[Help needed] We've opened the gates of hell
HCI Today summarized the key points
- •This article addresses the conflicts that arise after an LLM-friendly design system is created, as PMs begin using it directly.
- •The author explains that the design system was reorganized to be LLM-friendly, with definitions, goals, variables, and principles clearly structured.
- •However, the author says that PMs skip the Validation Phase, create their own designs, and miss dependencies on response-based screens and different flows—leading to confusion.
- •The comments argue that this is a result of organizational culture and the pursuit of efficiency, pointing out that the value of UX (user experience) and design is being underestimated.
- •Overall, the takeaway is that we shouldn’t stop at protecting the design itself—we also need to make collaboration rules and validation procedures clearer about when and how the system should be used.
This summary was generated by an AI editor based on HCI expert perspectives.
Why Read This from an HCI Perspective
This article is a case study showing how, as design systems are organized to be more LLM-friendly, it becomes clearer who in an organization has design authority—and through what paths. For HCI/UX practitioners, it demonstrates that ‘tool efficiency’ does not automatically translate into ‘the appropriateness of decision-making.’ It’s especially meaningful because it lets you read the costs that emerge when the validation stage, contextual understanding, and responsibility boundaries weaken.
CIT's Commentary
From a CIT perspective, this situation is less about personal conflict and more about interaction design and organizational governance. As you make the design system easier for LLMs to use, it effectively becomes an internal self-service tool. The problem is that when accessibility improves but validation mechanisms and definitions of usage scope are not designed together, the more outputs you produce, the more collaboration friction and quality variance you tend to get. So the key is not to block PMs, but to clearly define—at the process and interface levels—things like ‘under what conditions an automated draft is valid’ and ‘when a design review is mandatory.’ This is also about preserving the dignity of design, but more broadly it resembles HCI-oriented safety design that reduces product risk.
Questions to Consider While Reading
- Q.In an LLM-friendly design system that allows self-service, how can we design the workflow so that the validation stage is naturally passed through without having to force it?
- Q.If we want to use drafts generated by PMs while still maintaining the value of design review and the boundaries of authority, how should we restructure role definitions within the organization?
- Q.Between improving the accessibility of a design system and preventing quality degradation, what metrics can HCI researchers use to evaluate the balance?
This commentary was generated by an AI editor based on HCI expert perspectives.
Please refer to the original for accurate details.
Subscribe to Newsletter
Get the weekly HCI highlights delivered to your inbox every Friday.