Behind the AI Curtain - Designing for Trust in Machine Learning Products
summary
A design team redesigns a predictive-analytics score for legislative bills by removing a misleading area chart, hide numeric details behind summarized copy, reorder contributing reasons to reflect algorithmic weight, and use situational user interviews (ask colleagues-how-you’d-explain prompts) as the success metric. A search finds 140 internal docs explaining the feature, proving internal confusion. Testing shows users can accurately explain the score to a coworker, so the redesign ships and increases prioritize explainability over raw transparency without breaking revenue.
AI-written summary · about
chapters (7)
- 0:00Designing for trust in AI
- 2:52Why trust in AI matters
- 12:13Less is more in predictive analytics
- 14:40Asking the right research questions
- 19:33Clear writing in UX
- 22:03Redesigning predictive analytics feature
- 31:11Key takeaways and AI design future
related talks
37mBehind the AI Curtain - Designing for Machine Learning Products
16mTrust by Design: Building Reliable AI for Sensitive Environments
47mDecoding the algorithm - why explainability and transparency matter when buildin...
26mUsing Explainable AI to garner trust and overcome adoption inhibitors
24mWhy and How to Build Explainability into your ML Workflow
39mGuide towards algorithm explainability in machine learning
24mDesigning for Transparency in Machine Learning
39mThe Search for Transparency and Accountability in the Age of...