Insurance Technology Diary
Episode 31: AI Aud(it)actiy
Guillaume Bonnissent’s Insurance Technology Diary

When I was little, every year I told Père Noël what I wanted for Christmas. I made list, attested vehemently to my goodness over the year, then sent off the lot to le pôle Nord.
This activity tended to pay middling dividends. Usually I would find several practical things that I hadn’t requested under the tree (underpants, always), along with one or two fun things from my list, but the Big Man hardly ever delivered on my extreme requests like ‘motorcycle’ or ‘boxing lessons’.
One year it became clear that my parents had intercepted my epistle to Santa. I had erred on the side of practical with the hope of improving my Christmas-list hit rate. I’d asked the Saint for two pairs of jeans. Specifically, I had requested one baggy, and one parachute.
Before I could lick the colourful festive stamp, though, my mother had explained that these two wide-legged items were the same thing. She suspected that she thought I didn’t even know what I was asking for. I confess in retrospect that she was right. My older sister always looked fashionable to my naïve eyes, so I’d simply asked for the kinds of trousers that I’d heard her talking about.
Flash forward. This week a company that makes and markets rather effective tools to detect fraudulent insurance claims has published what it calls its ‘GenAI Maturity Assessment’. The eight-question test is predictably disappointing, because it does everything to foment GenAI FOMO, but nothing to address the primary question everyone should ask: do we need generative AI in our processes, or is it just fashion?
The Assessment dictates an affirmative answer without even asking the question. For example, question three asks: “Does your organization have a dedicated budget for GenAI for the upcoming financial year?”
I hope even the most ‘mature’ AI users will answer ‘no’. The use of GenAI is not an end unto itself, so its use should not be pursued as one. Would you set a dedicated budget for Systems Resources Management or (for that matter) bin-emptying? No one would extract those costs and set a “dedicated budget,” but they’re on the same order.
Another question lists nine departmental functions – from marketing to legal – and asks in which auditees “are currently using or planning to use GenAI.” The strong implication is that you are not a mature GenAI business until the tech has penetrated all these functions.
Only the final question addresses the potential benefits of GenAI. It asks auditees to rank, from “none” to “exceptional,” any “measurable outcomes… from the integration of GenAI into your analytics processes” in areas including data processing, customer satisfaction, general operational costs, compliance, and even employee satisfaction.
If you haven’t already reaped some benefit from GenAI in these areas, you are at risk of immaturity, according to the Auditors.
When I was just a goose writing to Father Christmas, my confusion about what was fashionable and what I needed was simply cute. But the grown-ups knew better. Similarly, mature businesses should know what problems they face, and leave it up to specialists to decide if they need GenAI to resolve them.
