Can generative UI serve users who cannot specify interfaces?

Kate Moran grounds the case for generative UI (genUI) in her own user-research experience: "I think back to the hundreds of user-research sessions I've conducted and observed where I've seen firsthand how bad most people are at identifying and articulating what they want from a digital product or service." On that basis she argues vibe coding "assumes a user who can think in terms of apps, dashboards, and interaction patterns," which she calls "a meaningful prerequisite, and not one most people meet." GenUI, by contrast, "serves the rest": "'Help me plan my trip' is a prompt anyone might type," and if the system answers with "a custom itinerary tool rather than a paragraph of text," someone who would never think to vibe code a trip planner still gets an interface "designed for their situation."

The mechanism Moran gives is that genUI's promise is not better code generation but the AI acting as designer: "not that AI becomes a better code generator, but that it becomes a designer. One that makes the interface decision so the user doesn't have to know they needed one." She extends this into a market argument against the "build your own everything" reading of vibe coding — the belief that "AI will let everyone build their own custom software for everything they need." She contends that reading overlooks the "institutional knowledge" dedicated software makers accumulate "from years of user research, iterative refinement, and observing how the tool performs across thousands of different organizational contexts," depth she says "LLMs don't inherently have... baked in."

This supplies a candidate explanation for the preference numbers in Do generated interfaces outperform text-based chat for most tasks? and Do full web pages beat markdown chat for LLM responses?: both measure that users prefer generated interfaces to chat text, and Moran's articulation-gap argument offers a reason that preference might hold — not that the interfaces look better, but that most users could never have specified them in the first place. The argument also depends on Does who initiates a UI element matter for accountability?, since "serves the rest" only follows once genUI is defined as the AI's own decision to build rather than the user's request.

Moran's claim rests on her recollection of "hundreds of user-research sessions" rather than on any reported study, count, or citation given in the excerpt — it is professional testimony, not measured data, and the excerpt gives no figures on how many users actually fail to specify interfaces or how reliably genUI tools succeed at reading "the moment" as she hopes. The implication, held to that evidentiary weight, is that genUI's value proposition is plausible for the stated reason but remains an argued position rather than a demonstrated one.