From Executor to Partner: A Regulated Learning Framework for Human–Generative AI Collaboration in Systematic Reviews

Systematic review requires sustained regulation of research goals, strategies, judgments, and adaptations. This conceptual paper introduces a framework for human-generative AI (GenAI) collaboration by treating systematic review as a regulated learning system that complements its methodological structure. Drawing on Zimmerman’s cyclical model, we map review activities onto forethought, performance, and reflection and explain how these cycles operate both across the whole review and within individual stages. We then adapt distinctions among self-regulation, co-regulation, and socially shared regulation to differentiate GenAI as a delegated executor, a co-regulatory influence, and a potential partner in shared regulation. Activity mappings, figures, and regulatory features specify these distinctions. Applying the framework in a systematic review of 60 GenAI-assisted evidence syntheses showed that reported applications primarily positioned GenAI as an executor of tasks within human-defined regulatory structures, with episodic co-regulation and limited evidence of shared regulation, leaving GenAI's potential as a regulatory partner largely unrealized. The framework helps reviewers integrate GenAI intentionally while retaining human intellectual and ethical accountability, and it informs the design of GenAI-supported review practice, the teaching and mentoring of novice reviewers, and the development of AI agents that support, rather than merely execute, regulation.