Co-Creating Educational Futures: How Speculative Prompting Shapes AI-Generated Design Fiction

Design fiction uses narratives to make possible futures tangible and open to reflection, while large language models (LLMs) introduce new opportunities for human-AI co-creation in speculative design. This paper examines how different forms of human–AI interaction shape narrative qualities and thematic patterns of design fictions in the domain of education. We co-created eighteen design fiction narratives about AI in education with GPT-4, guided by six speculative prompting approaches, and analyzed them through a heuristic-based evaluation complemented by an inductive thematic analysis. Our findings suggest that the qualities of these narratives do not emerge from the model alone, but from the interplay between human framing, prompting strategies and model behavior. Different forms of interaction foreground different concerns, narrative perspectives, and speculative tensions, shaping how educational futures are imagined and contested. Building on these insights, we present the Speculative Design Fictionist, a prompt-based design instantiation that externalizes selected strategies and evaluates considerations as a scaffold for reflective human–AI co-creation. Rather than automating speculative design, the instantiation illustrates how generative AI can support collaborative inquiry into educational futures through structured yet open-ended interaction.