Panelists, Martha Conlin, a Partner with Troutman Pepper Locke, John LaBarbera, a Partner with Kennedys, and Dan Waterloo, an academic and Adjunct Professor with the Illinois Institute of Technology delivered a cutting edge presentation on the acceleration of AI in insurance.
This session was designed to both demystify AI and underscore how quickly it is reshaping insurance.
Dan Waterloo from IIT started with fundamentals, explaining AI as a “function machine” that turns inputs into outputs via increasingly complex transformations, moving from simple regression examples to modern neural networks and large language models. Attorneys Martha Conlin and John LaBarbera then walked through practical insurance applications, including underwriting, policy origination, claims handling, fraud detection, document summarization and internal knowledge management, to address what it takes to build or deploy private and secure AI platforms. A recurring theme was the growing appeal and risks associated with the use of open-source models in a heavily regulated sector.
Much of the discussion focused on the regulatory and liability landscape now forming around insurer use of AI. Attendees were briefed on emerging federal and state regimes, including AI-specific laws and broader privacy statutes, as well as regulations governing automated decision tools. On the insurance side, the NAIC’s AI Model Bulletin and early state adoptions were highlighted as setting expectations for governance, consumer notice, vendor oversight, and bias mitigation.
The panel also flagged three major liability fronts: regulatory violations where AI-driven decisions run afoul of privacy or fairness rules; disparate-impact exposure when models are trained on biased historical data; and privacy risks tied to large-scale model training and uncontrolled employee use of third-party AI tools.
The overarching takeaway was that although AI offers efficiencies for insurers, it must be paired with rigorous oversight, transparency, and caution.
A video replay of this presentation can be accessed in the AIRROC On-Demand Library at https://airroc.memberclicks.net/airroc-on-demand.
