Time to chat about the bot: is the insurance sector ready for the AI transformation?

AI is here. What are we going to do about it?

“AI is clearly going to disrupt the industry, just as tech has done for banks and other related industries,” says Swiss Life Network CEO Michael Hansen.

He argues that the key lies in how we answer fundamental questions about the relationship between insurer and client: “How much do we actually want to know about ourselves – and how much should our insurer know? If AI can predict that an individual will die at 50, will the insurer cover them only up to that age?"

There is hesitation throughout the sector. According to a report by Capgemini, 62% of executives recognise that artificial intelligence and machine learning technology is improving underwriting quality and reducing fraud, but only 43% of underwriters trust and regularly accept automated recommendations from decision-support predictive analytics tools. A vast majority (83%) of property and casualty insurance executives believe predictive models are critical to the future of underwriting, but only 27% say their business has advanced capabilities.

The trust is still not there. The path to unlocking data-driven and actionable insights must begin with a secure data ecosystem.

The Capgemini report finds that 53% of policyholders are concerned about the amount of personal information collected by insurers. But almost two-thirds say they would be willing to share more data in exchange for transparency, financial benefits, and reassurance that their information is secure.

Talk to Skye: AI and data

Brand Engagement Network (BEN), NASDAQ: BNAI, a leading provider of customer-engagement AI technology through human-like avatars, is at the forefront of the transformation. BEN delivers pharmacy-driven patient engagement solutions such as the AI chatbot Skye, which assists patients taking the diabetes drug Metformin.

Skye presents an affable character and makes no claim to be all-knowing. Nor, her developers stress, does she hallucinate. Skye is loaded only with pharmaceutical specifications from the client drugmaker, the results of clinical trials, and data from relevant health organisations such as the American Diabetes Association. When the patient offers information or asks a question, she draws on a deliberately limited and carefully delineated set of data. “More than generative AI, we like to call it retrieval AI,” says Paul Chang, BEN's global president.

BEN's proprietary large language model enables the chatbot to understand questions and produce a human-like response. Head of business development Andy Sharma recalls how once during development of an application, a bot didn't know the answer to a certain question. After it was connected to the internet, the next time it had an answer. “This was impressive, but not what we want," he says. "Our AI does not need to know everything. We want knowledge to be narrow and deep, not wide and shallow.

BEN says its smaller-footprint AI models – they use a fraction of the parameters used by larger models such as ChatGPT – offer scalable and efficient solutions tailored to specific use cases. There could be Skye-like assistants for up to 20,000 drugs, as well as uses in the insurance sector, starting with the crunching of customer data as part of the sales process.“

We see this as an augmentation business, not a replacement business," Sharma says. "In a typical 90- minute call to sort out medical insurance, you let the avatar do the first 45 minutes, going through all the options available and collecting information to come up with picture of what makes best sense for the consumer. When the agent gets on the call, they can immediately have a meaningful, intelligent conversation.”

Will the data customers provide be reliable? Research suggests that people trust chatbots more than humans with sensitive information about health issues, for example. “The avatar communication experience has been shown to be positive when someone is in a state of anxiety - there is no judgment,” Sharma points out.

Beyond data gathering

Apart from taking the lead in managing customer data, AI offers various other avenues to be explored. In addition to lead generation, it can draw real-time data from wearable devices from providers such as Garmin and Apple to facilitate dynamic underwriting. This approach ensures fair underwriting for healthier individuals as well as businesses that are seeking to enhance employee wellbeing. AI can also assist in claims processing, providing timely and accurate information to both insurers and clients.

Says Hansen: “How much an individual is doing to stay healthy and what a company is doing to support their employees in staying healthy is generally not taken into consideration. Feeding the AI tool with such information would ensure fair underwriting for lives being lived well.”

According to Daniel Mark Olsen, CEO and partner at Vicorda, an employee health and well-being screening and monitoring solution provider, the application of AI to the proliferation of available data will benefit individuals, employers and insurance companies, as long as it is done transparently, and users retain control over their personal data.

“We see AI as a refining tool that can transform the multitude of inputs we receive through surveys, wearables and other data sources into unique user experiences and secure valuable data that make a difference for individuals, their workplaces, and for insurance and pension providers,” Olsen says.

“If users can see the value created by leveraging their data to provide individualised solutions, data-based advice and personal benefits, this will ensure a healthy ecosystem where the user is at the centre of the solution.”

Other aspects of insurance ripe for transformation include the manual review process, which is long and often frustrating for both clients and insurers. AI can streamline this process by ingesting rules engines and underwriting procedures, providing a single point of contact that offers efficiency and consistency. A client relationship manager (insurer or broker) could be replaced with an avatar. Today a customer needs to interact with various people in the insurance company depending on their situation or need. “One 'face' of the avatar could do it all,” Hansen says.
 

No going back

To develop a holistic approach in which AI offers tailored advice by processing personal data, the insurance industry must address several key concerns. AI systems must be transparent, reliable, and aligned with regulatory standards. BEN says its AI systems comply with industry standards such as the US Health Insurance Portability and Accountability Act (HIPAA) as well as offering a RememberMe feature, which allows customers to control whether or not their conversations are stored.

As Chang sees it, we can move forward safely, but staying in the same place is not an option. “AI tech is going to impact our daily lives, just as the internet has impacted our daily lives," he says. "We are applying AI in very specific ways to reduce the burden of mundane repetitive tasks. Not long ago, people collected road tolls inside small booths, working in the heat and the cold, surrounded by exhaust fumes. Tech has swept that away and nobody misses it. In 10 years, we might wonder how we ever managed without our AI assistant.”

Olsen has no doubt that AI will play a crucial role in both individual preventative health as well as strategic health at the organisation level. “This will provide the insurance and pension industry with the opportunity to move even further away from their historically reactive functions and roles and come directly into the living rooms of individual customers, while becoming an even stronger strategic partner for their corporate clients,” he says.

With the caution that is innate to the sector, Swiss Life is looking at the advantages that harnessing AI could bring to its operations. The possibility of partnering with BEN, Vicorda and other innovative companies in the field is very much on the agenda.