Beyond the Algorithm: A Stewardship Framework for the Hand Surgeon Adopting Artificial Intelligence.

Artificial intelligence is entering hand surgery through imaging, outcome prediction, and patient communication. Neural networks read scaphoid and distal radius radiographs. Machine learning models predict outcomes after carpal tunnel release. Large language models are being tested for patient communication and chart drafting. Adoption, however, has outpaced validation. Most hand surgery artificial intelligence tools are tested only on data resembling their training set, deployed in workflows that have not been audited, and rarely remeasured after release. The hand surgeon remains the clinical decision maker into whose workflow these tools are integrated, and is therefore accountable for the patient outcomes they shape, even when the model's internal workings remain opaque. We propose a four-part stewardship framework for the hand surgeon adopting these tools: external validation on the population the model will actually see; transparent disclosure of intended use and known failure modes; defined escalation paths when the model and surgeon disagree; and prospective audit after deployment. We set out the specific questions to ask a vendor, the reporting checklist items that support each question, and the performance metrics a surgeon needs to interpret before adopting a tool. This framework has implications for residency training, board certification, the editorial standards of hand surgery journals, and the regulatory pathways that bring such tools to the clinic.