Agentic AI Teammates in Medical Research - From Tools to Collaborators - and the Accelerating Digital Divide

AbstractAgentic AI may lower barriers for individual clinician-scientists while amplifying institutional differences in who can turn ideas into reliable research. In medicine, emerging agentic systems do not just generate text - they can execute multistep workflows by retrieving information, using tools, querying databases, iterating analyses, and producing auditable outputs with limited human orchestration. For clinician-scientists, this may be transformative as it lowers the barrier from clinical question to computational analysis and expands participation in research. But the same shift intensifies a preexisting digital divide. Early examples in biomedical discovery and clinical data-facing research remain limited, but they suggest a shift from assistive AI toward agentic scientific workflows. The equity issue is therefore not only access to AI, but conversion capacity: the ability to translate AI access, structured data, secure compute, governance pathways, and methodologic support into trustworthy scientific outputs. Existing inequities in funding, personnel, prestige, and infrastructure will persist. Agentic AI may add a further divide: Small teams may be empowered where shared infrastructure exists, while data-rich institutions may compound their advantage where conversion factors remain locally concentrated. We argue that access to agentic research capacity should be treated as a shared scientific resource and potentially coordinated at national and international levels, not left to local advantage. The key task is not only to improve agent performance, but to shape the conditions under which agentic science remains trustworthy, plural, and broadly accessible, enabling shared scientific freedom.