AI Adoption, Data Governance Gaps, and Training Needs Among Hematologists in Türkiye

A national, cross-sectional survey evaluated the knowledge, clinical and research utilization, governance, and educational needs surrounding artificial intelligence (AI) among practicing adult and pediatric hematologists in Türkiye. Utilizing the verified member registry of the Turkish Society of Hematology, 146 of 639 invited hematologists completed the self-administered instrument (22.8% response rate), with 65.8% practicing adult hematology.

Clinical adoption of AI was widespread, with 87% reporting use in the prior three months, yet only 14% had received structured AI training. In research and manuscript writing, 55.5% reported using generative AI; among these users, 81% encountered fabricated references or incorrect information. Serious data governance challenges were identified: 48.6% reported inputting patient data into non-approved public chatbots, comprising 40.4% in anonymized form and 8.2% involving identifiable personal or institutional details. Formal governance frameworks were scarce, with only 4.8% reporting written institutional AI policies, 64.4% reporting none, and 30.8% unaware of their institution's policy status.

Respondents exhibited strong responsible-use attitudes, with 90% endorsing output verification, 73% supporting declaration of AI use, and only 8.2% expressing feelings of professional threat. Measured educational need was high and internally consistent (Cronbach's α = 0.92). Multivariable analysis revealed that higher digital literacy was independently associated with research and manuscript-writing AI use (adjusted OR 1.87 per point), while attitudes remained comparable between adult and pediatric hematologists. The authors conclude that AI integration is outpacing formal governance and training, highlighting the need for specialty-specific guidelines, structured literacy programs, and supervised, institution-approved tools.