Generative artificial intelligence (GenAI) is contributing to changes across the insurance industry, reshaping both back-office and client-facing activities across the insurance value chain. Large language models (LLMs) and artificial intelligence (AI) agents are now being applied to underwriting, claims handling, customer service, actuarial modeling, valuation and reporting, risk assessment, pricing support, legal compliance, marketing, and daily professional productivity. GenAI can reduce operational effort, improve access to unstructured information, accelerate reporting and documentation, and support more consistent workflows, but it also introduces new technical, ethical, regulatory, and professional challenges that actuaries may need to understand and govern.
As adoption expands, five contextual challenges shape the need for this guidance:
- Professional roles and responsibilities are shifting as actuaries move from direct production of analyses toward supervision, validation, and governance of AI-enabled workflows.
- The opacity of modern LLMs creates trust, interpretability, and explainability challenges for practitioners who must justify methods, assumptions, and outputs to executives, regulators, boards, and policyholders.
- Validation practices will likely need to evolve beyond traditional statistical testing to address GenAI model outputs, including summaries, extracted information, generated explanations, and image-based assessments.
- AI systems introduce distinctive safety and security risks, including prompt injection, data leakage, adversarial attacks, hallucinations, jailbreaking, and bias amplification.
- Successful productization typically benefits from robust deployment, monitoring, version control, incident response, and retraining processes so that systems remain reliable, as data, language, customer behavior, and regulatory expectations change.
Against this backdrop, the Society of Actuaries Research Institute has published a report that focuses on the practical governance questions raised by GenAI and AI agents in actuarial and insurance settings. It is structured into a Practical Guide and an Educational Handbook. The Practical Guide establishes the ethical, regulatory, supervisory, and standards-related landscape relevant to responsible GenAI use, including themes such as fairness, transparency, accountability, privacy, traceability, and security, alongside mandatory documentation requirements. The Educational Handbook provides the conceptual foundation to guide actuaries in informed GenAI practice.
Main topics of discussion:
- Introduction to GenAI: Impact on actuaries and the insurance industry.
- Practical Guide: Ethical and regulatory frameworks.
- Practical Guide: Existing standards and volutary guidance frameworks.
- Practical Guide: Governance and ethics checklist within the AI lifecycle.
- Educational Handbook: Key concepts about AI systems and their validation.
- Educational Handbook: Case studies with application of the AI ethics and governance framework.
Download the full paper (PDF).
Sponsored by the SOA's Actuarial Innovation and Technology Strategic Research Program.