Risky Science Podcast · Risk Market News

Black Box Problems, Machine Judgment and the Rules Nobody's Written Yet With Daniel Schwarcz

·53 min·5 clips

Chris Westfall frames the episode around a line that has become common in insurance: that AI will improve efficiency and customer outcomes. He treats that language as a starting point, not a finish line. Daniel Schwartz joins from the University of Minnesota Law School, where he teaches insurance law, contract law, tort law, and financial regulation, and where his academic work sits at the intersection of AI governance and insurance regulation. The conversation quickly moves beyond slogans. Westfall asks whether the industry is really thinking about these issues yet, and Schwartz says many insurers are, but not in one uniform way. He separates broad hype about replacing entire agent networks from the more ordinary uses of AI already being considered inside insurance operations. He also points out that different lines of insurance and different uses of AI carry different risk profiles. Some systems support employees. Some automate parts of the workflow. Some sort easy cases from hard ones. The interview keeps circling back to how regulators will judge those uses. Schwartz stresses that the governance language around AI is still broad and unsettled. Auditing and internal controls matter, but the standards are not always exact. That uncertainty creates friction for insurers that want to use AI aggressively. It also creates legal bills, since firms need lawyers to help them work through the rules. The discussion puts special weight on the state-by-state structure of insurance regulation. That structure makes a national AI strategy harder to manage and easier to get wrong. Westfall and Schwartz also talk about the legal risks of AI-driven underwriting. They connect those risks to the possibility of shadow insurance and regulatory arbitrage. Schwartz explains that the industry is often drawn to AI because prediction is already part of the business. That makes the technology seem familiar, even when the legal environment is not. The episode also introduces what Schwartz calls Obamacare for homeowner's insurance, a phrase that signals the way policy design, market structure, and public regulation can blur together. The interview does not treat that idea as a slogan. It uses it to explore how insurance markets are shaped by institutions, public rules, and competing incentives. Westfall keeps the conversation grounded in practice. He presses for concrete distinctions, asks what kinds of deployment matter most, and returns to where the system breaks down. Schwartz answers in terms of assumptions, institutional behavior, and the limits of current rules. By the end, the episode has built a clear picture of why AI in insurance is not just a technical story. It is a legal one, a regulatory one, and an organizational one. The conversation ends with the expectation that insurance will change, and that the broader economy will change with it.

As heard by us

A careful look at AI in insurance when the rules lag the pitch.

Risky Science Podcast frames AI in insurance as a governance problem rather than a product pitch. Daniel Schwarcz lays out the gap between what insurers say AI can do and what law, regulation, and ethics are likely to allow, with steady attention to underwriting, state-by-state…

Read the full review in PlayNext →

Why you'd press play

You want the legal, regulatory, and ethical questions behind AI in insurance, not another efficiency slogan.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Risky Science Podcast