it, you’re not directing AI — you’re following it.

The engineers with the strongest design taste in five years will be the ones who maintained it deliberately — who kept exercising the judgment even when AI made it unnecessary in the short term. That’s the tradeoff: short-term velocity for long-term standard. Most engineers are making that tradeoff without realizing it. You’re now aware. What you do with that awareness is the actual decision.

Continue Reading