Description: Data science is being used to create more algorithms used across a variety of applications, including underwriting and pricing. As data science gains traction, it is imperative that we as actuaries consider the ethical implications of use of these algorithms. This session will discuss techniques for evaluating whether your algorithms are inadvertently making biased decisions and considerations for ensuring that your algorithms minimize risk of making biased decisions. Learn to identify potential ethical issues with their algorithms and techniques to address them.

Description: "Explore a general introduction to the opportunity of applying a relatively newer area in machine learning called uplift modeling, which has been successfully used in retail (Wayfair etc). The concepts are parallel to the intervention programs in health care, where we are interested in measuring the treatment effects and ultimately directing the right members to the most effective programs, driving the highest value for the programs both from a financial and health outcomes perspective. Uplift modeling can also be viewed in context of causal modeling where we are interested in assigning cause to a treatment relative to a member not receiving the treatment. Check out an introduction on the concept, seehow it relates and builds on existing work, and use a demo to showcase the potential of integrating this ML approach to current work."

Description: Explore insights to non-traditional actuarial roles from a panelist with experiences in data science, finance and entertainment. The panelist will be sharing their perspectives on their non-traditional actuarial role in relation to the actuarial profession.

Description: In this video Jing Lang emphasizes the importance of ethical considerations in AI utilization, particularly focusing on privacy, bias, transparency, and creativity. It highlights the need for collaboration among various stakeholders to address these concerns through regulations, diversity promotion, and ongoing discussions. Sources of bias in AI, such as underrepresented datasets and historical biases, are identified along with strategies to mitigate bias, including careful assessment of inputs and retaining algorithms. Contributors: Jing Lang, FSA, FCIA, MAAA FLMI; Scott Allen, FSA, MAAAA, Henry Chen FSA, FCIA, MAAA; Jon Forster ASA, MAAA

Description: Dana M. Peterson, chief economist for the Conference Board, has established herself as a visionary, and the go to expert providing insightful analysis and predictions to corporations, associations, and government organizations looking for clarity and foresight into how economic drivers might impact their businesses in the short- and long-term. With over two decades of experience in the field, she serves as the globally recognized think tank’s public face and primary spokesperson on all things related to the economy.

Description: This session will examine bias from a social science perspective. It will discuss the three types of systemic biases using the National Institute of Standards and Technology (NIST) model. The NIST model details the three types as historical bias, societal bias, and institutionalbias. The session will also provide several examples of each type to provide a framework for identifying the social structure of bias.

Description: The rapid evolution of life insurance underwriting has created pricing and operational challenges and raised new questions about established approaches to assumption development, both of which demand a fresh perspective. As underwriting methods evolve, we find ourselves navigating uncharted territory. In this session, we delve into accelerated underwriting, focusing on two critical areas. First, we explore assumption setting in this new world, where numerous interaction effects can make it feel like we’re asked to consider everything, from every angle, all at once. Considerations for assumption setting must encompass early performance indicators—what signals inform our assumptions? Additionally, we address novel underwriting approaches and how to adapt assumptions for these paradigms. Furthermore, in the context of post-COVID claims experience, we discuss distinguishing meaningful insights from noise. Second, we examine declinable risks and their impact on mortality. In this new world ushered in by accelerated underwriting, previously excluded risks—low-frequency, high-severity declinable cases—are now part of the landscape. Quantifying their mortality impact challenges us, as historical data lacks systematic tracking. To quantify declinable risks, we draw upon research findings from analyzing 27 million life insurance applicants, providing valuable insights for attendees to navigate this dynamic landscape and make informed assumptions for pricing and risk management.

Description: Actuarial concepts of credibility can be extended to encompass additional information provided by new machine learning estimates. New statistical ideas also indicate how to generalize to blend arbitrarily many estimators, while not double-counting common information. For example, historical claim costs and risk scores both leverage diagnosis code information. We introduce two new estimators, one based on partial correlation and the other based on the machine learning-derived Shapley statistic, to address these limitations.

Description: In this session, we will do a deep dive into the potential of large language models; we will explore how new AI tools can be combined with traditional modeling to automate aspects of risk assessment. Industry experts focused on operationalizing this technology will talk through their approaches, lessons learned and demo use cases. We will also share insights around creating a robust AI strategy based on business objectives and regulatory considerations. Learning Outcomes: • Applications of large language models in risk assessment use cases • Regulatory considerations around the use of generative AI

Description: How are regulation and innovation intertwined? How to regulate artificial intelligence/machine learning in the insurance industry? We will cover how regulators are striking the balance between protection of their constituents and supporting innovation. We will zoom into the case of predictive analytics and the regulatory attempt at avoiding undue discrimination. Let's discuss what this really mean.