Artificial intelligence (AI) is disrupting the insurance business. Though insurers are somewhat late to the party, compared with other industry groups, some have already implemented a broad range of applications for such AI staples as natural language processing (NLP), robotic process automation (RPA), machine learning (ML) and deep learning. I want to take a dive into three areas that are undergoing the greatest change as a result of this trend: customer experience, back-office processing, and compliance and risk management.
Insurers are applying data analytics to identify pain points in customer interactions and help make the customer-interaction process more fluid and personalized. For example, some insurers are cross-referencing call center recordings with chatbot data to gain new insight into customer sentiment and agent service quality.
The focus on personalization has already given rise to new types of insurance, such as the single-item policies offered by insurtech newcomers Slice and Trov. These policies allow customers to customize coverage all the way down to a single item, such as a camera or smartphone, or a single event, like short-term renter’s insurance for the length of a vacation stay.
Customers willing to share personal data may even earn discounts. Progressive, for example, offers discounts to drivers willing to provide real-life risk data collected by a car-mounted sensor.
Chatbots, virtual advisers and voice neural interfaces are being used to create a more seamless, automated and personalized enrollment experience. Instead of requiring applicants to wait to talk to an agent, provide information and sign documentation, insurers such as Lemonade are using machine learning and behavioral science to speed up the process, allowing customers to enroll for some types of insurance in as little as 90 seconds and get claims processed, via smartphone, in less than three minutes.
Back-Office Processing and Fraud Detection
Behind the scenes, insurers are applying AI to improve underwriting and using RPA to perform routine repetitive tasks, such as claims processing, freeing staff to focus on more strategic types of work.
Machine learning systems are being used in the back office to analyze photos of accident damage or scrape social media sites to see whether disability claimants are posting pictures of themselves skiing, for example. The ability to quickly identify all the connections between claimants and the doctors, attorneys, pharmacies, body shops, attorneys and other parties involved in the processing of a claim can reduce payouts of fraudulent claims dramatically. The same process can also help identify repeatable patterns in fraud activity across millions of claims. Combining this information with third-party data sources and leveraging predictive and cognitive capabilities can help reduce costs associated with fraudulent-claims investigations.
Compliance and Risk Management
Not all repetitive tasks suitable for automation are office processing tasks. AI has proven to be particularly well-suited for automating compliance, uncovering cyber risks, and improving predictive and risk analytics. Feedzai, an AI company, has positioned itself as the solution for efficient, automated know-your-customer (KYC) and onboarding processes, PCI-compliant credit card processing, and continuous fraud monitoring. Deep machine learning is increasingly viewed as the future of cyber defense systems, enabling them to evolve at the speed of change of cyber threats. The possibilities for managing fast-paced change in the risk landscape with AI truly appear to be endless.
There is a catch to all this, of course. To benefit from AI – especially when it depends on sophisticated big-data analysis – insurers must have made some foundational technology investments over the past few years to get control of their data. While some have made the digital leap, and others, like Lemonade, have a head start by virtue of being born digital, many others have aging and inflexible legacy systems that prevent them from accessing much of the valuable data they have accumulated.
It is crucial for insurers to create the right data architecture and a data lake – a large, flat pool of high-quality data used for large-scale analysis, as opposed to older, slower, hierarchical database technology – to benefit from the information they already possess. Without the massive, robust and clean sets of data required for AI, companies can’t even contemplate implementing these systems. In addition, insurers need to recruit the right talent to translate all that raw data into positive business outcomes and relevant experiences for customers. This could prove challenging, since the demand for data scientists now far outstrips the supply – but it is the direction in which insurers must direct their efforts and investments.
With new, nimble and digitally savvy competitors entering the insurance market in droves, customer experience and service will be a critical part of driving new sales and retaining customers, and AI applications are driving that trend. The good news is that companies burdened with legacy systems have broad and increasing access to third-party solutions and services that can help with the transition. With every new market development, I am increasingly convinced that AI will likely be one of the most important technologies of our time. This is why we have been tracking it as an emerging risk in our PreView series as well. Insurers need to develop AI capabilities now to be successful, or even remain competitive, in the future.