Designing for AI for UX Designs
Designing for AI is the practice of creating interfaces for products that use AI, such as chat assistants, recommendations and voice or spatial interfaces. Key patterns include setting expectations, showing confidence or sources, letting users correct or undo results, handling errors gracefully, and building trust in uncertain outputs.
Stage: AI and Emerging UX · Part of the UX Design roadmap · Last reviewed 19 November 2026
What to learn in Designing for AI
- Design a chat interface with clear prompts, loading states and error handling
- Show sources or confidence so users can judge an AI answer
- Add controls to edit, regenerate and undo AI output
- Set honest expectations about what the AI can and cannot do
- Adapt interaction patterns for voice and spatial or multimodal contexts
Designing for AI interview questions
Why is it important to let users correct AI output?
AI can be wrong or misread intent, and users need a way to fix it without losing work. Editing, regenerate and undo controls build trust, reduce frustration and generate feedback that can improve the system.
What makes voice interfaces hard to design?
Voice has no visual cues, so users cannot see options or system status, and speech recognition can fail. Designers must give clear prompts, short confirmations, error recovery and ways to ask for help.
Where Designing for AI fits in the UX Design roadmap
Designing for AI belongs to the AI and Emerging UX stage (intermediate level), which takes about 4 hours. What changed.
Other units in this stage
Back to the full UX Design roadmap