This week’s guest is Dr. Hansi Singh, an Earth system scientist who’s worked at the U.S. Department of Energy, taught in academia, and now leads a startup called Plannette.AI, which blends physics and AI to deliver long-range weather forecasts for industries including finance and insurance.

In this conversation, Dr. Singh explains why AI’s biggest weather-forecasting success have been within the seven-day window—and why pushing beyond that horizon remains so hard. We explore how AI can enhance—not replace—traditional physics-based models.

We also get into the practical side: how finance, insurance, and even energy traders are using AI-driven forecasts, what the rise of AI agents means for accessibility, and why transparency and back-testing are critical to overcome the industry’s skepticism toward “black-box” models.

Dr. Paul Wilson, Head of Catastrophe and Climate Research at Twelve Securis, a leading insurance-linked securities asset manager, sits down for a live recording of the Risky Science Podcast.

We sit down with Dr. Seth Baum, Executive Director of the Global Catastrophic Risk Institute and research affiliate at Cambridge University’s Centre for the Study of Existential Risk. We explore how societies understand and prepare for global-scale threats—from climate change and pandemics to nuclear conflict and artificial intelligence. Dr. Baum explains why uncertainty is the defining feature of catastrophic risk, why markets struggle to price the unthinkable, and why collective action and governance are essential to tackling the crises that no private market can solve.

Join the Risky Science Podcast for a live discussion with Dr. Paul Wilson, Tuesday, September 23, 11 a.m. ET.
Register Here

We speak with Dirk Schmelzer, Partner at Plenum Investments in Zurich. Dirk has spent more than 15 years managing catastrophe bond and insurance-linked securities funds, and he brings a practitioner’s perspective on how catastrophe models are actually used in portfolio management and investment decisions.

We’ll explore how models have evolved, where they still fall short, and how issues like climate change and artificial intelligence are reshaping the conversation.

This week we speak with Karen Clark, founder of Karen Clark & Company, about the evolution of catastrophe modeling and the shift toward higher-frequency, climate-driven events.

John Seo, founder of Fermat Capital, about the lessons of Katrina for catastrophe bonds and models 20 years later.

• (00:00) - Introduction

Less than a year after the devastating Los Angeles fires, I’m joined by Michael Wara from Stanford University.

We explore why Michael is skeptical about California developing a public wildfire model, despite being part of the strategy group that studied it. We'll dig into how the newly approved private wildfire models are about to transform California's insurance market. And we'll discuss something that's crucial but often overlooked: how community-scale risk mitigation efforts can and should be integrated into these models.

We talk with Dr. Victor Gensini, a professor at Northern Illinois University and one of the leading experts on severe convective storms. Dr. Gensini works with the Insurance Information Institute and has just launched a new center for convective storm research, bringing together academic research and industry needs to tackle this modeling challenge.

We'll explore why these storms are so much harder to model than hurricanes, what new data sources are filling the gaps in our understanding, and why we're still five to ten years away from having reliable catastrophe models for severe convective storms. 

In this episode we talk with Dr. Tina Dura, a coastal hazard researcher at Virginia Tech, as she unpacks a threat that most risk models still underestimate: Sudden land subsidence from a long expected Cascadia subduction zone earthquake.