Ethics

Ethical Use of AI for Actuaries

Actuaries understand quantitative modeling. It seems only natural that they would adopt Artificial Intelligence (AI) as an extension to their already extensive training and skill in predictive and probabilistic modeling. However, in addition to requiring more learning to be effective, AI imposes requirements that are very different from current analytical modeling. AI is fueled by data—and great volumes of it in a multitude of forms. In addition, because of its ability to both operate in an unattended mode as well as mask its internal logic, AI can pose ethical risks in its implementation.

The scope of this paper is to provide a technical overview of the tools and disciplines currently in AI as well as the forces at work that financial institutions such as insurance companies are using to modernize their analytical processes. The aim is to provide some guidance to the actuarial profession for understanding ethics in relation to using AI, recognizing the potential for doing harm and how to avoid it. The purpose of this report is to highlight the ethical risks arising from the application of AI in actuarial practice and to have tools to use to identify and manage it.

Practical Examples of the Impact of AI in Data Management

How application of AI provides access to data without the need for code. Embedded AI provides recommendations, not merely black-box solutions by understanding similarity across datasets. The paper describes explicitly what AI techniques apply to the different problems. Machine Learning, Recurrent Convolutional Neural Networks RCNN, Semi-Structured Data Parsing, Hidden Markov Model, Gene Sequencing algorithms and Knowledge Graph.

Solving Data Integration at Scale

Despite shiny new AI and data science tools, the problem of data integration at scale hasn't gone away. But promising new approaches from vendors like StreamSets and FlureeDB are worth a closer look.