De-mystifying the use of artificial intelligence for insurers

There has been a lot of hyperbole about the impact artificial intelligence will have on the insurance industry over the next decade. Is AI a panacea for all the ills of a slow, paper-based system rife with inaccuracies, inefficiencies and inconsistencies/? Are we looking forward to a future of automated applications, interconnected devices and an AI-controlled claims cycle?

McKinsey’s report “Insurance 2030— The impact of AI on the future of insurance” paints the picture of a future where “insurance will shift from its current state of detect and repair to predict and prevent.”

To appreciate the value of AI for insurance, you have to consider how AI applications derive insights from raw data, and how these insights can be applied in the real world. You must consider why and how — and if — AI-controlled systems can outperform their human counterparts. To see the true potential of AI, you have to look behind the curtain.

Inside the black box

AI is a broad area of study covering many diverse technologies and with equally diverse applications. Artificial superintelligence (a system that can outperform a person in any area) and artificial general intelligence (a system that can match the capabilities of the human brain) are still very much the stuff of sci-fi movies and dystopian fantasies. Today’s technologies are characterized as Artificial Narrow Intelligence (ANI), which means it can match or exceed the capabilities of a person in a specific task, making it a perfect fit for select insurance applications.

Machine learning

To function effectively, ANI systems require a vast set of data, significant processing power, and advanced machine-learning algorithms. By looking for patterns and applying statistical analysis to data, the system “learns” how to interpret information and model behavior without being given specific instructions. Once the system has learned to interpret the data it can consume and analyze significant volumes of data faster and more accurately than any person. It can also uncover trends and make predictions that a human analyst would not see.

The insurance industry has always generated vast quantities of data, making it the perfect area to test and develop learning algorithms. With the growth of the Internet of Things (IoT), the volume of data is only going to increase, as will the sophistication of the insights available to those who can successfully analyze this information.

Sensors on vehicles, wearables and connected devices provide data to fine-tune actuarial calculations, predict the probability of a risk event, reduce its severity and even prevent the loss from occurring.

For example, several insurers and InsurTech vendors provide technology that plugs into an automobile, collects driving data and matches the data to known patterns to identify the driving behavior and compare it to more or less safe drivers. It’s in the early stage and not perfect by any means such as does the device know who is doing the driving?  However, it’s more data and which leads to better underwriting and coverage options.

Machine learning has also successfully been applied for fraud detection, lead generation and marketing optimization. It helps insurers to move away from a “one size fits all” approach to improve sales by offering personalized advice and product recommendations. It can identify policies that are about to lapse and propose the best strategy for retaining business.

Cognitive computing

Cognitive computing applications emulate human interaction by interpreting visual or auditory stimulus and responding in real time. A successful system must analyze, understand and generate natural language — taking complicating factors such as regional accents and emotional cues into account.

The most practical and frequently implemented application in this area is the chatbot, which can walk a customer through the application process, answer questions on products and manage straightforward claims. Many insurers have already automated parts of the claims management process, and this trend looks likely to continue.

Chatbots are available 24/7, cut down on redundant paperwork and free agents from repetitive tasks, allowing them to focus on more unique queries, improve customer satisfaction and streamline the claims process. In 2016, Jim (the chatbot used by Lemonade) reportedly settled a claim in three seconds without generating any paperwork.

Intelligent process automation

Robotics process automation involves constructing a robot with a single task that it can repeat endlessly, without making a mistake or growing weary. Robots built to this model have already been applied very successfully in manufacturing. The incorporation of AI allows for intelligent process automation. The system can now learn and adapt, making it possible for more and more robotic systems to operate in daily life.

Self-driving cars are predicted to massively reduce the risk of accidents and traffic congestion by eliminating human error. Enhanced surgical robots will become more widespread, 3D printing could be extended to the creation of buildings and all manner of tools, and autonomous drones look set to deliver more and more goods. Insurers will need to adapt to these new opportunities and develop new products to respond to changing risk profiles.

AI will positively disrupt the insurance industry

AI has the capacity to turn the vast quantities of data generated over the life of a policy into actionable insights. It can also transform the process of finding new customers, marketing products, optimizing customer service interactions, and improving underwriting decisions. Ultimately, AI has the potential to fundamentally change the business of insurance.

Fred Lizza (FLizza@stclaim.com) is CEO of StrategicClaim, a provider of a SaaS-based claims platform for carriers, agents and policyholders designed to expedite reporting and resolution of auto and homeowners insurance losses.