In Search of Trustworthy AI
About this episode
Craig Hatkoff discusses the importance of securing trustworthy information from AI and presents a four-stage model for verifying AI's output.
Craig Hatkoff has spent four decades at the intersection of innovation, culture-building, and institutional transformation. He pioneered commercial mortgage securitization at Chemical Bank, co-founded the Tribeca Film Festival alongside Robert De Niro and Jane Rosenthal after 9/11, and .in 2010 co-founded the Disruptor Awards with Harvard Professor Clayton Christensen and Irwin Kula. His latest initiative is Dragon Camp, which provides a methodology for using CGI (Collaborative General Intelligence) as a practice framework for creating a viable human-AI partnership. In this episode focused on unlocking the potential of AI in a humanist manner, the first order of business is to secure trustworthy information from AI. To that end, Craig discusses a four-stage model for verifying AI’s output. The first element is to leverage output from multiple AI sources, rather than just one, in order to guard against what have been called “AI hallucinations” or “fabrications.” To do so, moves organizations beyond stage 1: single-source vulnerability. Stages 2 through 4 then ramp up from cross-checking via multiple AI models (stage 2), to human intervention to verify (stage 3), culminating in…
More episodes of New Books in Economic and Business History
- Jake Dyble, "Managing Maritime Risk in Early Modern Europe: General Average in Law and Practice in Seventeenth-Century Tuscany" (Boydell Press, 2025)
- Sarah McNamara, "Ybor City: Crucible of the Latina South" (UNC Press, 2023)
- Margaret O’Mara on the Clintons, Tech, and Memory
- Eileen Otis, "Walmart: Made in China" (Stanford UP, 2026)
- Navigating Landmines at Work: Differences Can Create Value
- Turning IBM's Culture Massively Around
Explore listener stats, chart rankings, contacts and more on the New Books in Economic and Business History podcast page.