(Image credit: Adobe Stock.)
In 2017, the artificial intelligence (AI) buzz grew in the insurance industry and we saw another significant uptick in insurance carriers starting IoT initiatives. That gives credence to Market Research Future’s forecast that this category will reach $9 billion by 2022. It’s no surprise that car insurance is leading the pack, with health and life insurance following closely.
Early experiments have taught insurers that they lack the right data management capabilities to cope with all these new data streams. Their challenges are related to larger data volumes, the speed of the data, the complexity of the data, and also the need to turn this data into new insights that improve employee workflow and have a positive impact on customer experience.
In 2018, we should expect many well-known IT vendors—such as IBM and Microsoft—to make significant contributions to Insurance IoT and AI solutions, as we should for many smaller vendors that have introduced breakthrough advanced analytics, machine learning, data management and artificial intelligence. For example, BigML isn’t exactly a household name, but you don’t have to be a PhD to use its collection of scalable and proven algorithms thanks to an intuitive web interface and end-to-end automation. Check out this video from DIA Barcelona last year for some real-world examples.
Strategy Meets Action (SMA) discussed its recent research (“2017 Data and Analytics Trends: Is the Insurance Industry in for a Wild Ride?”) in a webinar outlining how 92 percent of insurers are investing in data and analytics initiatives. Furthermore, at the core of those initiatives is a Big Data platform based upon Hadoop (see figure 1).
As the insurance industry progresses, it’s interesting to ponder what we will see in the areas of machine learning, artificial intelligence and cognitive computing and how they will impact this industry. Each relies on big data’s foundational capabilities. In a recent McKinsey article, “Ask the AI experts: What are the applications of AI?,”members from Google, Baidu Research, Silicon Valley AI Lab, Arraiy, and Bosch Research and Technology Center, highlighted how enterprise systems will shape AI relative to everything from self-driving cars, forecasting weather to the kinds of crops to grow in certain places, during certain times. Each of these examples have implications for the insurance industry, especially impacting how we assess risks and investigate claims.
Adam Coates, from Baidu Research Silicon Valley AI Lab summarizes it best:
“AI is substantially driven right now by three critical pieces. One is data, another is computing power, and the third is talent. As much as the field is hot, there still are not enough engineers who know how to apply these machine-learning algorithms with a really high level of skill. It’s getting better, but it’s still a scarce talent.
“If you’re interested in solving AI problems for your business, then I think it’s important to think hard about whether you want to try to construct a machine-learning team within your company to solve a specific problem or whether you can now use enterprise platforms.”
While technologies continue to mature, an absolute constant is the continued growth of data along with its variety and velocity. We see investments and innovation in open source capabilities such as Apache Spark (machine learning, natural language processing) and Apache Zeppelin (data science platform) which are quickly becoming a foundational element, like Hadoop.
Many would refer to Hadoop as “plumbing”; certainly, to keep one’s house in order, one needs plumbing and we need it to interconnect and inter-operate—especially when we deal with highly complex data management and data volume issues. The insurance industry depends on data to operate, grow and protect customers. To support “preventive analytics” the data needs to move and be analyzed “freely.”