For decades, the insurance industry remained convinced that disruptive technology couldn’t touch it. Yes, a few startups would arrive to automate tedious administrative tasks, but that was welcome. Anything new would arrive in the form of assistance, not an existential threat.
In other words, very few people imagined a world where robots could handle the complexity of risk assessment.
Well, the robots are coming, and they’re not interested in mere paperwork. In fact, they’re not even interested in underwriting as we know it. There’s an exciting future in store for the insurance industry, one in which our entire understanding of how to assess risk will be flipped on its head. Forget historical data and statistical models. This new insurance landscape uses data from the internet, social media, news reports, IoT devices, and more.
Complacency breeds disruption, and the insurance industry has grown awfully comfy. But what exactly does artificial intelligence – “robots” – mean for the future of insurance, and why are they now necessary? Is there even an immediate need for alternatives to historical data?
Companies have new needs and the insurance industry is not equipped to meet them
Cytora, one of the insurance industry’s leading analytics providers, notes that there’s been a reversal in the percentage of tangible versus intangible assets held by S&P 500 companies. Twenty years ago, they held 75 percent physical assets and 25 percent in intangibles. Today, those figures are flipped leaving companies with liabilities they never had to account for. This brings a startling new demand for insurance products that cover events like cyber breaches, brand reputation, and non-damage business interruption. How do you assess such things?
Well, if you rely on historical data, it’s difficult. Make no mistake, the big insurance players have cobbled together packages for high-profile liabilities like cyber breaches, but they’ve left companies with unique insurance needs unsatisfied. So much so that these companies have turned to insuring themselves.
Aside from growing demand, there’s also simmering dissatisfaction. People are starting to ask, “Why should we rely on these traditional methods?”
No two businesses are the same which means their risk profiles are definitely not alike. Moreover, there’s a similar shift in what it means to do business. People run businesses out of their homes or over the internet. They sell software or information as opposed to a physical product. And with brand value so vital to success, a blow to a company’s reputation can be just as, if not more, damaging as stolen inventory or a factory fire.
How can historical models help us understand the risks of processes we’re only beginning to embrace?
The answer is they can’t, but the large amounts of external data we have can. As Oliver Ralph notes in the Financial Times, it’s not that the industry doesn’t recognize the importance of data – if any industry knows the value of data it’s the insurance industry. What it’s done is underestimate not only the amount of data out there, but the opportunity to use it. In fact, machine learning will have to replace statistical models since statistical models can’t process the amount of data we pump out each day.
Where will this external data come from?
Everywhere, and that’s not an exaggeration. The challenge will be cleaning this data up so we can use it to tell a coherent story. Ralph’s piece points out several examples of AI’s potential for insurance. Lapetus wants to use facial recognition and social media to determine how healthy you are. Aerobotics claims it can assess plant health with drone technology which has implications for crop insurance. The possibilities are as broad as the data that’s available.
One doesn’t even have to go as high tech as drones to understand data’s impact on insurance. Even small businesses with humble insurance needs stand ready to benefit from changes. Data analytics allows small businesses to find insurance packages tailored to their unique business. This way, they avoid the cost of overinsurance and the risk of underinsurance. And they continue to avoid this risk via digital insurance platforms that alert them when their risk profile (based on both internal data and external data) has changed, necessitating the need for more or less insurance. For example, when an online store starts selling into other countries, their existing Product Liability policy may not cover them. What is more scary is that the store-owner may not even know that the existing policy is no longer sufficient.
Should a sudden spike in poor restaurant reviews raise concerns about a potential lawsuit? After how many new hires do you need to bump up your coverage for employee liabilities? Does online chatter about a poorly performing product mean a recall is looming? With the right technology real-time, external data allows small business owners to answer these questions and make smarter decisions.
Artificial intelligence’s role in insurance decision-making
If all of this sounds exciting, it’s only the tip of the iceberg in terms of what AI can accomplish in the insurance industry. A 2016 PwC report, “AI in Insurance: Hype or reality?”, provides an overview of some of the fascinating ways AI will affect decision-making.
For instance, the customer experience heavily affects consumer choices. Machine learning techniques (i.e. decision tree analysis and Bayesian learning) will use data to predict how people behave in order to understand their needs, design personalized experiences, and present customized offers.
In terms of providing financial advice, machine learning and simulation modelling will use microeconomic factors (i.e., household finances) and macroeconomic factors (i.e. market conditions) to make recommendations and alterations, according to the report.
What are the implications for the humans of the insurance industry?
The property and casualty insurance industry in Canada employs 124,900 people across the country, according to the Insurance Bureau of Canada’s 2017 factbook. This includes brokers, support staff, underwriters, and more. However, this number does not include all the parallel industries that support the P&C industry, such as recruiters, real estate professionals, lawyers and many more service providers.
Right now, the employment conversation revolves around the reduction of support staff roles as digital assistants as insurance CRMs grow in popularity. Sooner or later, these advances in AI will start to affect individuals who perform core insurance functions. The question is whether these individuals will adapt their jobs to work alongside new technology or become obsolete altogether.
The robots are coming, and they’re ready to revolutionize an industry that desperately needs it.
Zensurance is Canada’s leading online commercial insurance broker. We offer a full range of insurance products to small businesses, with a particular focus on digitizing businesses and technology startups. We understand what it is to work with new technology, and know the most common risks of which you should be aware. Based on that (and a lot of analytics), we recommend the ideal insurance coverage for your business.