Most Insurers Expect AI to Transform Their Business but Remain in Early Stages of Adoption
Nearly 60% of insurers expect artificial intelligence to significantly transform their business models within one to three years, but only about 20% consider their organizations at an advanced stage of AI implementation, according to a survey by AM Best.
The gap between expectation and execution points to an industry that recognizes AI’s strategic importance but continues to grapple with data quality, legacy systems, and security concerns, the report said.
The survey, which included 152 rated carriers and managing general agents, found that 53% of respondents described their approach as cautiously keeping pace with the industry, while 27% said they aim to be “successful followers” learning from other companies. Just 20% identified as first movers and industry innovators, AM Best found.
Governance Frameworks Taking Shape
As AI adoption accelerates, insurers are building out oversight structures. Sixty-three percent of respondents reported having a formal AI policy in place, and 47% said they have robust governance processes related to AI, according to the survey. Regulatory developments remain a source of concern for 42% of respondents, while 40% expressed a neutral stance.
AM Best noted that the NAIC introduced its Principles on Artificial Intelligence in 2020, followed by a 2023 model bulletin addressing potential inaccuracies, unfair biases, and data vulnerabilities. The bulletin emphasizes responsible governance and risk management to ensure fair outcomes for insurance consumers, the report said.
Operational risk was cited as the single biggest risk in implementing AI by 41% of respondents, followed by data privacy at 22% and regulatory risk at 15%, AM Best found.
Data and Legacy Systems Pose the Biggest Hurdles
The survey identified data readiness (45%), security and privacy (43%), and legacy system integration (41%) as the three largest challenges insurers face in deploying AI. Talent and capacity constraints followed at 38%, while unclear business case and return on investment concerned 34% of respondents.
Legacy systems create significant barriers because they were not built for AI-style data integration and often store information in inconsistent formats, the report said. AI systems can produce unreliable outputs when underlying data is fragmented, poorly governed, or lacking context — a particular risk in insurance, where tools draw on large volumes of both structured data and unstructured content such as submissions, claims narratives, and policy documents, according to AM Best.
Insurers carrying significant technical debt from short-term system patchworks may face high future costs to make their infrastructure compatible with AI, the report warned.
Productivity Gains Lead, but ROI Remains Elusive
Improved employee productivity was the top operational goal for AI investment, selected by 68% of respondents. Lower operating costs (47%) and enhanced underwriting capabilities for risk selection and pricing (37%) rounded out the top three priorities, according to the survey.
In terms of functional areas, IT and developer productivity showed the highest rate of live AI use at 38%, followed by administrative and operations functions at 27%, AM Best found. Claims, underwriting, and actuarial departments showed varied levels of exploration and piloting.
Despite growing investment — 66% of respondents plan to increase AI spending in the next 12 to 24 months — only 13% expressed confidence in their ability to measure AI’s return on investment. The largest tangible gains so far have been in workforce productivity and satisfaction, where 63% of respondents reported small improvements and 11% reported significant improvements, the survey found. By contrast, 78% said AI has had no impact on premium growth.
On the workforce front, insurers largely view AI as an augmentation tool rather than a replacement for employees, AM Best said. Thirty-seven percent of respondents said staff would be redeployed to higher-value work, while 30% anticipated no material change to staffing levels, the report said. Only 9% expected a net reduction in headcount.
At the same time, 56% of respondents disagreed or strongly disagreed that their employees possess sufficient AI-related skills and training, underscoring a talent gap that could slow adoption further.
Obtain the full report here. &


