AI in Healthcare: Bias, Privacy, and Trust in Kube-Native Systems
summary
Shows a Kubernetes-native stack that embeds ethics into deployment: Kubeflow pipelines run Fairlearn and IBM AIF360 audits after each training job, Prometheus tracks fairness drift, and an admission-controller plus OPA Gatekeeper blocks promotions that fail checks. Describes an actual hospital readmission model that underperforms on older patients, where continuous fairness checks detect drift and auto-rollback restores the last good model. Recommends explainability sidecars, federated learning with differential privacy, and audit logs for traceability.
AI-written summary · about
chapters (11)
- 0:00Introduction and Welcome
- 1:14The Role of Kubernetes in AI
- 2:48Challenges and Risks in AI for Healthcare
- 4:22Leveraging Kubernetes for Ethical AI
- 5:53Fairness in AI Models
- 8:04Explainability in AI
- 10:16Ensuring Privacy in AI
- 12:13Scaling Ethical AI
- 14:25Regulatory and Ethical Alignments
- 16:02Practical Implementation Strategies
- 20:02Conclusion and Key Takeaways
related talks
30mStand up for ethical AI! How to detect and mitigate AI bias using Kubeflow
22mAI in Kubernetes The Promise
36mKeynote: Ethics in AI: Building Fair and Just Intelligence Systems
25mSecuring AI Pipelines: Real-World Attacks on Kubernetes-Based AI Infrastructure
46mCan you trust your AI?
28mLife is not fair, but your ML pipeline can be!
12mTrust and transparency of AI for the enterprise
36mInfusing Trusted AI using Machine Learning Payload Logging on Kubernetes