Explainable Artificial Intelligence and Continuous Delivery: Towards the intelligent organization
summary
Argues that adoption stalls because deep learning black-box models disconnect ML from business, and proposes CD4xML — a concrete variant of CD4ML that mandates explainable models so features and predictions map to domain language. Recommends combining Domain-Driven Design and Data Mesh to align data/products with domains, and using white-box techniques (additive and rule-based models), saliency maps and gradient and post-hoc methods where needed, so CI/CD pipelines deliver models whose prescriptions directly drive and close the loop on business actions.
AI-written summary · about
chapters (11)
- 0:00Explainable AI intro
- 1:54Analytics evolution stages
- 4:06MLOps and CD4ML today
- 8:15ML delivery limits
- 10:49Domain ML connections
- 11:55Defining explainable AI
- 16:11XAI methods taxonomy
- 20:26XAI techniques key
- 23:42White-box models bridge
- 24:24CD4xML for explainable ML
- 25:28Building intelligent org
related talks
33mContinuous Delivery Principles for Machine Learning
18mContinuous deployment of Machine Learning models
17mContinuous deployment for ML: The new software development lifecycle
27mContinous integration of ML products and UX design
14mGetting Machine Learning applications into production using MLOps and CD4ML
41mCan Data & ML really learn from software engineering?
33mAI with a DevOps mindset - experimentation, sharing ..
54mArtificial Intelligence seen from the software development lifecycle perspective