From affordance offers to uptake depth: examining hybrid intelligence in human–AI collaborative academic writing

The significant and widespread influence of generative Artificial Intelligence (AI) in education has intensified attention to the emerging concept of Hybrid Intelligence (HI), which emphasises learners’ capacities to co-construct knowledge and advance task performance in coordination with AI. In higher education, academic writing has become a critical testbed for HI, as generative AI systems are increasingly embedded in students’ composing practices. Although contemporary research has examined learners’ interactions with AI in academic writing, HI perspectives that simultaneously account for AI response affordances and human uptake remain limited, particularly in explaining how offered affordances are differentially enacted in subsequent writing. Accordingly, this study investigated HI in AI-assisted academic writing using an episode-based design that systematically linked learner prompts, corresponding AI responses, and learners’ subsequent writing actions. Forty undergraduate students completed a 30-minute AI-permitted essay task, yielding 178 collaboration episodes. The findings showed that learners’ prompt intent primarily predicted both AI affordance offers and uptake depth, while adding response-side affordance indicators and richness did not improve model fit beyond the prompt-side model. The study advances both the conceptualisation and operationalisation of HI in learning, while providing new insights into how AI affordance offers and human uptake interact to shape learning-relevant collaboration.