One in nine broker submissions received by insurers are either declined or left without a quote because operational constraints prevent carriers from processing risks they would otherwise be willing to underwrite, according to a new global insurance study.
The finding was contained in research conducted by ISG, a global technology research and advisory firm, and commissioned by mea Platform, which examined how insurers are redesigning their operations around artificial intelligence.
The study found that 83 percent of insurers are prepared to allow AI to handle repeatable operational tasks, as carriers look to improve their ability to process more business without compromising human oversight of consequential underwriting and claims decisions.
The research surveyed senior executives across underwriting, operations, claims, technology and transformation functions in North America, Europe and Asia. It examined 20 operational activities, including submission intake and triage, quote generation, bordereaux processing, claims adjudication and compliance screening.
According to the report, the operational bottleneck is becoming a constraint on insurers’ ability to capture business. Carriers surveyed said some risks within their underwriting appetite are still going unquoted because their existing operations cannot keep pace with the volume or complexity of submissions.
The study found that where AI is already being deployed in insurance operations, 61 percent of respondents reported productivity improvements, while 51 percent reported faster cycle times.
Insurers also expect AI-enabled operational changes to reduce operating costs by 16 percent over the next two years, although the research suggests that increased capacity to handle more business could be a larger incentive than cost savings alone.
When asked what could most improve the way brokers view their organisations, 64 percent of respondents identified pricing, while 52 percent pointed to ease of doing business and 51 percent cited AI-driven speed and completeness of responses to submissions.
Insurers cautious about AI decision-making
Despite the willingness to automate repetitive processes, the research showed that insurers remain cautious about allowing artificial intelligence to make high-consequence decisions without significant controls.
About 75 percent of respondents said they would trust an insurance-specific model or a governed hybrid system for high-consequence underwriting and claims decisions conducted with limited human oversight.
Only six percent said they would trust a general-purpose AI model on its own for such work.
The report defines a governed AI model as one operating within an insurer’s own rules, including controlled policy wording, endorsements, underwriting appetite, claims guidance, referral thresholds and permissions. Such systems also allow designated employees to override or stop the AI’s actions.
The distinction reflects insurers’ preference for using AI to handle repetitive processes while retaining human involvement in decisions that could materially affect risk selection, pricing or claims outcomes.
The study found that 86 percent of respondents believe consequential decisions that differentiate an insurer from its competitors should remain with people.
Although AI-led operational redesign is becoming a priority across the industry, the research indicates that most insurers are still some distance from fully AI-driven operations.
Some 96 percent of insurers surveyed have AI-led operational redesign on their agenda. However, only 13 percent currently have an advanced operating model in which AI plays a central role in workflow design.
That figure is expected to rise to 52 percent within two years.
Fully AI-native operations, defined by the study as AI executing defined processes end-to-end while humans establish policies and manage exceptions, currently account for less than one percent of the market. About 12 percent of insurers expect to reach that stage within two years.
Ashish Jhajharia, insurance SME and principal analyst at ISG, said the gap between insurers’ plans and their current adoption shows that the industry is still working through the operational requirements needed to deploy AI at scale.
Martin Henley, chief executive of mea Platform, said insurers are prepared to hand repetitive work to AI, provided the technology understands insurance and operates within their own rules.
“Insurers are ready to hand the repeated work to AI, and they will do it on one condition, which is a model that understands insurance and operates inside their own rules,” Henley said.
The report added that 27 percent of insurers remain in the evaluation stage rather than actively building AI-led operating models.
