The Most Interesting Thing in A.I. Building Trustworthy AI - with Nicholas Thompson and Navrina Singh
27 snips
Nov 19, 2025 In this engaging discussion, Navrina Singh, Founder and CEO of Credo AI, shares her insights on building trustworthy AI systems. She explores the lessons from Microsoft's Tay disaster, emphasizing the need for early governance in AI deployment. Navrina argues that proper regulation can enhance innovation and help mitigate bias while keeping trust at the forefront. They also discuss how AI might widen inequality and the importance of representation in development teams. Finally, Navrina highlights the potential for AI governance to evolve significantly in the coming years.
AI Snips
Chapters
Books
Transcript
Episode notes
Publish Clear Transparency Reports
- Require transparency reports that state what you tested, what you cannot guarantee, and which risks you accept.
- Use measurable testing for proprietary components and disclose limitations for open-source or foundation models.
Products Reflect Their Builders
- Product fairness depends on who builds and tests the systems as much as the data itself.
- Navrina warns that homogenous development teams produce products that fail large segments of users.
Equality Hinges On Data Choices
- AI's impact on equality depends on human choices around data, languages, and access.
- Navrina warns Global South languages and datasets are underrepresented, risking wider inequality.
