Moving Forward with Explainable AI: Adoption Rate & Essential Dimensions for a Reliable Use
summary
PwC reports 4.5 operationalized AI use cases per company and finds 46% fear productivity loss, 44% fear lost customer confidence and 25% fear legal risk when AI is unreliable. A practical framework combines organizational, regulatory and technical axes into five dimensions: comprehensibility, performance, faithfulness, information type, granularity. Prototype tooling uses SHAP and LIME explainers, measures surrogate fidelity by quantifying model‑surrogate distance, supports classification/regression only, and flags that fully automated explainability remains unrealistic without human review.
AI-written summary · about
chapters (10)
- 0:00Welcome
- 1:45Speaker backgrounds
- 3:11Blind faith in AI risks
- 7:31Explainable AI dimensions
- 9:35Performance dimension metrics
- 11:40Comprehensibility challenges
- 13:27Local vs global explanations
- 14:33Faithfulness evaluation
- 15:56Information types in explainability
- 18:02Challenges and future outlook
related talks
26mUsing Explainable AI to garner trust and overcome adoption inhibitors
28mThe Need for Explainable AI
35mWhy Do You Need an Explainable Ai?
15mIntroduction to Explainable AI
24mWhy and How to Build Explainability into your ML Workflow
39mThe Search for Transparency and Accountability in the Age of...
32mOn the Importance of Explainability
43mBring in the Lawyers: Explainable AI Driven Decision-making