Why Transparent AI is the only AI that can be trusted
summary
Argues that trust requires knowledge representation and reasoning integrated with machine learning rather than black‑box deep nets alone; compares NVIDIA’s end-to-end pixel attribution for autonomous driving (helps visual focus but not action context) with a nanopore DNA project where qualitative modelling yields 84% unsupervised classification versus ~30% for ML attempts. Cites adoption gaps—less than 15% deployed AI widely—and executive surveys showing demand for transparent, causality‑based decisions.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 2:12Challenges
- 5:30General AI
- 6:51Singularity
- 10:26Neural Network
- 13:12AI going wrong
- 15:14Machine learning algorithms
- 22:52Responsible AI
- 25:06Responsible AI design
- 26:26GDPR
- 27:40Create your own AI
- 28:53Benefits of Transparent AI
- 29:34Questions
related talks
30mTrustworthy AI - opening up the black box
39mThe Search for Transparency and Accountability in the Age of...
32mHow to Build Trustworthy AI with Open Source
37mBuilding a case for Fair and bias-free AI
44mAI, Corporate Responsibility & Democratic Legitimacy
46mCan you trust your AI?
28mThe Need for Explainable AI- 39mSponsored Session: The Search for Transparency and Accountability in the Age of AI: RAI & XAI as Essential Tools