The Noise Is Everywhere. That’s the Problem.

Jul 1, 2026

An editorial by James W. Moore

The most dangerous thing in insurance right now isn’t AI. It’s the noise surrounding it.

Everywhere I turn for professional information: trade publications, conference main stages, consulting firm white papers, industry events, the AI conversation has been captured by two voices that have almost nothing useful to say to each other. One is convinced the industry is about to be hollowed out by machines. The other is promising transformation so complete that it makes your legacy systems sound like a competitive liability you should be embarrassed to still own.

Neither one is talking to you. Both are performing.

If you’re exhausted by it, good. That means you’re paying attention.

We Have Seen This Before

In 1889, Thomas Edison published a piece in the North American Review titled “The Dangers of Electric Lighting.” He argued that alternating current was a constant menace, that danger lurked in every wire, and he vowed to fight AC and refused to adopt it. He had a financial reason to say so: his patents and his business model were built on direct current, and George Westinghouse was threatening both. On the other side, electricity promoters were selling pure transformation: positioning electrification as a symbol of modernity rather than a technology to evaluate on its merits.

Both narratives were running at full volume. Both had credible spokespeople. Both were wrong about the thing that actually mattered: whether this specific technology was relevant to a specific operation, at a specific moment, for a specific set of business reasons.

The companies that got it right didn’t pick a side in that debate. They asked a different question. Does electric power actually fit our manufacturing process, our workforce, our capital position? Some electrified early and gained real advantage. Others waited until the economics were clear and competed effectively anyway. The ones who lost made the decision for the wrong reasons — either chasing the narrative or dismissing the technology wholesale because the loudest voices promoting it were obviously overselling it.

One hundred and thirty-seven years later, the incentive structure that produced that noise is still running. It just has more channels.

Why the Noise Is Everywhere

Fear sells. Hype sells. That is the whole explanation.

The apocalypse narrative sells conference registrations, magazine subscriptions, and consulting engagements. If the threat is existential, you need guidance, and guidance has a price tag. The utopia narrative sells vendor contracts, implementation projects, and keynote slots. If the upside is transformational, you need a partner, and partners have price tags too.

I don’t think most of the people producing this content are being deliberately dishonest. I think they believe what they’re saying. But both narratives are optimized for a specific outcome: your attention and eventually your budget, not your clarity.

The result is a professional information environment where the loudest voices have the strongest financial incentive to stay loud. A measured assessment of what AI can realistically accomplish in a mid-size regional carrier’s claims department over the next eighteen months doesn’t get a keynote slot. A prediction that AI will either save or destroy the industry absolutely does.

This isn’t a conspiracy. It’s an incentive structure. And it operates the same way whether the channel is a social media post, a trade publication feature, a conference main stage, or a white paper with a major firm’s logo on the cover. The medium changes. The economics don’t.

Capability Is Not a Strategy

Here is the thing both narratives share: they are talking about capability.

What AI can do. What AI might do. What AI will do to you if you don’t act, or what it will do for you if you do. Capability is the currency of the noise economy because capability is dramatic, demonstrable, and almost impossible to disprove in the abstract. A system that processes a thousand claim photos per minute is a genuinely impressive capability. A keynote predicting AI will displace 40% of insurance jobs is a genuinely alarming capability claim.

Neither one answers the question that actually matters: is this relevant to the specific problem in front of me, in my organization, with my people, my technology stack, my regulatory environment, and my budget?

Relevance is the question the noise economy cannot afford to ask. Because relevance creates accountability.

If a vendor commits to a specific outcome for your specific operation, they can be held to it. If a speaker predicts a specific disruption for your specific market segment, they can be checked. Capability claims float free of accountability because they exist at a level of abstraction where nobody is responsible for whether they apply to you. “AI can transform claims processing” is not a promise to anyone. “This system will reduce your claims cycle time by 30% within twelve months” is.

