AI-Mediated Creative Agency: A Framework for Creative Self-Efficacy in Human-AI Collaboration

Generative AI is reshaping creative practice by expanding what individuals can produce while complicating how effort, authorship, and mastery are understood. This article introduces AI-mediated creative agency (hereafter, mediated agency) as a framework for explaining how human–AI collaboration reshapes creative self-efficacy and the horizons of creative possibility. Mediated agency refers to the experience of acting creatively when intention, process, and outcome are co-produced by human and algorithmic elements, producing new forms of attribution ambiguity, opacity, and uneven control. Extending Bandura’s account of self-efficacy, the framework reconceptualizes his four sources of efficacy information as relational processes interpreted across three dimensions: control over process, authorship, and creative identity. Grounded in sociocultural theory and situated within distributed agency, co-creativity, and technology adoption perspectives, the framework presents testable propositions and educational implications for studying and supporting creative self-efficacy in AI-mediated contexts. Sustaining creative self-efficacy, and over time broader creative confidence, requires attending not only to what AI enables people to produce but to how mediation shapes their experience of contribution, capability, and future creative possibility.