Episode Summary
Executive Summary: Scott Galloway interviews Josh Tarangol about his book AI for Good, arguing that AI’s real value is not hype-driven utopia or apocalypse but disciplined problem-solving in healthcare, government, warfare, and media. They discuss sepsis prediction, IRS modernization, Palantir, autonomous systems, China’s AI pricing strategy, job displacement, and the need for policy and human judgment to shape outcomes.
Main Topics: AI beyond hype: separating technology from tech companies: Tarangol says his book was inspired by the noise around AI—both utopian promises and doomsday fears—and by Danny Hillis’s advice to separate the technology from the companies selling it. The real story is practical experimentation by people solving concrete problems. Healthcare use cases and sepsis prediction: The Cleveland Clinic case study shows AI being used in a narrow, measurable way: predicting sepsis, refining the model in practice, and reducing mortality. The discussion emphasizes incremental gains over miracle solutions. Government modernization and the IRS: The conversation focuses on how AI and machine learning can modernize bureaucratic systems, especially the IRS and federal logistics. Operation Warp Speed is cited as proof that government can move fast when the mission is clear, but political backlash and DOGE-style disruption can also derail progress. Palantir, procurement, and public-sector software: Tarangol defends Palantir as an effective software company that broke through Pentagon procurement inertia. He argues the firm’s founder culture, persistence, and product quality matter more than ideological objections. AI, jobs, and the pace of labor-market disruption: The speakers debate whether AI will cause a job apocalypse. Tarangol says the data is not yet conclusive, but the main uncertainty is how fast adjustment will occur. He expects productivity gains over time, with the real risk being a rapid transition. War, drones, and asymmetric conflict: Ukraine is presented as a live laboratory for AI-enabled warfare, with cheap drones, image recognition, and remote operations reshaping combat. Tarangol argues future wars will be fought more from command centers and screens than traditional battlefields. China, cost undercutting, and AI dumping: Galloway proposes that China may use subsidized, lower-cost AI models to win market share, similar to its past industrial strategies. Tarangol agrees this is plausible and notes concerns over distillation, IP theft, and the lack of profitability in U.S. frontier AI.
Key Arguments: AI is most useful when applied to specific, measurable problems rather than framed as a universal solution or existential threat. Healthcare shows AI’s best current promise: not perfect diagnosis, but meaningful improvements in outcomes such as reduced sepsis mortality. Government can be improved by AI if agencies actually want better service and modernization, but political dysfunction can undo progress. Palantir succeeded because it had a real product, persistent execution, and a culture of urgency, not merely because of political branding. The future of warfare will be increasingly asymmetric, drone-driven, and data-intensive, advantaging smaller, more adaptive forces. AI’s labor-market effects are still ambiguous in the data; the key question is whether disruption unfolds over 3, 5, or 30 years. China could pressure U.S. AI firms through cheaper models and possible model distillation, especially if American companies cannot show clear ROI. AI will likely raise productivity, but it may also deepen inequality if benefits accrue mainly to firms, elites, and the wealthy. Education should emphasize identity formation and critical thinking, not just credentialing, because AI can easily automate the latter. Journalism’s moat is original reporting and new information; AI is more likely to commoditize low-value content than replace serious newsrooms.
Data Points: Sepsis deaths in the U.S.: About 300,000 per year - Tarangol cites sepsis as the deadliest condition in America, exceeding several major diseases combined. Sepsis mortality reduction at Cleveland Clinic: 41% - Reported reduction after deploying and refining the Bayesian sepsis prediction model. Lives potentially saved at Cleveland Clinic: More than 1,000 - Derived from the 41% mortality reduction in the hospital system. Cleveland Clinic system size: 80,000-person system - Used to describe the scale of the hospital network where the AI model was deployed. IRS modernization timeline: Starting in 2014 - Tarangol says IRS leaders quietly began modernizing the system around this time. Operation Warp Speed software cost: About $16 million - Tarangol describes the cost of the digital-twin supply chain project built to manage vaccine distribution. AI tool adoption in enterprise: 67th AI tool - A sponsor read says companies keep adding AI tools, creating more security blind spots. LinkedIn ads ROAS: 121% - Mentioned in sponsor copy, not the interview content. Consulting firm hiring outlook: Hiring through the rest of 2026 - Tarangol notes some large firms still expect to hire despite AI-related concerns. China vs. U.S. model gap: Previously 9-12 months ahead - Tarangol says the gap between U.S. frontier models and Chinese models was once thought to be this large, but is shrinking.
Pivotal Quotes: "You need to separate the tech from the tech companies because then you're going to start finding interesting things, like real experimentation and real people with values." — Josh Tarangol: Explains the origin of his book and why he moved away from AI hype coverage. "What I'm looking for with AI is not is it going to solve my problem, but is it going to improve the solution that I had before?" — Josh Tarangol: Describes the Cleveland Clinic’s pragmatic approach to AI in healthcare. "The value of identity formation in the age of AI is never going to be greater." — Josh Tarangol: Warns that education must do more than credentialing if AI can automate many tasks.
Implications: AI’s near-term impact will come from boring but powerful operational gains in healthcare, government, warfare, and media. Winners will be groups that pair technology with mission clarity, human judgment, and good governance; losers will be those chasing hype or ignoring policy and culture.