Artificial intelligence is proving to be a boon to insurance companies that seek to make the most out of the treasure trove of data they possess. Properly harnessed, the technology can benefit both insurers and customers through more accurate and personalised products and services.
In late 2019, major Japanese insurer Mitsui Sumitomo Insurance Co. revealed that it had partnered with Silicon Valley-based AI start-up dotData to create a digital platform known as MS1 Brain. Launched in February, MS1 Brain analyses customer data such as contract details and history, accident information, and lifestyle changes, to predict customer needs and offer the products and services that match those needs. The platform also helps generate more targeted customer communications, including personalised videos on products and services.
“MS&AD, the world’s fifth-largest property and casualty company, wanted to leverage digital transformation as a core strategy to help create richer, more rewarding customer experiences for Mitsui Sumitomo Insurance, their insurance arm,” Shinichiro Funabiki, senior executive officer, CIO, CISO, and CDO of MS&AD Insurance Group, told Insurance Business.
“The idea of leveraging data to provide customers more tailored product recommendations meant that machine learning and AI would be critical tools in the arsenal. In order to achieve their vision of creating their ‘MS1 Brain’ recommendation platform, MS&AD worked with dotData to deploy dotData’s AutoML 2.0 platform – dotData Enterprise – to quickly scale and build AI and ML models in days, instead of months.”
Read more: AI to reorganise Asia's insurance landscape
Meanwhile, Ryohei Fujimaki, PhD, founder and CEO of dotData, discussed the various insurance applications of AI, as well as advice for insurance firms that want to adopt such technologies.
“There are many use-cases for using AI and machine learning (ML) in the insurance industry. In many cases, immediate consideration may be for managing risk,” Fujimaki said.
“Fraud detection, assessing policy risk, and monitoring claims processing are typical applications. There are also areas that insurers have applied statistical analysis such as underwriting, which AI or ML can advance. In addition, it has become increasingly important to use these technologies for customer-facing benefits. For example, insurers can use AI and ML to optimise portfolios for clients and identify the right mix of product versus cost and benefit for the consumer. These can also function as recommendation engines for the insurance industry.”
However, insurers must keep in mind several risks and potential pitfalls when integrating AI capabilities into their operations.
“The biggest risk that insurers face in developing AI capabilities is resource constraints,” said Fujimaki. “Data science, the underlying practice at the heart of AI and machine learning, is resource-heavy, slow, and very time-consuming. It’s common for single AI projects to last months and never make it into production. Insurers must embrace automation to help accelerate development time and create better products. Platforms with AutoML 2.0 capabilities can help shorten development times from months to mere days by automating the feature engineering parts of data science that are often at the heart of AI project delays.”
Fujimaki added that it is also critical for insurers to keep an eye on transparency, especially due to stringent industry regulations in many markets.
“One of the potential risks of automation is the creation of models that are difficult – if not impossible – to understand and interpret,” he said. “This is especially critical for insurers that tend to be highly regulated and must comply with oversight. Any AI solution deployed by an insurer should have easy interpretability to make oversight and compliance easier.”
