(Image source: Frank Neugebauer/Genpact.)
Chatbots have evolved to serve underwriters using a combination of artificial intelligence (AI), machine learning (ML), and computational understanding (a fancy way to say reading and writing like people do). This means that insurers can leverage advanced technologies, along with their human workforce, to make underwriting an effective mix of humans and technology. But, as with any innovation, there are risks and barriers that insurers must overcome to fully benefit from this technology.
How Chatbots Talk
Understanding how chatbots can help humans be more productive is paramount to turn conceptual ideas into reality. Equally important is comprehending the operational aspects of chatbot functionality.
The underlying “engine” of a chatbot is a piece of technology that understands how to reason—specifically, reasoning that mimics human reasoning. How does a chatbot do this? By understanding significant numbers of algorithmic patterns.
Take Apple Siri for instance: it can respond to millions of different sentence structures across different languages. For chatbots, it means understanding linguistic patterns and replying with appropriate responses. The chatbot is only as good as what it understands from its human teacher—for this technology, input equals output.
What This Means for Insurers
Insurers can use chatbots to complement the work that humans are already doing. For instance, chatbots answer frequently asked questions that are time-consuming for humans—thus freeing up their time for more value-added tasks.
There’s also an opportunity for cross-selling and up-selling. This type of action requires the chatbot’s AI engine to meet complex conditions. Once the conditions are met, the chatbot provides suggestions to the underwriter. For instance, if a customer adds a vehicle to their policy, humans can train the chatbot to check if they have homeowners’ insurance. If not, it prompts a suggestion. Make no mistake, most humans do this well; however, humans are not infallible—they make errors. The chatbot ensures underwriters don’t miss opportunities like this and creates “omni-awareness,” that is, awareness of many things concurrently. Chatbots don’t get distracted by e-mail, phone calls, or text messages. More importantly, underwriters can focus on risk and client service.
Where’s the Risk?
While some companies are experimenting with chatbots providing the quote function—and potentially other underwriting functions—to customers directly, there are significant risks in doing so. The problem is customers have key characteristics that are (hopefully temporary) roadblocks to widespread chatbot adoption. These include:
- Customers don’t understand how a chatbot works, and a company can’t train its customers to understand.
- Customers are unforgiving—if a human gets something wrong, customers blame the human. If a chatbot gets something wrong, the customer blames the company—which creates reputational risks.
- Insurance is a hard use-case—there are virtually infinite intents a customer can have with a chatbot. This makes it challenging to train, which increases the likelihood the chatbot will get the information incorrect.
Chatbots have come a long way and continue to evolve rapidly. For now, it’s ideal to start with a more amenable audience and keep chatbots internal, as customers are unpredictable and untrainable. There are a range of challenges customers can bring to a chatbot, making responses a bit of a hurdle. That said, there are many benefits of a robust chatbot program, including cross and up-selling, and freeing up human talent to complete more complex tasks.