Twelve percent. That is the share of 1,028 high-earning Americans who told us in September they would let an AI assistant complete an insurance purchase on their behalf. We commissioned a survey because, quite frankly, we were curious.
For the past year we have noticed a larger number of prospects who have already done preliminary research with AI, most notably ChatGPT. To a certain degree, they understand what disability insurance is (a notoriously complex product) and what it costs before the first discussion. So, how far would they go with AI? Would they let a machine actually buy it on their behalf, or do they still prefer human involvement? The answer will shape how financial services firms do business.
Centiment fielded the survey for us from September 11 to 17, screened to employed adults in households earning $100,000 or more, with 340 respondents individually earning $200,000 or more. The full results, every crosstab and the question wording, are on the study page.
The line sits somewhere in the middle
The willingness to allow AI to act on one's behalf unsurprisingly declines the farther into the process you go. A large majority (74%) would let AI research their coverage options, and 66% would allow it to take the action of obtaining and comparing real quotes from different insurance companies. But only 26% would let it choose the company, and a meager 12% would let it complete the purchase. So there are clearly still hard limits to how much someone is willing to trust an algorithm with their money and financial protection.
When we asked who they would trust more for an honest answer about insurance, 49% picked the insurance agent and 22% picked the AI. Another 24% said both equally, and 5% said neither (ouch). I suspected that the agent would still win, but the margin was slightly larger than I would have bet, perhaps giving human expertise a fighting chance moving forward.
That said, 45% of respondents (the AI and both-equally answers combined, before rounding) trust an AI at least as much as an agent. Among the 77% who say they use AI for financial decisions, it is 55%. Technology workers are the highest of the industry groups at 65% (n=192), and they happen to be a fast-growing part of our client base.
My primary takeaway is that the AI will increasingly make the first impression, explaining how the coverage works, giving ballpark figures, and answering questions. But then the buyer still wants to go find a human they can ask whether it is right and hold accountable. This lines up with our experience over the past year.
The line is moving based on age and occupation
19% of respondents aged 25 to 34 (n=160) would let an AI assistant buy insurance on their behalf, against 4% of those 55 to 64 (n=183), more than a four-to-one ratio. The same two groups sit at 54% and 33% on trusting an AI at least as much as an insurance agent, and at 80% and 42% on having asked an AI a financial question in the past year. Younger generations are more accustomed to AI being a part of their lives, and therefore more trusting of its suggestions.
Occupation also has an impact. Technology workers (n=192) are the most comfortable with AI at every step. 80% have asked an AI a financial question in the past year, 65% trust it at least as much as an agent, 64% would let it take at least one transactional step (the application, the company choice, or the purchase), and 25% would let it fully complete the process, which is double the sample as a whole. Finance and accounting (n=108) comes next on the transaction question at 53%, though only 39% trust an AI at least as much as an agent.
One result was counter-intuitive. Healthcare (n=165) was the least willing of any industry to allow a transactional step at 32%, yet 11% would still let it complete the purchase, close to the sample as a whole. That matters to us, since physicians and other clinicians have been our core clients for fifteen years, and they are the slowest group to let an AI transact. The legal group was too small to lean on (n=34), but had the highest result of any industry for using AI as the first stop for a major financial decision at 29%, while also being the least likely to allow the AI to buy on their behalf at 3%.
Technology workers are also the best covered. Only 24% have no disability coverage at all, against 42% in other professional services and 45% among people working outside the listed industries. Healthcare, the field we have served longest, came in at 29% with no coverage.
What the same people believe about their own income
We used the back half of the survey to ask about our core product, and per usual, the answers reveal a lack of understanding regarding financial protection. 31% of those earning $200,000 or more have no disability coverage of any kind. Of those who do, most have a group plan through their employer, and those plans do not inspire confidence. Across the 546 respondents with employer coverage and some form of variable pay, 26% believe the plan replaces that pay and another 15% do not know.
When asked which is their largest financial asset, only 28% of those surveyed said their remaining career earnings, vs 39% for retirement accounts and 32% for home equity. As an illustration, a 35-year-old earning $200,000 with 30 working years ahead has an estimated $6 million of future career earnings, not counting raises or taxes. You'd need one heck of a 401(k) or piece of property to compete with that.
What I would do with this if I were buying
Spend an hour with ChatGPT. It already knows more about me than I probably realize. It's unfathomably knowledgeable and is only getting better. Everyone already knows this. 62% of the people we surveyed have already asked an AI a financial question in the past year. I would ask it how a disability insurance policy works, why I need it, what it costs, and then I would ask who it recommends I contact for official quotes (hopefully it says Seaworthy Insurance).
Then I would begin working with a licensed human being, who should be able to confirm my independent research and provide some accountability for any decisions made. Many of our clients are taking this path, and for the time being, it appears to be the best one. We plan to run the same questions again next September and publish what moved.
Frequently asked questions
Should you use ChatGPT to research disability insurance before talking to an advisor?
Which occupations are most willing to let an AI buy insurance for them?
Do younger buyers trust AI more than an insurance agent?
More from insights
- Storefronts for Machines To our knowledge, Seaworthy Insurance is the first disability insurance brokerage to expose an agent-callable quote action over MCP. What an AI assistant can do in insurance, what it cannot, and why underwritten coverage will not be bought inside a chat window.
- AI Sent Me Back to School at 43 Why AI sent a twenty-year disability insurance operator back to school for the CLU designation. AI made surface knowledge free and, in doing so, made verifiable human expertise the thing that now carries the weight.
- The Three-Minute Underwrite AI's least visible talent is removing the cost of a human reading your file. 1.2-second claim denials, three-minute quotes, and what that speed misses.