Photo: iStock.com/Martin Barraud
The proliferation of big data, predictive analytics and artificial intelligence (AI) is a gold mine of opportunity for innovation. However, deployment of cutting-edge technology also opens the door to new forms of legal, ethical and social risk. The potential for harm is ever present, as illustrated by the title (and contents) of Weapons of Math Destruction by Cathy O’Neil. And although there have been numerous laws and regulations enacted to protect the public, actuarial responsibilities go beyond meeting the letter of the law.
It is important for the actuarial profession, insurers, data scientists, regulators and the public to have a set of ethical and responsible data use and model construction guidelines. These guidelines provide greater assurance that appropriate procedures are used for data acquisition and manipulation, and for the building, use and maintenance of predictive models.
The Society of Actuaries (SOA) recognizes the need for a rigorous learning and certification process to ensure the drive for innovation is balanced with appropriate behavior. Therefore, it created a first-of-its-kind Ethical and Responsible Use of Data and Predictive Models Certificate Program.
Through seven self-guided modules, three instructor-led webinars and a graded, take-home final assessment, participants will gain a deep understanding of:
- An ethical framework based on fairness, safety, transparency and accountability
- The ethical issues around using data and the types of questions to ask to mitigate the risk of using prohibited or high-risk data or combinations of data
- The inherent bias in data and the techniques for mitigating associated risks
- How the use of predictive analytics and AI creates professional and reputational risks different from traditional actuarial model building
- The importance of model explainability and the difficulties associated with models that are viewed as black boxes
- How teamwork and communication are critical for model development and deployment to reduce the risk of unintended biases and unanticipated outcomes
- How model governance can be used to achieve the above
A past program participant, Daron Yates, FSA, MAAA, of Foresight Solutions, says, “Noninsurance examples helped tremendously in understanding the broader ethical issues while the insurance illustrations made it more relevant to actuaries.”
This five-month certificate program is designed for actuaries and data scientists who build and deploy models, actuaries who supervise or sign off on the work of modelers and regulators who must ensure that models are appropriately used. This program does not assume a knowledge of predictive analytics in particular, but it does require a familiarity with financial modeling and working with actuarial data. If you are interested in this certificate program, learn more on the SOA website or reach out to firstname.lastname@example.org with any inquiries.
Copyright © 2022 by the Society of Actuaries, Schaumburg, Illinois.