UX Collective: Why AI is exposing design's craft crisis

The argument Dolphia makes in this piece is blunt: AI didn’t create a crisis in design — it exposed one that has been building since the early 2010s. The root cause, in the author’s reading, is an industry-wide decision to position design work as separate from technical implementation.

Early bootcamp culture explicitly framed this as a feature. Designers were told they didn’t need to write code, understand data models, or follow the mechanics of API design. That separation looked fine while interfaces were deterministic — when every screen state could be specified in advance and handed off to an engineer. The problem emerged when most product decisions started involving trade-offs that only make sense if you know what’s technically feasible.

The statistics Dolphia marshals are striking. Entry-level design hiring at major tech companies fell to just 7% of all hires in recent years — a drop of around 50% compared to 2019. Roughly 62% of developers report spending significant time reworking designs because of communication gaps. Figma’s own demo site was found to have over 210 WCAG accessibility violations. All of this preceded the current wave of AI tools by years.

AI sharpened the problem because the tools are not actually easier to use than the craft they replace — they require more domain knowledge, not less. Using a generative design tool well demands prompt engineering skills, the ability to recognize when a model is producing something technically impossible, and enough fluency with accessibility or implementation constraints to catch errors before they compound. Designers who lacked that fluency with traditional tools now face a much shorter feedback loop to visible failure.

What the article recommends is not that designers should become engineers. Dolphia argues for what she calls strategic literacy: understanding how data models constrain interface behavior, what API latency means for interaction design, what technical debt implies for future design iterations. Designers at companies like Vercel and GitHub who developed that fluency have held influence through the current disruption; those who didn’t have lost ground.

The piece draws a useful distinction between knowing how to code and knowing why technical decisions look the way they do. The former is a skill; the latter is judgment that shapes whether a designer can participate in the room where decisions get made.

This article is useful for mid-career designers who want a sharper account of why senior positions have become harder to reach, and for team leads trying to understand why some design hires contribute to product direction while others don’t. It’s equally relevant for anyone building design education programs or onboarding tracks.