TL;DR
AI makes production faster than evaluation, forcing UX designers from creators into janitors — but they can turn it around.
AI lets teams build prototypes and ship code at absurd speed, yet deciding whether the result is actually usable has become the slowest part. The design often ships before anyone has even thought about whether it should exist.
Fast is not the same as useful
Designers used to start from a blank page. Now they pick through what AI dumped on them — five versions of a workflow in one afternoon, and someone has to keep, change or discard each one. The authors call this the custodial era of UX: design as cleaning up after AI, not creating from scratch.
The root problem is that generating has become cheaper than evaluating. Building a plausible-looking thing takes minutes; judging whether it helps real users takes understanding goals, cognitive load, mental models, accessibility — the slow, expensive work.
So teams pile up UX debt: the interface runs well enough for a demo, then users get confused, abandon it, flood support. Worse, AI accelerates old dysfunctions — teams that communicate badly or follow broken processes now ship their mistakes faster.
Don't wipe the floor yet — triage first
The authors' answer is not to make teams give up AI — that would never stick. Instead, three moves:
First, build shared judgment: when handed an AI-generated feature, resist polishing it. Run a cost-benefit analysis — what problem does it solve, who needs it, is it worth the complexity? Cut mercilessly, then explain your reasoning so the team learns to judge next time.
Second, make evaluation keep up: faster user testing, standard templates and checklists, AI to handle research logistics — but never substitute synthetic users, and keep humans responsible for conclusions.
Third, feed UX knowledge back into generation: put design principles, content standards and accessibility requirements into context files so the AI starts from a stronger base.
Parting thought: 'custodian' has another meaning — someone entrusted with something valuable. UX's job is not to guard the AI slop, but to guard real users' needs.
Curated from high-quality sources, with concise summaries and key takeaways.
