Practice · Fast-moving · Intermediate

AI User Experience

Designing interfaces for systems that are probabilistic, occasionally wrong and sometimes slow.

What AI User Experience is

Good AI UX sets expectations, shows sources, makes correction easy, streams progress, and never presents an uncertain output as settled fact.

How it works

Patterns include drafts rather than final answers, inline citations, confidence signalling, undo and edit affordances, visible progress for long agent runs, and explicit consent before consequential actions.

Why it matters

Most user distrust of AI features comes from interface decisions rather than model quality — unexplained answers, no way to correct, no indication of uncertainty.

Common uses

→Assistant interfaces
→Agent approval flows
→Search and answer interfaces

Strengths

✓Turns imperfect models into trustworthy products

Watch for

✓Few established conventions
✓Over-hedging annoys users

Continue exploring

Human in the Loop Response Streaming AI Citations AI Hallucination