At Amazon’s prototype grocery store, Amazon Go, customers can walk in, pick up what they want, and walk out without ever waiting in a checkout line or pulling out a wallet. Amazon will automatically charge their account and send them a receipt. Run out of paper towels? No problem. Amazon Prime customers can place an order from their phone and get same-day delivery.
These are the kinds of no-hassle experiences that consumers have come to expect — and artificial intelligence (AI) is powering many of them. Amazon Go, for instance, uses computer vision, fusion sensors, and deep learning to track when items are removed from or put back on shelves.
Unlike Amazon, and many other product and service providers, insurance companies have fewer opportunities to interact with customers, and those opportunities are less rewarding. Insurance is traditionally a “selling” process, not a “shopping” process, and claims are the necessary outcome of an unfortunate event. For insurers looking to turn the tables on a less-than-ideal dynamic, artificial intelligence can transform both the customer experience and the claims process.
Chatbots have the answers
What problems can AI solve for insurers? To improve their customer experience, many are investing in chatbots powered by natural language processing. Chatbots are fast and efficient, and customers can interact with them in the way they are most comfortable communicating — via whatever mobile device they have in their pocket. And, unlike humans, chatbots can help more than one customer at a time and are available around the clock. But they are not strictly a utility play. When they are designed to have personalities that align with the brand powering them, they move from being simply transactional to transforming the customer experience.
In a recent Accenture study of the insurance industry, 68 percent of respondents said their companies use some sort of AI-powered virtual assistant in at least one segment of their business. Geico’s virtual assistant, Kate, for example, answers basic policy and billing questions within an app. Digital insurer Lemonade takes things a step further. Their chatbot, Maya, sells inexpensive homeowners’ and renters’ insurance, and their claims bot, AI Jim, makes Amazon Prime’s same-day delivery look slow — it recently settled a simple claim in three seconds.
As chatbots become more commonplace, they are making their way into behind-the-scenes claims processes as well. Tableau’s prototype chat software, Eviza, has a voice interface so users can drill into its signature data visualizations simply by asking questions out loud. Clara Analytics offers askClara, a chatbot the company bills as a “24/7 personal assistant to the claims handler.” Like customer-facing chatbots, it can answer routine questions about a given set of claims.
Machine learning makes sense of data
Insurance companies are sitting on a trove of the one thing AI requires to be successful — data. And AI technologies like machine learning have the ability to make that data actionable. Machine learning can look at data in a number of different ways. It can rank information, putting what it thinks you are looking for at the top of a list; classify information like images; make recommendations; and associate something with a numerical value. It can also group similar things together and detect anomalies.
For example, by reviewing data from closed claims, machine learning algorithms can identify both straightforward claims for automatic processing and complex claims that are more likely to require human intervention. By identifying commonalities in closed claims that resulted in litigation, it could predict which new claims might take a similar path and recommend preventative measures. Anomaly detection plays a big role in identifying fraud of all types. It could, for instance, be used to flag abnormal pharmacy prescribing patterns and alert an adjuster that some kind of clinical review might be necessary. The possibilities are limitless.
Tractable, a U.K.-based AI technology company, is putting machine learning to use in an interesting way. Armed with a database of hundreds of thousands of photos, it is using computer vision and machine learning algorithms to assess images of damaged vehicles so auto collision insurers can determine whether to repair or replace a part.
The road to AI adoption
Insurers are eager to adopt AI technologies — a recent report by Tata Consultancy Services estimates insurance companies will each spend an average of $90 million dollars on AI by 2020 — but implementing it may not be straightforward. Claims are seldom as simple as the one Lemonade’s AI Jim processed in three seconds. The claims process can be complex and highly regulated, and for mature companies, it is often powered by traditional, less flexible, technologies.
For each point along the claims process, insurers must determine what type of AI could bring improvements, whether their existing systems can be adapted to incorporate it, and if those improvements are worth the investment. They need to identify partners, hire or train for new skill sets, and put new development processes and infrastructure in place.
The good news is that AI does not have to be tackled all at once, and pilot projects do not have to be comprehensive. There are plenty of places to start, and getting started is key to laying the foundation for future innovation.
Alex Sun is the CEO and President of Mitchell International, a leading provider of technology and solutions in the insurance industry.