.avif){kind=small}
The MIT AI Risk Initiative at MIT FutureTech and the Society of Actuaries (SOA) Research Institute have released a new report, AI Risks in the Actuarial Profession: Insights from Senior Leaders in Phase 1 of the Foresight Project.
Download the full report here: AI Risks in the Actuarial Profession
The report is the first output from the Foresight Project, a multi-method, multi-phase research initiative examining current and emerging AI risks to the actuarial profession. The project aims to provide evidence to inform decisions about AI governance, professional education, and workforce development.
For this first phase, MIT researchers conducted exploratory interviews with 13 senior actuarial leaders representing insurance companies,, consultancies, actuarial technology firms, academia, and professional bodies worldwide.
The interviews explored how generative and agentic AI are being adopted, the greatest AI risks, and how increasingly capable AI could reshape the profession, according to senior leaders in the profession.
Senior leaders described an actuarial profession in the early stages of integrating generative and agentic AI.
Although many actuaries are experimenting with AI for research, writing, coding, and administrative work, systematic integration into core actuarial workflows remains limited. Adoption also varies substantially between individuals and organizations, with pockets of advanced use alongside more widespread, superficial experimentation.
A recurring tension was the pressure from organizational leadership to adopt AI without clear implementation guidance, sufficient resources, or established governance frameworks.
Some leadersalso questioned whether organizations were adequately assessing AI's return on investment and the difficulties of converting promising demonstrations into reliable production systems.
Senior actuarial leaders identified several prominent AI risks, including:
These concerns broadly aligned with findings from previous MIT research involving 272 AI risk experts, which identified Finance and Insurance as a particularly vulnerable sector.
However, the interviews also revealed differences in priorities. Actuarial leaders placed greater emphasis on reputational damage, professional accountability, overreliance on AI, and the human consequences of adoption, including loss of foundational skills and employment.Some of these risk priorities may also shift as AI systems mature and organizations develop more effective safeguards.
One of the strongest areas of concern involved the future actuarial talent pipeline.
Senior leaders worried that AI could automate the foundational tasks traditionally undertaken by junior actuaries, reducing both entry-level employment opportunities and the practical experiences through which professional expertise develops.
Although automation could allow junior actuaries to progress toward more advanced work faster, it could also weaken their understanding of actuarial fundamentals.
This creates a challenge for the profession: How can actuaries develop the expertise needed to oversee and validate AI-generated work if AI increasingly performs the tasks through which that expertise was traditionally acquired?
Leaders emphasized the importance of preserving actuarial first principles, coding literacy, critical judgment, and communication skills. However, they provided differing perspectives on the expertise that the profession should prioritize, including deeper technological and model-building expertise and higher-level actuarial judgment and interpretation.
Leaders also anticipated changes to the boundaries of actuarial work.
As increasingly capable AI systems make technical and analytical tasks more accessible, other professions, including data science, statistics, and finance, can increasingly involve work traditionally performed by actuaries. While this trend preceded recent AI developments, it is now being exacerbated by it.
Leaders described several possible directions for the profession, including developing more advanced machine learning models, expanding into emerging risk domains, and focusing on the interpretation of complex risks and their implications for organizational decisions.
The interviews highlighted opportunities for actuaries to apply their expertise beyond traditional insurance markets, including climate change, geopolitical risk, and other complex societal challenges.
While the impact of AI on actuarial work has so far been relatively gradual, some leaders anticipated more disruptive changes as AI systems become increasingly capable and autonomous.
These changes could also affect the profession through broader economic and societal disruptions that destabilize insurance markets and create risks that are difficult to anticipate using historical data.
These findings suggest that preparing for AI may require actuaries to look beyond the immediate implications for productivity and consider how the underlying risks they are responsible for managing could change.
The interviews identified several opportunities for professional bodies, employers, and educators to prepare for these changes.
These include developing clearer guidance and standards for responsible AI adoption, reforming actuarial education to preserve foundational understanding while reducing rote learning, and expanding professional development to encompass technological expertise, systems thinking, and higher-level judgment.
Participants also emphasized opportunities to draw on the profession's established standards of accountability and ethics, which could provide a foundation for the safe and responsible use of AI.
This report provides an initial qualitative exploration of the perspectives of senior actuarial leaders. As the study involved a small, purposively selected sample, its findings are intended to identify emerging themes and research priorities rather than represent the views of the wider profession.
Subsequent phases of the Foresight Project will validate these findings by surveying SOA members, examine differences in AI adoption and risk perceptions, and investigate how AI could affect actuarial skills, employment, and professional responsibilities.
Together, these findings will help establish an evidence base to support the actuarial profession in navigating current and emerging AI risks.
This project was funded by a gift from the Society of Actuaries (SOA) Research Institute.
The report was authored by Kun Zhao, Alexander K. Saeri, Peter Slattery, and Neil Thompson of MIT FutureTech.
The authors thank Dale Hall and Ryan Silhavy at the SOA Research Institute for their valuable contributions to project planning, recruitment, and review, and for their support throughout the research.
We are also grateful to the senior actuarial leaders who contributed their time and insights to the interviews.