The conversation stays at the capability level because relevance introduces accountability.

Two Ways to Get It Wrong

I’ve watched the capability trap catch executives in two different directions, and I’ve watched both mistakes compound quietly over time.

The first is what I’d call the line-item problem. The board wants an AI story. The annual report needs a technology narrative. The peer carrier down the street issued a press release last quarter. So the implementation gets approved, the vendor gets selected, the announcement goes out — and somewhere in that sequence, nobody stopped to ask whether this particular capability was relevant to this particular operation at this particular moment. Often, it isn’t board pressure alone driving it. A VP needs an AI win for their annual review, so they buy capability and retrofit a problem afterward.

That’s not innovation. That’s signaling disguised as innovation. The capability was real. The relevance test never happened. The bill arrives later, when the system underperforms against expectations that were never grounded in operational reality, when the adjuster workarounds multiply, when the board asks why the results don’t match the deck.

MIT’s Project NANDA, a Media Lab study based on more than 300 initiatives and 52 executive interviews, found that roughly 95% of generative AI pilots and enterprise initiatives failed to deliver measurable ROI in 2025. That is not a technology failure. That is a relevance failure. Capability was purchased. Relevance was assumed.

The second mistake is the dismissal. The executive who sat through three CRM rollouts, two blockchain pilots, and one very expensive robotic process automation (RPA) initiative has earned some skepticism. Pattern recognition is a survival skill in this industry. And much of what’s being promoted right now genuinely isn’t relevant to their operation. They’re not wrong about the hype.

Consider the carrier that watched a blockchain pilot fail in 2019 and quietly used that experience as the reason to pass on a claims AI evaluation in 2024. The pattern felt familiar. The conclusion didn’t follow. Blockchain failed because distributed ledger technology solved a problem the industry didn’t actually have. AI applied to claims triage is solving a problem every carrier has had for decades. Same noise, different technology, different answer.

“This vendor is overpromising” is a reasonable observation. “This technology has nothing to offer us” doesn’t follow from it. The dismissal that looks like hard-won wisdom is often just the opposite reaction to the same noise. The apocalypse crowd pushed, and the skeptic pushed back, and neither one ever ran a relevance test. They just landed on different sides of a debate that was never framed correctly to begin with.

Discernment is a competitive advantage. Abdication isn’t.

The Question Worth Asking

The executives I respect most right now are not the loudest ones on the conference circuit. They’re not the ones with the most aggressive AI narratives or the most emphatic rejections. They’re the ones who have learned to ask a different question before anything else.

Not: What is AI capable of?

But: What problem am I actually trying to solve, and is this the right tool for it?

That question doesn’t generate keynote invitations. It doesn’t produce impressive demos or quotable predictions. It produces something more valuable: decisions that can be evaluated against reality by people who are accountable for the outcome.

That accountability now has regulatory and financial teeth. The SEC brought enforcement actions in 2024 against firms for false or misleading AI claims — what regulators are calling “AI washing.” D&O underwriters are increasingly focused on what executives say publicly about their AI capabilities, not just what they implement. Overstating AI to satisfy a board narrative is no longer just a strategic miscalculation. It is a disclosure problem, and the insurance industry’s own underwriters are paying attention.

I’ve watched this industry navigate technology transitions before. The ones who came out ahead weren’t the early adopters or the holdouts. They were the ones who ignored the noise long enough to ask whether the technology was actually relevant, and then made decisions they could defend with something other than a press release.

The noise hasn’t changed. That question has always been the move.

Sources

James W. Moore is the founder and publisher of InsuranceIndustry.AI, an independent editorial publication covering AI’s impact on the insurance industry. He has 40 years of experience across carriers, agencies, and wholesale operations.

Subscribe to AI Insights Newsletter

AI Disclaimer: This content was created with assistance from artificial intelligence technology. While content is based on factual information from the source material, readers should verify all details directly with the respective sources before making business decisions.

468