FutureAiSummit'25: insurance in the age of AI

Summary

On the Red Stage panel at FutureAiSummit'25 we discussed where AI meets the insurance industry. With me were Ege Örer, head of IT and innovation at Aksa Sigorta, Prof. Dr. Selim Yazıcı, author of the book Fintech and Insurtech, and Yılmaz Sonışı, a management consultant who teaches AI at the Turkish Insurance Institute Foundation's Broker Academy. I moderated and also shared my own view as a technology entrepreneur and insurance broker. I defined AI as systems that imitate the human abilities to think, learn and predict, and explained how we have moved from simple algorithms with pre-set rules to structures that learn patterns from large data and make predictions.

I showed how AI in insurance has moved from a supportive technology to a first-line decision-maker in places, with the Tractable example: Allianz, Tokio Marine and Covéa integrated this application, which estimates repair cost from a photo of the accident, into their decision processes; adjusters first used the output as a reference, then for some claims it became the decision itself. Covéa announced it would assess roughly 150,000 claims this way in 2025. From there I moved to my main thesis: AI is not the same AI for any industry, because every industry has its own knowledge and its own rules. An AI trained on a sector's knowledge doesn't just speed up processes; it fundamentally changes how risk is assessed, how price is set and how claims are managed. This is as much a redefinition of culture, ethics and the human touch as a technological revolution.

The panelists' field experience made the picture concrete. Ege Örer described AI through three legs: the readiness of employees, of infrastructure and of governance. At Aksa, 500 of 1,200 employees actively use AI tools, the company runs 16 live AI projects and has a vision document reaching to 2030; it set up an AI committee that checks compliance with data-protection law, security and group standards, and started training from top management. Yılmaz Sonışı stated the reality of the intermediary channel: Turkey has roughly 17,000 agencies and 220 brokers; you cannot make a non-digitalised firm use AI, and having ChatGPT draft an email is not digitalisation. The first question owners still ask is about cutting headcount, when the real question is how to issue 54 quotes a day instead of 12.

To the question of whether an insurance industry without people is possible, the professor recalled the fences put up around the first welding robot at Ford, and Ege Örer the automated board line that took over the work of fifteen people in 1995; in both cases people found new work. I gave the closing: chat applications writing convincing essays or code is no sign of real intelligence. We are still talking about a technology that imitates human faculties, and the ability to imagine has not been handed over yet. In insurance specifically, AI will speed up processes; but it will prepare us for a new era by changing our job descriptions, without cutting the workforce, perhaps even expanding it.

In this talk

• The definition of AI: systems that imitate the human abilities to think, learn and predict; the shift from rule-based to learning structures
• The Tractable example: repair cost from an accident photo; integration by Allianz, Tokio Marine and Covéa, 150,000 claims in 2025
• AI is not the same for any industry; it has to be trained on the sector's own knowledge and rules
• Beyond speeding up processes: a fundamental change in how risk assessment, pricing and claims are viewed
• A redefinition of culture, ethics and the human touch; the subject is not technology alone
• The Aksa Sigorta experience: readiness of employees, infrastructure and governance; 16 AI projects, an AI committee, training from the top
• The reality of the intermediary channel: 17,000 agencies and 220 brokers; no AI without digitalisation; productivity over headcount cuts
• Regulatory differences: the EU's caution, the US's permissiveness, China's biometric-data limits; the EU AI Act and data-protection law
• Two lessons from history: the welding robot at Ford and the automated board line in 1995; people find new work
• Convincing output is not real intelligence; the ability to imagine remains human
• The workforce in insurance won't shrink, job descriptions will change; the sector is preparing for a new era

AI is not the same AI for any industry, because every industry has its own knowledge and its own rules.