Responsible AI in Practice: Real-World Examples and Challenges
summary
Panel argues responsible AI needs operational rules not slogans, describing concrete measures: use IBM trustworthy AI toolkit to detect dataset bias during feature engineering, augment underrepresented groups with synthetic training data, set an interdisciplinary governance board like IBM’s ethics board, and expect enterprise rollouts to take 12–24 months with training, procurement controls and audit trails. Cited failures include biased hiring tools and facial recognition errors, and one vendor uses guardrails as a customer-facing differentiator.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 2:11Panelists' backgrounds
- 4:01Defining responsible AI
- 8:18Private sector state
- 12:30Who is responsible
- 14:39Implementation challenges
- 16:47Large vs small companies
- 19:42Industry-specific challenges
- 22:53Public pressure and transparency
- 25:25Education and awareness
- 26:52Business case for AI
- 28:36AI ethics training
- 29:19Developer tools
- 31:07User feedback education
- 32:30Closing and Q&A
related talks
34mResponsible AI: The Key to Ethical, Safe and Inclusive Software Development
26mResponsible AI-What, Why, How and Future!
39mBuilding a Responsible AI Governance Program
37mBuilding a case for Fair and bias-free AI
40mSocially responsible AI: Practical tools for robust design
15mResponsible AI 2.0 : NO AI without RAI
22mA story arc for responsible AI
44mPanel: Building responsible, human-centric AI models