Why Do You Need an Explainable Ai?
summary
Argues that lack of transparency causes real legal and bias failures and that explainability is practical not philosophical, using LIME, SHAP and counterfactual explanations as the actual tools. Describes an implemented open source stack (Java and Python library plus container/Kubernetes service) called trusty core that inserts an explainability layer and a human‑in‑the‑loop risk filter into existing ML pipelines. Provides loan/resume/claims examples, a hub‑and‑spoke governance model, and code + blog links; demo fails but code is available.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 3:40AI bias and transparency problems
- 8:00Transparency in AI systems
- 9:51Black box AI models
- 10:34Explainable AI for transparency
- 12:02Methods of explainable AI
- 15:42LIME and SHAP algorithms
- 17:26Counterfactual explanations
- 19:11Understanding PDPs and their applications
- 21:16Explainable AI use cases
- 24:49Implementing explainable AI
- 26:55Trusty AI introduction
- 28:42Trusty AI workflow
- 31:10AI risk management model
- 32:58Ethical AI with XAI
- 33:39Adopt explainable AI
related talks
16mBuilding Trust through Explainable AI
24mWhy and How to Build Explainability into your ML Workflow
41mThe Explainability Problem: Towards Understanding AI
15mIntroduction to Explainable AI
01h 29mNavigating ML Explainability: From Theory to Implementation
26mUsing Explainable AI to garner trust and overcome adoption inhibitors
32mOn the Importance of Explainability
28mThe Need for Explainable AI