Bill Gates, co-founder of Microsoft, said, “The advance of technology is based on making it fit in so that you don’t really even notice it—so it’s part of everyday life.” By this measure, the technology that actuaries use has not succeeded. We notice it all the time. For example, how often is a project late because the model took longer than anticipated? In this issue of The Actuary, we explore what can be done to make technology fit in so that it is not noticed. But first, let us start with some historical context.
The evolution of actuarial practice over the past 40 years has been revolutionary. Teams of actuaries now build, implement and maintain extremely sophisticated models using technology that does things we only could have dreamed of in the 1980s. We work on requirements that did not exist—and were totally unanticipated—as well as in newly created disciplines. Sometimes we try to do this on our own, but increasingly we are working with software engineers and other professionals. What problems do we face in addressing the resultant challenges, how can we summarize our accomplishments, and where do we go from here?
As models started to play a larger and larger role in our professional lives, there were those who predicted the shrinkage of our profession. After all, the computer could do everything we used to do. How wrong they were, as our profession continued to grow at a rapid clip. The ability to understand what needs to be done and analyze the results, together with the increasing needs and requirements, has required more and more actuaries. Today, some say artificial intelligence (AI) and machine learning (ML) will replace actuaries. I believe they will likewise be proven wrong as more actuaries than ever will be required to understand what can be done, assure this work is done effectively, and interpret the results. The machine will never do all of this on its own.
What then is the challenge? Well, as models have become more sophisticated and requirements more detailed, many actuaries have become caught up in the minutiae of doing their existing jobs correctly. Often this is necessary to produce an accurate, reliable and replicable result for the audience. But this has come at the expense of communicating our findings effectively, making our processes repeatable and producible, and being ahead of the curve on the next big thing that will truly add value that we should be working on. In other words, we need to do a better job at maintaining the relevance and sustainability of our profession and work products.
Oftentimes when we hear of these types of challenges, we are told about all of the soft skills we need to improve on—and there’s no doubt that we, as a profession, need to improve our communication and collaboration skills. But in this issue of The Actuary, we show concrete steps—hard skills—we can take to make us more relevant and more sustainable in the face of the emerging challenges of the 2020s.
The article by Tom Peplow, MSc, stresses the need for actuaries to work closely with other professionals (e.g., software engineers). Different groups of professionals can all learn from each other and develop a much broader knowledge base. Actuaries have realized that combining our skills with those of software engineers results in models that are more dependable, more reliable, better controlled and more flexible—models that are more likely to stand the test of time. Rich Lauria, FSA, CFA, MAAA, shares the history of ERM and how it has affected and been impacted by model development—for example, how regulations have evolved and how models have kept pace. Andy Smith, FSA, MAAA, writes about the numerous challenges with building models today and some potential solutions. These three articles position us well to respond to the challenges of model development, model maintenance, version control and model updates during this time of rapidly changing technology. Perhaps most important, an actuary gets to see how other actuaries have successfully responded to many of the challenges being faced today.
Then we come to today’s new frontier: ML and AI. The importance of and need for a multidisciplinary team to tackle the challenges in this space cannot be emphasized enough. With a properly assembled and managed team, the whole is much greater than the sum of its parts. If you want the diamond team, think about how diamonds are formed: You put a lump of graphite under enormous pressure for a long time, and then you have diamonds. Well, in ML and AL, you need domain expertise, technical proficiency, data science expertise, strong interpretive skills and strong communications skills. Taken together, you have the diamond team.
But actuaries may wonder how they can put all of these skills to practical use. Dave Czernicki, FSA, MAAA, and his coauthors share many of the use cases in which actuaries have an interest today. These use cases can all be practically implemented with the right team and resources. Yet the advantages of AI and ML go even further. In the article that Adam Haber and I wrote, we show an innovative way to get a much better handle on post-level term mortality and how to interactively model the shock lapse, mortality and post-level term premium jump.
The techniques described in this issue of The Actuary can bring big strategic advantages to those who successfully implement them. We wish you the best as you read these articles and assure you that the authors would be pleased to answer any questions you may have. We hope you agree that successful implementation of the ideas presented will make the technology fit in, so you don’t notice it.
Statements and opinions expressed herein are those of the author and are not necessarily those of the Society of Actuaries.
Copyright © 2020 by the Society of Actuaries, Schaumburg, Illinois.