AM Best Survey: AI Ready, Insurers Not

The insurance industry stands at a technological crossroads. AM Best released a new survey revealing that while artificial intelligence appears ready for deployment, most carriers are not prepared to use it effectively. The disconnect between enthusiasm and execution is widening across the sector.

Nearly 60% of respondents expect AI to significantly transform their business models within the next one to three years. Yet the same survey identifies data readiness, security concerns, and legacy system integration as the largest impediments to deployment. This is not a technology problem. It is an infrastructure problem.

The survey results appear in a new Best's Segment Report titled "Artificial Intelligence Appears to be Ready, But Most Insurers Are Not." More than 150 rated insurers and managing general agents participated. The findings expose a gap between ambition and operational reality.

Forty-one percent of organizations report actively using AI across core business areas. Nearly 20% claim advanced implementation stages. A majority have formal AI policies in place. On paper, the industry looks prepared. In practice, the data tells a different story.

Legacy systems create significant barriers. Many were simply not built for this type of data integration. They store information in inconsistent formats lacking standardization. Kaitlin Piasecki, industry research analyst at AM Best, noted that AI systems depend heavily on high-quality, clean, and well-structured data. When that foundation is missing, the entire initiative falters.

AI systems can produce unreliable outputs when underlying data is of poor quality or fragmented across legacy systems. Sridhar Manyem, senior director of Industry Research and Analytics at AM Best, emphasized that insurers who have invested in modernizing their legacy systems will find AI integration easier. Those who haven't face a steeper climb.

Security concerns compound the challenge. Insurers view potential breaches of AI systems by bad actors as a significant threat. Data readiness ranks equally high. Change resistance and third-party model risk worry them less. The fear is not about adoption. It is about vulnerability.

Approximately two-thirds of respondents plan to increase AI investment in the next 12-24 months. Improving employee productivity, lowering operating costs, and assisting with underwriting functions lead the list of goals. For those who have implemented AI solutions, 63% reported small improvements in workforce productivity and satisfaction. Only 11% reported significant improvement.

Staffing expectations remain mixed. Thirty-one percent said there would not be any material change to staffing. Thirty-seven percent expect employees to be redeployed to higher-value work. The technology is not replacing jobs wholesale. It is reshaping them. (This is a relief to anyone who has spent years building actuarial models by hand.)

Jason Hopper, associate director of Industry Research and Analytics at AM Best, cautioned that return on investment in AI would be difficult to measure at this stage. Cost benefits will likely take years to materialize. Insurance roles requiring judgment, critical thinking, and accountability remain beyond AI's current capabilities.

Corroborating research from LIMRA and Equisoft reinforces these findings. Their 2025 report found that 78% of global life insurers believe data readiness is the biggest challenge to getting value from AI. Forty-six percent say they are not ready to implement AI solutions. The pattern is consistent across the industry.

Machine learning remains the most widely adopted AI technology. Natural language processing and large language models show rapid future growth potential. Eighty-seven percent of respondents are currently using AI in some operational areas like underwriting, operations, and new business. The technology is present. The infrastructure is not.

Regional differences emerge in the data. Australian insurers led in data readiness maturity assessments. North America scored somewhat lower than other regions. Sixty-six percent of United States life insurance carriers feel unready for AI. Organizational alignment ranks as the strongest dimension. Sourcing and integration rank as the weakest.

The physical reality of insurance work matters here. Agents and underwriters interact with clunky interfaces daily. They navigate multiple systems to pull client data. They spend hours on manual data entry. AI promises to automate these tasks. But automation requires clean data pipelines. Most carriers do not have them.

Carriers that have already implemented AI solutions encounter project challenges due to unexpected technology issues. Scaling challenges arise. Erroneous assumptions made during planning create negative impacts. The gap between pilot programs and enterprise-wide deployment remains substantial.

Data governance presents another hurdle. Many organizations report that governance guidelines have been created but adoption and accountability remain low. This creates an opportunity for progress globally. Without governance, AI outputs lack consistency and reliability.

The survey does not suggest abandoning AI. It suggests preparing for it. Modernizing legacy systems takes time. Building data governance frameworks requires investment. Training staff on new workflows demands resources. The technology will wait. The market will not.

Whether insurers can bridge this gap before competitors do remains the real question. The technology is ready. The industry is not.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn