Podcast
Responsible & Explainable AI: Practical Guide to Bias Detection, Fairness & Governance
Open original DataTalks.Club episode
Responsible & Explainable AI: Practical Guide to Bias Detection, Fairness & Governance
Original Episode
Use these links for the canonical episode and media sources.
- Open the original DataTalks.Club podcast page
- Watch on YouTube
- Listen on Spotify
- Listen on Apple Podcasts
Episode Overview
How do you detect bias, enforce fairness, and govern AI systems in production without sacrificing business outcomes? In this episode, Supreet Kaur — AVP on Morgan Stanley’s Data Strategy and Products team, founder of DataBuzz, and mentor at Columbia and Rutgers — walks through a practical roadmap for responsible AI and explainable AI grounded in real-world examples.
People
Use these links to connect the episode to guest notes.
Chapter Summary
Use these checkpoints to decide whether to open the source transcript.
- 0:00 - Episode Introduction: Responsible and Explainable AI
- 2:14 - Career Journey: Master’‘s, Consulting, and Founding DataBuzz
- 3:54 - Data Strategy Role: Building AI Products at Morgan Stanley
- 4:43 - Responsible AI: Definition, Trust, and Stakeholder Collaboration
- 6:42 - Credit Decision Bias Example: Explaining Disparate Outcomes
- 8:20 - Explainable vs Responsible AI: Post-mortem Tools vs Governance Mindset
- 10:30 - Glass-Box Approach: Explainable AI Techniques Overview
- 11:36 - Data-Level Fairness Checks: Skewness, Missingness, and Coverage
- 12:48 - Exploratory Data Analysis for Bias Detection
- 14:39 - PII Handling: Age, Gender, Masking, and Use-case Justification
- 17:20 - Feature Necessity: Product, SME, and Compliance Decisioning
- 18:27 - Automating Data Quality: DQ Tools, Alerts, and Monitoring
- 19:03 - Model Explainability Tools: What-If, Skater, and AI Explainability 360
- 23:24 - Local Interpretability: LIME, SHAP, and Surrogate Models
- 24:22 - Ethics vs Profitability: Balancing Fairness and Business Objectives
- 27:38 - Cross-Functional Governance: SMEs, Compliance, and Leadership Roles
- 32:29 - Accuracy vs Interpretability: Managing Model Complexity Trade-offs
- 35:28 - Human-in-the-Loop: Limits of Automation and Responsible Oversight
- 37:31 - Detecting Drift & Feedback Loops: Demographics, Overfitting, and KS Tests
- 42:39 - Regulated Industry Perspectives: Finance, Pharma, and Risk Sensitivity
- 44:07 - Hiring Tool Case Study: Historical Bias and Remediation Lessons
- 50:17 - AutoML Risks: Democratization, Oversight, and Responsible Usage
- 52:08 - Community & Mentorship: DataBuzz Resources and Networking
- 53:50 - Data Career Landscape: Analyst, MLOps, Consultant, and Strategist Roles
- 56:44 - Ethics Training: Professional Responsibility for Data Practitioners