A new study from Hi Marley, a platform for property and casualty insurance, has found a significant gap between how insurers and policyholders view the impact of AI on customer relationships.
Eighty-five percent (85%) of insurance executives surveyed expect AI to strengthen customer trust, while just 40% of policyholders agree, with nearly as many believing AI will actually weaken the trust they place in their carrier.
The study surveyed 550 senior P&C insurance decision-makers and 2,000 policyholders across the US and Canada. Sixty-five percent (65%) of insurers said they expect to increase dedicated AI investment across a broad range of functions. Only about half of policyholders said they are mostly or completely confident their insurer will use AI responsibly.
Two-thirds of policyholders said their biggest concern is not AI itself but the fear of being unable to reach a human when it matters. Ninety percent said knowing when they are interacting with AI is very or extremely important to them.
That concern is task-specific. Roughly two-thirds of policyholders said they would trust AI-delivered answers to general insurance questions or coverage details. That trust drops to just 45% when it comes to information needed to file a claim correctly, a finding that points to a clear line in how policyholders are willing to accept AI assistance.
Stephanie Behnke, principal industry and market insights at Hi Marley and co-author of the report, said the key challenge for carriers is identifying where AI reduces friction without removing judgment and empathy.
"Carriers must identify where AI reduces friction without removing judgment and empathy. As carriers deploy AI-enabled communication, building and maintaining trust with customers depends on reassurance that policyholders will be able to access a human to help with important decisions, confirm the accuracy of information, and provide support in critical and emotional situations," Behnke said.
Despite the skepticism, policyholders who have used AI-supported channels to communicate with their insurer responded positively: 91% said they were satisfied with the experience. The report found that acceptance and comfort with AI-enabled communication depend heavily on where AI is used, how transparent it is, and whether a human is accessible when it matters most.
Dave Fernandez, chief claims officer at Kingstone Insurance, drew a clear boundary in the report.
"AI can support coverage decisions, liability determinations, payment decisions and negotiations, but it shouldn't be the decision-maker," Fernandez said.
Saurabh Saraf, chief digital officer at Amica Insurance, raised a concern that goes beyond trust: if AI adoption becomes universal across the industry, the technology itself stops being a differentiator.
"AI adoption is going to be high across the industry, and to some extent it levels the playing field. We have to make sure that doesn't turn into a race to the bottom, where talking to us feels no different than talking to any other carrier. The real question isn't whether AI helps, it's what else we do to stay differentiated once everyone has access to the same baseline capability," Saraf said.
Alex Burgess, vice president of industry strategy at Hi Marley and co-author of the report, said a successful AI interaction is defined by outcome rather than just accuracy.
"A successful AI interaction is not simply one that provides an answer, but one that helps the policyholder understand what to do next or moves the issue closer to resolution. More automation will not differentiate the customer experience; carriers who identify where AI can improve the exchanges, where human support is needed, and how to keep the conversation progressing will have an advantage," Burgess said.
For carriers scaling AI-enabled customer communication, the report's practical takeaway is sequencing: start with interactions where AI's role is clear and the consequences of errors are low, then expand as technology, governance and customer experience mature.
For brokers advising clients on carrier selection, the trust gap identified here is a live market differentiator. Carriers that can demonstrate transparent, human-backed AI deployment may find it easier to retain policyholders who are comfortable with AI assistance but unwilling to lose human access entirely, a distinction that is likely to matter more, not less, as AI becomes standard across the industry.
