Understanding Learner Engagement and Persistence in AI-Assisted Interactive Climate Fiction Learning

Abstract

Advances in generative artificial intelligence (AI) have enabled educators to create increasingly sophisticated interactive learning materials, yet relatively little research has examined how learners engage with and persist in such interventions over time. This exploratory case study investigated learner engagement and persistence within a five-stage AI-assisted interactive climate fiction (cli-fi) intervention delivered through H5P activities embedded within the Moodle learning management system as part of an extracurricular sustainability initiative at an English-medium university in China. Participation, activity scores, completion data, and time-on-task measures were analysed descriptively across a five-week intervention. While learners who engaged with the intervention demonstrated high levels of engagement, reflected in strong activity scores and substantial interaction with the materials, participation declined markedly across successive stages, falling from 25 learners in Story 1 to six learners by Story 5. Descriptive improvements were observed in environmental knowledge scores among the small number of learners completing both pre- and post-tests, although these findings were inconclusive. The findings highlight a clear distinction between engagement and persistence, suggesting that engaging digital learning experiences do not necessarily translate into sustained participation. The study contributes to emerging research on AI-assisted learning design by emphasising the importance of considering persistence as a distinct outcome when designing and evaluating multi-stage digital learning environments.