Episode #163
Jun 12, 2025
The Intuition Behind Generative AI
Frank Schmid from Gen Re
SPONSORED BY
Frank Schmid from Gen Re shares his expertise on the intuition behind generative AI. Discover how this emerging technology continues to reshape insurance decision-making, why narrative still matters, and what insurers can do to prepare for an AI-driven future.
All Episodes Available On
The insurance industry stands on the brink of an AI revolution, and few understand this transformation better than Frank Schmid, Chief Technology Officer at Gen Re. In Episode 163 of the InsurTech Geek Podcast, Frank joins host James Benham for a compelling discussion about the intuition behind generative AI, the shifting nature of predictive models, and how insurance companies can build a sustainable AI future.
From cloud architecture to human-in-the-loop decision-making, this episode offers actionable insights for any insurance leader or tech enthusiast looking to stay ahead of the curve.
From Economist to InsurTech Leader
Frank Schmid’s journey is anything but ordinary. Born and raised in Germany, his academic path spans a Ph.D. in Economics, a postdoc in corporate finance, and time at Wharton. His career launched at the Federal Reserve Bank of St. Louis, where his passion for Bayesian modeling eventually drew him into the insurance world.
Frank’s move to NCCI and later AIG provided a deep dive into the complexity of long-tail insurance like workers’ comp—a sector he describes as both challenging and intellectually stimulating. His current role at Gen Re represents the culmination of this experience, now focused on leading AI adoption at scale.
AI as a Prediction Machine
One of the episode’s most powerful insights is Frank’s framing of generative AI as a prediction machine. Unlike traditional models that reveal causality (e.g., generalized linear models), generative AI excels at making accurate predictions without necessarily explaining them.
“We have very, very powerful prediction machines, but we don’t understand the machines.” – Frank Schmid
This abstraction can be unsettling for industries like insurance that thrive on transparency and regulation. But as Frank explains, it’s all about trust in the performance output rather than the process.
Human-in-the-Loop is Non-Negotiable
Both James and Frank champion a “human-in-the-loop” model for AI adoption. Rather than replacing human decision-making, AI should augment it—handling repetitive tasks like data aggregation and summarization while leaving critical judgment calls to professionals.
“Software and data are there to support decisions, not replace them.” – James Benham
This balance is crucial in insurance where decisions often involve financial, ethical, and regulatory complexities.
Building an AI-Ready Organization
Frank outlines a clear, three-step framework for companies aiming to adopt AI effectively:
- Cloud First: Migrate infrastructure to the cloud.
- Modern Data Architecture: Establish pipelines and governance.
- AI Layer: Integrate generative AI on top of a trusted data stack.
At Gen Re, this has translated into merging infrastructure and security teams for better alignment and using platforms like Microsoft Azure and Power BI to consume and deploy AI tools across the enterprise.
“The adoption of generative AI is like electricity—it will take many years, but it’s inevitable.” – Frank Schmid
Real-World Applications of AI
Among the most impactful use cases discussed:
- AI copilots for coders: Used to streamline programming with tools like Databricks Assistant.
- Document summarization: Especially for medical claims, dramatically reducing processing time.
- Treatment recommendations: AI tools suggest and evaluate medical treatment paths in workers’ comp.
These aren’t experimental projects—they’re in production, proving real ROI and improved operational efficiency.
Be Strategic, Not Hasty
Frank’s parting advice to insurance leaders is to be methodical and patient. AI’s full potential will unfold over years, not months. But the steps taken now—especially around cloud, data, and organizational readiness—will determine who leads and who lags in the AI-driven future.