Bridging the gap between AI and UI
summary
Reviews concrete model-interpretability tools and shows how they plug into product UIs: uses LIME and SHAP to attribute features, inspects RNN attention with ELMo/BERT, and indexes embeddings with Annoy for nearest-neighbor retrieval. Presents a working prototype with a why did I get it button that displays positive, negative and missing keywords so users correct preferences; voluntary feedback flows back to the recommender and yields stronger engagement and better next-time suggestions.
AI-written summary · about
chapters (5)
- 0:00Start
- 1:59Jurassic Park vs Sarah Jessica Parker
- 2:39What does it mean?
- 10:57Methods
- 13:27Human-Machine Interactions - Ofra Amir
related talks
31mA more human-centered approach to AI
55mBuilding trusted AI products
26mHow to develop useful and unbiased AI
34mInterpretable AI or How I Learned to Stop Worrying and Trust AI
39mThe Search for Transparency and Accountability in the Age of...
01h 00mData, visualization, and designing AI
27mMisplAIced: Good UI Needs Good UX
39mMind-Reading with Adaptive and Intelligent User Interfaces