Public Policy and Actuarial Science in the Age of AI
The Institutes Griffith Foundation sought the insight and expertise of Kevin Ahlgrim, PhD, on the impact of AI on the actuarial profession and on the public policy questions that have come to fore. Below please find the questions we posed and Dr. Ahlgrim’s responses.
Why are actuaries important in the insurance industry, and how is AI changing their work?
Actuaries are indispensable in the insurance industry because they assess and manage financial risks. Artificial Intelligence (AI) is revolutionizing actuarial work by increasing efficiency and accuracy. AI systems and tools are being used to analyze massive datasets and facilitate real-time, dynamic pricing with greater accuracy.
How do actuaries traditionally use data in their work, and what's an example of this?
Actuaries traditionally use historical data to perform tasks like calculating rates or establishing reserves. For example, in life insurance, premiums are determined by reviewing recent mortality experience, which is often analyzed by factors such as age, gender, tobacco use, or occupation. This data has, historically, been structured to differentiate premiums based on observable variables.
What are two main ways AI is changing actuarial work, and what public policy issues do these changes raise?
AI is revolutionizing actuarial work in two main ways:
Explosion of Data: AI enables actuaries to analyze massive and diverse datasets, including unstructured data like social media posts and videos, which were previously difficult to analyze.
Dynamic Risk Assessment: AI facilitates real-time, dynamic pricing with greater accuracy and improves understanding of emerging risks.
These changes raise public policy questions related to data protection, cybersecurity threats, potential discriminatory practices, and the lack of transparency in AI models (often referred to as "black boxes").
How is the explosion of data benefiting actuarial work, and what concerns does it raise?
The explosion of data benefits actuarial work by providing vast amounts of datasets for analysis, leading to new insights on risk. For example, telematics in auto insurance provides volumes of new data on drivers' habits, such as cornering speed, acceleration, and device distractions. However, this raises concerns about data protection and cybersecurity threats. Public policymakers are considering questions related to appropriate collection of personal data by insurers and questions regarding necessary disclosures to consumers about company practices.
What advantages does dynamic risk assessment offer, and what potential issues should policymakers be aware of?
Dynamic risk assessment offers advantages such as real-time, dynamic pricing with greater accuracy and improved understanding of emerging risks. For example, climate change is rapidly altering property risks not just for coastal property, but also farmland and properties exposed to wildfire risk. In a dynamic environment, predictive modeling techniques are explicitly designed to adapt to new information quickly and improve risk assessments. However, policymakers should be aware that AI systems may yield conclusions that are not easily interpreted or may inadvertently lead to discriminatory practices. This can happen due to embedded bias in historical data or specific AI techniques. Unlike traditional actuarial approaches, AI models often lack transparency, raising concerns about equity and fairness.
Is a balance between innovation and consumer safeguards important when addressing the use of AI in insurance-related actuarial work?
Yes, AI is rapidly affecting many industries and occupations so, striking a balance between promoting innovation and efficiency while safeguarding consumers is essential to maintain a stable and equitable insurance market in the evolving landscape of AI-driven actuarial work. Policymakers must keep abreast of both the benefits and risks of AI systems in actuarial practice to achieve this balance.
Where can public policymakers and others go to learn more about AI's impact on actuarial work?
Those seeking further information about AI's impact on actuarial work can explore resources from actuarial organizations like the Society of Actuaries (SOA), the Casualty Actuarial Society (CAS), and the American Academy of Actuaries (AAA). The SOA and CAS are the two primary organizations in the US responsible for actuarial education, including credentialing, continuing education, and research. The American Academy of Actuaries (AAA) relies on CAS and SOA members to help provide advice and guidance to the public on matters related to financial security and risk.
The web sites for these actuarial organizations include much discussion on AI. In addition, each organization has publicly available publications whose recent issues have been focused on the impact of AI on actuarial work. Here are some specific resources:
The Actuary Magazine (SOA). In addition, the SOA just recently published a collection of six essays entitled “Risks Emerging from AI Widespread Use”. Three of these risks are related to AI related discrimination.
Also, the NAIC Working Group on “Big Data and AI” may be a useful source of information. As part of their work, the NAIC has adopted a Model Bulletin which outlines principles for insurers to use artificial intelligence (AI) responsibly.
Kevin Ahlgrim is the current Chair of the Department of Finance, Insurance and Law at Illinois State University. Prior to his academic career, Kevin worked as a health actuary. He is an Associate of the Society of Actuaries and a Member of the American Academy of Actuaries. His research has appeared in the Geneva Papers on Risk and Insurance Theory and Risk Management and Insurance Review.
