Generative artificial intelligence (GenAI) has fundamentally transformed creative processes in educational contexts, challenging traditional definitions of creativity. This paper examines the evolving relationship between human creativity and AI’s generative capabilities, revealing that effective human-AI co-creativity in education depends critically on balanced agency between students and AI systems. We argue that optimal creative outcomes and self-efficacy will emerge when both students and AI tools contribute meaningfully to the creative process through dynamic interaction rather than when students merely edit AI-generated content.

Building on Boden’s (1994) fundamental distinction between Improbabilist Creativity (generating novel combinations from familiar elements) and Impossibilist Creativity (involving transformational shifts that redefine conceptual spaces), we apply these concepts to understand GenAI’s current capabilities and limitations. GenAI primarily operates in the realm of Improbabilist Creativity, while the deeper transformative aspects of Impossibilist Creativity remain distinctly human.

Based on this reasoning, we propose a three-dimensional framework for GenAI integration in education:
1) Distributed Creative Agency: Educational approaches should recognize and cultivate a genuine co-creative relationship where both human students and AI systems possess meaningful agency. This involves designing learning experiences that position AI not merely as a tool but as a creative partner with its own generative capabilities, while students learn to enter a dialogue with AI systems, creating space for emergent outcomes neither would achieve independently.
2) Pedagogical Integration: Curriculum should explicitly teach the complementary strengths of human and AI creativity, with assignments designed to leverage co-creative processes that consider both human and AI contributions. This includes developing assessment approaches that evaluate the quality of the collaborative relationship and the emergent creativity that results from this shared agency.
3) Metacognitive Development: Educational activities should foster students’ ability to reflect on the dynamic interplay between human and AI agency, developing judgment about how to navigate this relationship productively. This includes structured reflection on how the distribution of agency influences creative processes and outcomes.

While students and AI systems bring different capabilities to creative work, the most valuable outcomes emerge when both contribute actively to an iterative process. AI systems offer novel combinations and patterns beyond conventional human thinking, while students contribute intentionality, evaluation, and contextual understanding. Critically, AI lacks the ability to deconstruct the design process across different artistic modalities—it currently cannot fully analyze why certain creative choices work in specific contexts or translate principles between domains.

As GenAI rapidly transitions from tool to active partner in the creative process, education must evolve beyond simply teaching students to use AI effectively. The future of creativity in educational contexts lies in fostering a mindset that embraces the dynamic interplay between human and AI agency—a vision where students thrive in creative environments where agency is distributed and negotiated.

Keywords: Generative AI, Distributed Agency, Co-Creativity, Education, Human-AI Collaboration.