Postcard from InsiderTech Chicago

InsiderTech Chicago brought together people ranging from insurance industry veterans and novices to tech entrepreneurs and software engineers. They all saw a future strongly shaped by developments in artificial intelligence (AI) – a technological tool that can unlock great opportunities for the insurance industry.

AI can be applied to most parts of the insurance value chain but the greatest benefit is likely to be felt in areas such as claims management, risk selection and distribution.

The world’s increased use of technology also means an exponential growth in the amount of information that is available for a claim. 

AI can help the claims management process by aggregating this information and turning it into data points. It can bring together a wide array of sources and filter the data into value-add information at the point of decision, making the claims process, from first notice of loss to final claim payment, shorter and more efficient.

“If you're a claims adjuster, if all that information comes to you versus you having to Google search, that's a substantial powerful thing that AI can do for you,” said Harish Neelamana, co-founder of DataCubes, a commercial InsurTech that helps digitise submission data.

Another step where AI can add value is in distilling information into a specific response for the underwriter or decision-maker, helping them triage claims by severity, said Neelamana.

These benefits of AI were also supported by Jamie Yoder, president of Snapsheet, an InsurTech that provides claims management technology for personal and commercial insurers.

“Our ability to [get involved] as early as possible in the process allows you to triage [claims] and deal with [them], and get a response back to the customer as quickly as possible.”

Matteo Carbone, founder and director of the IoT Insurance Observatory, highlighted some of the benefits of telematics in the auto insurance industry’s claims management process. “An auto telematics value proposition focused on the claims monitoring has a terrific effect on the acquisition of low-risk clients,” he noted in his keynote address.

Carbone added that the technology also allows for better risk selection, as he found that even if two customers seem equal based on their characteristics, the one who accepts the telematics value proposition is 20 percent less risky.

Despite providing numerous advantages to the industry, the consensus was that AI must be approached with caution. The technology poses unique regulatory questions as legislators face an uphill battle trying to understand it while models become ever-more sophisticated.

“As we go forward, the regulatory bodies will get more and more comfortable with how machine learning is used and how we apply it to make sure we're not causing some of the social issues that we want to avoid,” said Sunit Shah, founder and CEO of Soteris, an InsurTech that delivers AI-based pricing software.

The bridge between regulators and big data specialists has come increasingly into focus. This has allowed a new wave of start-ups to arise which have identified the value in simplifying the characteristics of an input/output process at a time when models are becoming increasingly complex and difficult to explain.

One differentiator between AI models is ‘black box’ versus ‘white box’. In white box models, the underlying AI processes are fully transparent and explainable. 

On the other hand, black box models lack transparency as the algorithms remain hidden – the underlying code is contained within a ‘box’, which makes it impossible to explain why any result or prediction based on the models turned out the way it did.

“Part of the challenge with AI is black box and transparency in that technology, so getting regulators comfortable with it is inherent in whatever you're going to develop. It might start with commercial insurance where there's less regulatory oversight, but we'll get there,” said Kevin Leslie, vice president of marketing, insurance at Equifax.

At Embroker, COO Julie Zimmer believed that regulators should not accept pricing models and underwriting guidelines that employ mainly black boxed AI, as their role is to keep the industry honest and protect consumers.

Selected quotes from InsiderTech

"Machine learning algorithms can look at hundreds of thousands of different variables between millions of customers, can find generalities between customers with a high amount of claims, and compare those customers to new customers in just one second."

Anil Celik, co-founder and CEO, UrbanStat

Most of the sexy stuff in InsurTech are all the ‘homeruns’ the companies are trying to hit but there’s a lot of value in just getting singles and doubles. There’s a lot of friction in the insurance process that can be ironed out, automated.

Anand Dhillon, co-founder and CTO, Cover

I don’t believe anybody is completely uninsurable. I believe that the tables are going to be turned because we will learn a lot about what would traditionally be considered uninsurable, what goes into those definitions. I don’t consider it cherry picking, I consider it pricing and market positioning.

Julie Zimmer, COO, Embroker

Related articles