Artificial intelligence in pediatric diabetes: Are we building decision support or decision replacement?

Artificial intelligence (AI) is increasingly embedded in pediatric diabetes care, driving advances in glucose prediction, insulin dosing, and remote monitoring. The central risk accompanying this progress is not technical failure, but decision replacement : the gradual transfer of clinical authority from human judgment to algorithmic output. As AI systems gain autonomy, opacity, and commercial incentives, scalability threatens to normalize algorithmic decisions as definitive rather than advisory. This editorial argues that preserving safe and equitable pediatric care requires a governed, human-centered AI paradigm in which algorithms remain explicitly subordinate to clinician judgment and family context. Such governance must address model limitations, accountability, and power asymmetries in development and deployment across diverse health systems.