AI tools dramatically accelerate product creation, but they also introduce a new kind of UX debt: interfaces that look polished and syntactically correct, yet still fail users in predictable ways. When AI generates components without understanding user intent, we get "AI slop" at scale.

After reviewing a growing number of AI-assisted and vibe-coded products, recurring patterns started emerging: generic interaction models, broken state management from a user perspective, inaccessible automated choices, and complex interfaces optimized for speed rather than human understanding.

This talk explores how AI-generated design quietly creates usability problems at scale, why these issues are becoming harder to detect, and how teams can prevent “AI slop” before it reaches production.