Operationalizing Responsible AI
summary
Teams describe concrete practices: a weekly ship room where product teams, fairness experts and OKRs force a documented fairness plan and systematic testing for speech and vision models; use of Fairlearn and InterpretML plus model cards and differential privacy for global, hybrid and tenant-specific models; an equality-of-opportunity ranking deployed in recruiter search that yields a threefold increase in queries with representative top results; defaults and flags (e.g., zip code proxy) and continuous monitoring to catch drift.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:00Meet the panelists
- 4:15History of Responsible AI at Salesforce
- 5:50How customers are thinking about Responsible AI
- 7:25How to Operationalize Responsible AI
- 14:55Responsible Expert Teams
- 22:05Examples
- 23:55Ethics Team
- 27:25Experience at LinkedIn
related talks
31mOperationalizing Responsible AI
39mResponsible AI
19mA developer's guide to responsible AI review processes
11mAddressing Privacy, Bias, and Fairness in AI Systems
36mKeynote: Ethics in AI: Building Fair and Just Intelligence Systems
40mResponsible AI Practices: Fairness in ML
34mResponsible AI: The Key to Ethical, Safe and Inclusive Software Development
40mSocially responsible AI: Practical tools for robust design