The Age of UX Custodianship: How to Tidy Up the Mess AI Creates

Image source: Nielsen Norman Group · Anna Kaley, Raluca Budiu

AI summary focused on the core points of the original article.

Read original

1. UX debt accumulates after rapid production

AI has lowered the cost of making things, but it has not lowered the cost of judging usefulness and reliability.

The faster prototypes and features are built, the less time there is to confirm user needs and usability.
When verification is skipped, the result returns after launch as UX debt.
- Interfaces that work but are unnecessarily complex
- AI messaging and awkward content that obscure the core value
- Features that emphasize implementability and demo effect over actual need

2. The custodial pattern repeats in the same order

Outputs that appear finished call UX in too late, making cleanup after the fact the default task.

`Idea generation` → `AI-based rapid production` → `UX evaluation omitted` → `user confusion arises` → `UX team cleanup` repeats.
When a prototype already exists, conversations are likely to tilt from “Is it needed?” to “When should it launch?”
UX’s role also shifts from designing from a blank slate to selecting, simplifying, and fixing results that have already been made.

3. Three strategies for bringing the custodial role forward

The goal is not to slow generation speed, but to place user-perspective judgment at a faster and earlier stage.

Build shared judgment
- Use cost-benefit analysis and categorization to distinguish features that can be implemented from those that are actually needed.
- The outcome leaves room to keep, simplify, combine, defer, or remove, and preserves the reason as team knowledge.
Increase evaluation speed
- Choose user testing and research methods that fit the level of risk, and prepare evaluation templates and heuristic checklists.
- AI assists with research operations such as planning and analysis, but does not replace the interpretation of real users and researchers.
Reflect UX in the generation phase
- Provide the design system, approved interaction patterns, and content and accessibility standards as context for AI to reference.
- Feed issues found in post-review back into the generation rules so the same defects do not repeat.

4. From cleanup duty to custodian of user needs

When the lessons learned in cleanup are accumulated into the organization’s judgment system, UX can prevent the next round of confusion.

“Custodian” has two meanings: a person who cleans up messes and a person who takes care of and protects something valuable.
UX’s long-term role is not limited to continuously fixing AI outputs, but to managing the criteria for what to build and what to keep.
Whether users directly manipulate the product or leave it to an AI agent, the responsibility to keep human needs at the center remains unchanged.

XPLORED View

The quality of UX in the AI era depends less on one designer doing cleanup and more on how quickly an organization connects judgment, verification, and learning.

Good reading situations

When AI prototypes and features come out quickly, but UX reviews always lag behind
When discussing how to connect a design system or UX guidelines to the AI generation process

Takeaways

Shared judgment and rapid classification that separate “Can we build it?” from “Do users need it?”
An operational loop that feeds UX knowledge from reviews back into the context and rules of the next generation

Read original

This article is summarized and interpreted by XPLORED based on publicly available original sources and related materials. Rights to the original source and materials included in it belong to their respective rights holders. If a rights holder requests a correction or removal, we will review and respond promptly.

Get the week's signal by email, every Saturday.

A short read on the tools and articles added that week.