AI Can Earn Trust If We Stop Treating Human Access as a Backup Plan

If you ask a P&C insurance executive whether AI will strengthen customer trust, 85 percent say yes. If you ask policyholders, only 40 percent agree, and almost as many think it will weaken trust. 

That 45-point gap was the first thing that jumped out at me when we fielded our recent research results. The headline is easy to read as "customers don't want AI," but that’s not what the data reveals. 

“Policyholders who use AI tend to like it.” This feels like a less exciting, but more accurate headline to me. The attitudes policyholders express about AI in the abstract look very different from the experiences reported by those who have actually used it. In fact, 91 percent of policyholders who have interacted with an insurance carrier’s AI have been satisfied with those interactions. Policyholders named speed, ease of use, and 24/7 availability as AI benefits. Only two percent said there was nothing positive about the experience. 

When asked what they need to trust AI communications from their insurer, respondents cited easy access to a human, human review of important decisions, and guarantees that the information is accurate as the top priorities. One respondent who struggled with an AI interaction put it plainly: "All of my interactions with AI never gave me any answer to my problem." Policyholders don’t have a problem with AI per se; they have an issue with interactions that don't resolve anything. 

My personal takeaway is that we won't close the trust gap by gatekeeping humans. We need to understand the customer’s specific need, the sentiment they are expressing and the stakes of the interaction, then route them to the person or AI agent best equipped to help. Human access shouldn’t happen only after AI fails; it should be built into the experience from the start.  

I think about all the times I’ve said “representative” somewhat loudly and clearly into an IVR only to be returned to a generic menu of options that didn’t fit my circumstances. Let’s never repeat that mistake.

The lesson from IVR shouldn't be that automation doesn't work. It's that automation designed primarily to keep customers away from people doesn't work. We have much better technology now, and much better information about what customers expect from it. It’s our responsibility to use both.