Kathryn Grayson Nanz organizes AI user experience around trust, clarity, control, transparency and meaningful benefits. She argues that familiar chat patterns are useful but not universal, and that an AI feature should solve a real user problem rather than exist only because the technology is available.
Kathryn Grayson Nanz recommends making sources, planned actions, costs and permissions understandable. Users should be able to stop operations, review consequential actions, distinguish generated content and understand what information is remembered. Explanations should communicate useful rationale without pretending to expose a model's internal thought process.
Kathryn Grayson Nanz also discusses undo, version history, visible changes and appropriate granularity of control. Separating read access from write authority and providing ways to revoke permissions can reduce surprises. Her examples are design principles and patterns, not a claim that a polished interface alone guarantees reliable or safe AI behavior.
Watch on YouTube
