AI in insurance: from hype to reality

AI has actually been inside insurance for a long time, working quietly in pricing, claims and fraud detection. Recently a threshold was crossed: AI moved out of the experiment phase and began to create measurable impact. In this piece I want to separate the hype from the reality, because in our industry these two get confused more than anything else.

Let me draw the line first. Hype is the expectation that AI will solve everything overnight and make people redundant. The reality is plainer: AI does certain tasks far faster and cheaper, but its value emerges only when it sits on the right process and the right data. Every example here says the same thing: what matters is not the technology itself, but how it is positioned.

What has changed?

Today AI’s use in insurance has settled into concrete areas:

• Generative AI drafts policy wording and handles customer requests.
• Predictive models offer personalized coverage.
• Vision AI assesses vehicle and property damage from photos.
• Agentic AI runs multi-step administrative processes autonomously, freeing people for more strategic work.

So this isn’t a single shiny tool; it’s a layer of capability embedded at different points of the process. In the coming period, assistants embedded in underwriting and claims platforms, agentic systems that take over routine broker and customer tasks, and models that foresee losses from IoT data will all become standard.

The big players are serious now

That this is not a fad is clear from the moves of the industry’s established names. Zurich set up a lab called the Zurich AI Lab to build scalable solutions to real problems. Nationwide announced a 1.5 billion dollar investment to accelerate its AI transformation, spanning everything from infrastructure modernization to customer experience. Steps like these from two long-established names show clearly that AI has moved from an experimental concept to the core strategic muscle of insurance.

There is movement on the investment side too. AI venture funding, which had cooled for a while, has picked up again, led by the corporate funds of players like MassMutual and Munich Re. So capital also confirms this field is no longer experimental. But capital flowing in is not maturity by itself; what matters is where the money goes, and with what discipline.

The real picture: maturity is uneven

Yet looking at the whole, maturity is far from evenly spread. The Evident AI Insurance Index 2025 assesses 30 large insurers for AI maturity, and it does so from the outside in: not from companies’ own claims, but from public data, patents, research and hiring trends. The aim is exactly to see reality rather than hype: who is truly becoming an AI-first organization, and who is still experimenting?

The results are striking. AXA and Allianz lead across all four dimensions. The talent gap is clear too: half of the industry’s AI professionals are concentrated in just ten companies, and Allianz alone employs around 10 percent of the total. USAA stands out on density, with a share of AI staff three times the industry average. A few players are also well ahead on research and patents. So the line that everyone is doing AI simply isn’t true; a few institutions are genuinely building a capability while most are still entry-level.

Does it make money? The question is still open

This is the most important divider. In the same index, only 12 of 30 companies disclosed a concrete business outcome, and just 3 shared a financial return. So investment and excitement are high, but proven payback is still early-stage. This doesn’t mean AI is worthless; on the contrary, it shows the winner will be the institutions that manage AI as an organizational capability, not those that merely deploy it. Without strong data infrastructure, talent development, ethical governance and strategic leadership moving together, the advantage isn’t sustainable.

The index details separate hype from reality too. On research and intellectual property a few institutions are well ahead: a large share of AI-related academic output and citations sits in a single company, and the majority of patents in a handful of US players. On responsible AI, only 12 of 30 have made their principles public. The picture is clear: AI maturity is not a slogan, it’s organizational building work spread over years.

The biggest gap: product

There is also a point everyone skips. McKinsey’s sector AI report shows insurance is scaling AI mostly in operational areas: IT, risk and compliance, claims and service operations. These matter, because efficiency and risk control are everyone’s shared agenda today. But the same picture shows a big gap: AI use on the product and service development side is almost negligible.

The meaning is this: today AI solves the claims-and-operations problem. Tomorrow the difference will be set by companies that can create next-generation products. While startups push aggressively in areas like usage-based models, dynamic pricing, policies personalized with health data and embedded insurance, incumbents’ product-development muscle isn’t yet strong enough. The center of gravity of competition is shifting from operational efficiency to product innovation.

This gap is actually an opportunity. Usage-based motor insurance, policies personalized with health data, instant and embedded products; all are waiting to be redesigned with AI. Startups move fast here; the incumbents’ advantage is their vast data and distribution power. The real question is whether they can turn that advantage into a product-development muscle.

Next: AI itself is also a risk

Another threshold is approaching. AI doesn’t only transform insurance; it is itself becoming an insured class of risk. Model errors, liability for autonomous decisions, data and copyright issues are creating new coverage needs. For insurers who see this early and turn it into product, there is serious room here.

This new risk class also sums up insurance’s irony in the AI age. One of the most intensive users of the technology is building itself a new market by covering the very risks that technology creates. AI becomes both our tool and our product.

What it means for emerging markets

In a market like Turkey, AI is rapidly entering the corporate agenda, but most companies are still clarifying their strategy and strengthening their teams. The opportunity here is to shorten the years-long roadmap of mature markets: move straight to ready infrastructure and models, and focus limited resources on the right few problems. What matters is not making the biggest investment, but making it in the right place. A small but well-aimed start often delivers faster than a scattered, showy one.

Where I stand

Behind all of this is a single principle: position AI not in place of people, but as a tool that complements human expertise. The winners won’t be those who run flashy demos; they’ll be those who turn AI into an institutional capability that manages data, talent, ethics and leadership together. And those who can carry that capability from operational efficiency into product innovation will pull ahead. Standing on the side of concrete results rather than show applies to AI too: the index already shows who is real and who is still experimenting.