Rise of the machines — and the humans they'll learn from
Increased use of data science, AI and machine learning are driving the skills needed by today's insurance pros.
There’s a tremendous amount of conversation in the insurance industry today about the impact of artificial intelligence (AI), machine learning and data science. When we think about these topics, we normally talk most about the changes affecting insurance products, consumer interactions, and the industry. Oddly, we talk a lot less about the implications for the professionals — the people who operate our businesses. As a “lifer” in the insurance business — 23 years in property & casualty preceded my current role as CEO of the Society of Actuaries (SOA) — I think about those people every day from the perspective of an organization that develops professionals for a key segment of the industry’s work.
What we see in the short- and long-term are major changes in the nature of the work we do, employee education and professional development along with corresponding changes needed in developing our talent.
While AI and machine learning are reaching nearly every category of insurance, I suspect it’s the property & casualty industry, and especially the personal lines — homeowners and automobile insurance — that are feeling the greatest impact first. Outside property & casualty, it’s probably group health insurance for the same reason… There’s a lot of repeated data.
Real-world applications
Take auto insurance. Millions of claims are flowing at any given time, for every fender bender that happens and with a very short-term contract, usually six months or a year at most. The same is true of homeowners insurance: Short-term contracts are filled with quick claims. Whenever hail hits a house, there’s a cost and a settlement. And that’s exactly what AI and machine learning depends on; a lot of rapid-fire, repeated data observations from which an algorithm can learn.
And so, we develop models, test them, put data against them, learn and adjust. We’ve heard from property & casualty insurers how that’s not only changing their business but their staffing patterns, too. They’re automating. But that takes people with new kinds of skills. This is true for claim estimation and claim settlement but also for underwriting. It’s increasingly rare, for instance, for a claim adjuster settling a hail claim to climb up on the roof. Now, drones are taking the photos, and machines are looking at them.
Machines also are reviewing photos for car claims, putting them in an algorithm, estimating costs, and settling claims.
On the group health side, very frequent payment of medical bills creates the same dynamic; repeated events in a short-term contract from which an algorithm can learn. It’s much easier for machines to learn in these types of insurance than in, say, life insurance, where decades often pass before each claim is made, giving the algorithm much less to learn from.
Technology and personnel
The global interconnectedness of risk — cybersecurity, climate change, pandemics — is another area where AI is affecting property & casualty insurance. Weather and viruses don’t respect borders. And the amount of data to be captured is staggering. For staffing and training, property & casualty insurers must be thinking about the implications this shift has for the types of people they will need to hire and the skills they need to encourage or develop.
The SOA identifies the skills needed by those who use actuarial talent, but the trends we’re seeing apply broadly to the insurance industry’s workforce and many other industries, too. The rise of data science, the increasing application of artificial intelligence and machine learning, and all their impact on professional work are driving the skills needed today. Of course, there is a strong and growing need for employees who have hard data analytics skills — programming, statistical analysis, database management, model construction, and interpretation of results.
Probably the key skill, however, is the need for judgment. After all, as Clayton Christensen, the great management theorist and developer of the theory of disruptive innovation noted last year, “Big data tends to gloss over or ignore anomalies unless it’s crafted carefully to surface these to humans… It’s only by exploring anomalies that we can develop a deeper understanding of causation.”
So, as we look at how we need to educate actuarial talent in the future, we’re also focused on how to help people apply judgment to the data, interpret findings, find causation, and draw insight from anomalies.
We’re also looking at the changing nature of how professionals in the industry will be educated and credentialed in the future. This means offering more ways for people to demonstrate and verify the knowledge they’ve gained through digital badges and micro-credentials. It also means offering new certificate programs, and it means partnering with employers in the industry itself to ensure we’re offering learning in the skills they need in ways that are attractive to consume.
At the SOA, we are moving ahead aggressively to identify and define the new skills that actuaries and other highly skilled, highly specialized, highly technical professionals can expect employers in the industry will be demanding in the next five to 10 years. We’ve identified the skills we think are going to be needed most, and we are rapidly changing our education system to make sure that people who come through our pathways have those skills.
Ways of working are changing, so our education and training are changing, too. The work of actuaries and other insurance professionals will be affected by those trends. On behalf of our members and all whom they serve, we are going to continue to offer leadership in education and credentialing for actuaries and to champion the future of the profession.
Greg Heidrich (gheidrich@soa.org) has served as chief executive officer of the Society of Actuaries (SOA) since 2007. He serves as chief executive for the organization and provides management oversight and direction to the full range of the Society’s activities.
These opinions are the author’s own.
Related: