Description: Artificial intelligence (“AI”) adoption in the insurance industry is increasing. One known risk as adoption of AI increases is the potential for unfair bias. Central to understanding where and how unfair bias may occur in AI systems is defining what unfair bias means and what constitutes fairness.

Description: Many seniors who own houses in the United States have most of their wealth in their houses. Home Equity Conversion Mortgages are designed to allow seniors to borrow money against their house. This paper describes a stochastic simulation approach used to estimate the amount of a level-payment annuity payable as long as the senior is alive and living in his/her home. From Actuarial Research Clearing House 1990, VOL. 2.

Description: In this paper, a basic varying parameter regression model, where the regression coefficients vary over time, is employed to predict the future expected claims. It is shown that the resulting credibility formula for the prediction is of the updating type. The credibility formula performs well compared to the basic credibility formula of BÄuhlmann-Straub when applied to the Hachemeister data. From Actuarial Research Clearing House 2003, Vol 2.

Description: This paper described techniques used to keep existing policies up to date. One technique involved increasing the reserve interest rate of previously issued business to the level being used for new business as permitted by law. From Transactions of Society of Actuaries 1980, Vol. 32.

Description: View the Avoiding Unfair Bias in Insurance Applications of AI Models

Description: This paper presents the application of the asset share technique of pricing to Home Equity Line of Credit insurance. The paper's objective is to provide useful tools when developing a new HELOC product including projections for marketing and underwriting, distinctions between desirable and undesirable product design features, and development of premium rate structures. From Actuarial Research Clearing House 1990, VOL. 2

Description: This paper reviews the general methodology of capture-recapture procedures and describes some applications such as estimating the size of the undercount in censuses and estimating the number of duplicate records on in a database. From Actuarial Research Clearing House 2006.

Description: This report provides actuaries and other professionals working with risk, data and analytics or program evaluation in the health care space with considerations and frameworks related to the use of determinants of health (DOH) in data and actuarial work streams. Such frameworks can leverage new data sources to enrich health care data analysis to move beyond traditional data elements, such as age, gender, zip code or health diagnosis.

Description: This report documents the work of the UCT with respect to development of an algorithm used to score preferred risk class structures, producing Relative Risk Scores and Prevalences. It is built for both knockout and debit-credit scoring methodologies.

Description: This paper presents the recursive definition of the aggregate claims distribution which does not require the usual computation of the convolutions of the distribution of claim amounts. It is independent of the number of lives in the group. which makes it particularly useful for large groups. From Transactions of Society of Actuaries 1980, Vol. 32.