Approaching AI in Banking
summary
Argues that AI in banking centers on operationalizing unit economics and segmentation into an auditable decision pipeline rather than flashy chatbots, with a hands‑on claim that banks turn Excel cash‑flows into a task graph wired to Dask and Jupyter tooling, parse legacy COBOL copybooks, model drivers with linear models and gradient‑boosting trees (neural nets are de‑emphasized), then run ILP/optimization to resegment customers; notes surprising infrastructure fact 5% hardware saturation and emphasizes human‑in‑the‑loop governance.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 1:04AI relevance in banking
- 3:10Proprietary vs non-proprietary AI
- 4:14Banking data flywheel
- 7:29Regulatory constraints in AI
- 9:20Building AI decisioning brain
- 12:13Excel as legacy system
- 13:39Legacy COBOL challenges
- 15:52Transitioning to model systems
- 18:25AI planning and optimization
- 20:52Dependency diagrams in credit
- 21:55Tech stack for AI banking
- 25:54Final thoughts
- 27:09Q&A: Task scheduling and streams
- 27:54Q&A: Multi-objective optimization
- 29:41Q&A: Model interpretability
- 32:55Q&A: Neural networks in decisions
- 33:59Q&A: Discrete optimization in credit
related talks
26mThe AI-Based Banking Era
16mMy Bank Meets AI: Smarter Decisions Beyond Fancier Spreadsheets
29mAI - Digital (r)evolution in banking
31mBanking Transformation with AI: Extracting Value and Pioneering Change
27mData-Driven Approach to Process Automation in Banking
15m60+ AI Use Cases! How CBA Is Shaping the Future of Banking
44mPanel Discussion: The BFSI Playbook - Scaling AI with Trust and Quality
46mThe AI dominated world – skills required for successfully navigating it