Designing With/For AI

Today’s activity is a discussion:

• AI as a tool for design: opportunities and challenges
  • Generative UI and Outcome-Oriented Design
  • How Generative AI Is Remaking UI/UX Design | Andreessen Horowitz
  • UX research: identifying jobs-to-be-done, ideating core user journeys, etc.
  • Implementation: code generation, graphics, motion, accessibility, etc.
  • Reflection on designs: misinterpretation
• Designing interfaces powered by AI
  • AI design patterns
    • search
    • recommendations
    • conversational interfaces
    • tool use
    • model as router
    • predictive text
    • etc.
  • Explanations, resilience
  • Manipulation, and privacy

Questions:

• How should we handle AI errors in design?
• How can we design interfaces that accurately show what AI can and can’t do?

Resources we discussed

• Air Canada chatbot costs airline discount it wrongly offered customer - CBS News
• [2401.10873] An AI-Resilient Text Rendering Technique for Reading and Skimming Documents and related AI-Resilient Interfaces

Prompts we used for AI feedback

“Here’s a screenshot of a website landing page. Give constructive criticism: first give a list of types of feedback that the designers might find valuable, then give specific examples of each type of feedback.”

“How can I make this website more accessible?”

“What are some possible restrictions that someone with visual impairments might encounter when using my website?”