ML Beyond Metrics: Challenges of Using ML Systems in the Real World
summary
The talk examines practical challenges of deploying machine learning systems, insisting on concrete evidence like dataset size/composition, which features drove decisions, and how bias was detected. It covers deployed pipeline demos that show whether false positives or failures were reduced, plus measured safety or liability figures for applications (e.g., autonomous driving) and any legal or insurance outcomes tied to them.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 2:57ML beyond metrics
- 4:04Ethics and responsibility
- 10:50Government role in AI
- 17:19Target metrics and discrimination
- 27:19AI decision interpretability
- 35:12Self-driving ethics
- 40:58ML impact on society
- 53:28Q&A
related talks
46mI know Machine Learning but... am I ready for the real world?
21mHow The Application Of Machine Learning Is Changing The Way Companies Use Data
06mML for Systems and Systems for ML
06mMachine learning is about your data and deployment
27mA Look Inside Applied Machine Learning
48mWhen Not to Use ML
44mAdvancing Cybersecurity with AI and Machine Learning: Opportunities and Challenges
09mML and AI systems will make mistakes