Definition. UX design that anticipates ML/LLM imperfection: sets right expectations, makes mistakes recoverable, attributes sources, anchors on familiar patterns, collects feedback.
Five principles
- Set right expectations — disclaimers on AI output; surface known limitations on the landing page.
- Enable efficient dismissal — Copilot: keep typing to ignore. Chatbots: trivially closeable.
- Provide attribution — citations (BingChat); inline quotes; community-recommended badges.
- Anchor on familiarity — resist exotic UI. Chat is flexible but high-effort.
- Collect feedback in-flow — thumbs / regenerate; variations / upscale; accept-modify-ignore (implicit).
Mapping in the agentic-coding context
- Eng-bot in Slack. Show confidence; cite which files/commits backed the answer; trivially closeable. Don't reply unless asked.
- PR-review agent. Never approve, only comment. Require human "resolve" on each thread. Collapse low-confidence comments by default.
- Issue → PR loop. Open a draft PR with a clearly-marked "agent-authored, plan attached" badge; explicit "looks wrong" button.
- Production access (MCP). Surface every action as "this will do X" before execution; default dry-run; one-click rollback.
Yan drawing on Microsoft's 18 Guidelines for Human-AI Interaction, Google's People + AI Guidebook, Apple HIG for ML
Engineering clients under-invest here because their users are other engineers — "they'll figure it out." They won't; they'll just stop using the agent. 1–2 week deliverable that lifts adoption more than another month of prompt tuning.