Artificial Intelligence Research Reports
If you’re looking for research projects and reports on artificial intelligence topics, you’ve come to the right place. The SOA offers research, authored by an individual or a team of authors, for download in just a few clicks. Most reports are in PDF format.
2025
- Actuarial Intelligence Bulletin
- March
- This bulletin serves as a platform for sharing knowledge and fostering collaboration around artificial intelligence within the actuarial community. Explore articles on strategic initiatives, practical tips, and research advancements, all aimed at empowering actuaries to leverage AI responsibly and effectively.
- Operationalizing LLMs - A Guide for Actuaries
- January
- This guide covers understanding, evaluating, deploying, and managing risks of Large Language Models.
2024
- Impact of AI on Mortality - Essay Collection
- September
- The Society of Actuaries (SOA) Research Institute’s Mortality and Longevity Strategic Research Program Steering Committee issued a call for essays to explore the application of artificial intelligence (AI) to mortality and longevity. The objective was to gather a variety of perspectives and experiences on the use of AI in mortality modeling, forecasting and prediction to promote discussion and future research around this topic. The collection includes six essays that were accepted for publication from all submissions. Two essays were chosen for prizes based on their creativity, originality, and likelihood of further thought on the subject matter.
- Comparison of Regulatory Framework for Non-Discriminatory AI Usage in Insurance
- August
- This report aims to provide readers with an up-to-date overview of the regulatory landscape, but the information contained here is likely to change.
- Impact of AI on Retirement Professionals and Retirees - Essay Collection
- August
- The Society of Actuaries Aging and Retirement Strategic Research Program Steering Committee issued a call for essays to explore the impact of artificial intelligence (AI) and large language models (LLM) on retirement professionals and retirees. The objective was to gather a variety of perspectives and experiences with AI and LLM in different retirement settings—both now and in the future. It is the goal of this collection to spur thoughts for future research and set the stage for upcoming efforts.
- Risks Emerging from AI Widespread Use - Essay Collection
- August
- The Society of Actuaries Catastrophe & Climate Strategic Research Program Steering Committee issued a call for essays to gather Risks Emerging from Artificial Intelligence Widespread Use. The purpose was to gather potentially catastrophic uses of artificial intelligence (AI) as submitted by essay authors with the goal to initiate discussion regarding risks and impact of AI and set the stage for future research.
- A Primer on Generative AI for Actuaries
- February
- This paper offers a technical overview of the mechanics of Generative AI, with several actual examples throughout. It is intended to be a practical introduction to the concept, not a comprehensive view. Actuaries should consider this a point-in-time view of this rapidly evolving industry and use the issues presented here to spark conversation within organizations and across the profession.
- Federated Learning for Insurance Companies
- February
- Federated learning (FL) describes a distributed machine-learning framework enabling multiple devices or organizations to collaborate on a machine-learning model without having to share their raw data with each other or with a central server. One promising application of FL lies in the insurance industry, where each firm harvests a vast amount of client and claims data. There has been little to no previous literature on the applications of FL on insurance data. This paper aims to fill the gap and to offer researchers and practitioners an introduction, the pros and cons of FL, as well as potential use cases.
- Using Interpretable Machine Learning Methods: An Application to Health Insurance Fraud Detection
- January
- This project establishes a foundational framework for implementing interpretable machine learning techniques in the context of health insurance fraud detection. Machine learning algorithms excel at constructing intricate models by discerning patterns in data, yet the risk of overfitting to training data necessitates rigorous testing by modelers and users. While certain validation practices for linear models apply to machine learning, the challenge of interpretability remains pronounced. This report endeavors to enhance transparency and understanding in the realm of health insurance fraud detection through interpretable machine learning.
2023
- Generative AI – A Roundtable Discussion – October 2023 Update
- November
- On October 7, 2023, the SOA Research Institute assembled an industry expert panel to discuss current issues in GenAI. The group was diverse in terms of employment, including company actuaries from life, health and property/casualty backgrounds, as well as consultants from various kinds of firms. This document summarizes the discussion that occurred during the three-hour meeting.
- Data Challenges in Building a Facial Recognition Model and How to Mitigate Them
- January
- This paper is an introduction to AI technology designed for actuaries to understand how the technology works, the potential risks it could introduce, and how to mitigate risks. The author focuses on data bias as it is one of the main concerns of facial recognition technology.
2022
- Avoiding Unfair Bias in Insurance Applications of AI Models
- August
- Here’s a framework for avoiding or mitigating risks of unfair bias when developing insurance industry AI models, developed from interviews with industry leaders on the topic.