Between Fear and Adoption: The AI Paradox in Medico-Legal Practice Among National Insurance Physicians

To assess artificial intelligence adoption patterns, trust levels, and perceived utility among physicians conducting disability assessments for a national insurance system, with a focus on the paradox between widespread use and limited trust in artificial intelligence for medico-legal decisions.

Artificial intelligence is rapidly entering clinical workflows, yet its integration into insurance medicine-where physicians must translate complex clinical findings into legally defensible administrative decisions-remains unexplored. Physicians conducting disability assessments operate at the intersection of clinical judgment, administrative burden, and legal accountability, making them a critical population for evaluating artificial intelligence adoption and its epistemic limits.

A cross-sectional survey was conducted in February 2026 among 428 physicians working in medical committees, appeals boards, and advisory roles at the Israeli National Insurance Institute (response rate 85.6%). The questionnaire assessed artificial intelligence usage frequency, trust in artificial intelligence-supported decisions (5-point scale), perceived risks, and preferred applications.

Artificial intelligence adoption was widespread (76% regular or occasional use; 27% daily use), yet trust remained low (mean 2.7 of 5). Daily users showed higher trust than non-daily users (odds ratio 1.81; 95% confidence interval 1.14-2.88). Administrative burden was the dominant adoption driver. Preferred applications included medical record summarization (85%), inconsistency detection (45%), and protocol drafting (35%). Most use was informal, relying on personal tools. Senior physicians expressed relational concerns; less experienced physicians emphasized technical risks.

Insurance medicine physicians distinguish between instrumental trust for administrative efficiency and epistemic trust for legally accountable judgment. Artificial intelligence integration should emphasize administrative augmentation while preserving human interpretative authority.