Human-AI collaborative reading in academic contexts: An exploratory case study

Abstract

The increasing prevalence of generative artificial intelligence (GenAI) in education presents new opportunities and challenges for second language (L2) learners' academic literacy development. While tools like ChatGPT can act as virtual experts, offering immediate feedback and engaging learners in reflective dialogues, little is known about how L2 readers interact with them in their academic practices and how their reading processes evolve. Taking a sociocognitive view, grounded in sociocultural theory and interpreted through the heuristic lens of the Direct and Inferential Mediation (DIME) model of reading comprehension, this exploratory case study examined how one international graduate student engages with GenAI to enhance academic reading. Data were collected over six weeks through screen-recordings of reading sessions, chat logs, stimulated recall interviews, and semi-structured interviews. Findings revealed that ChatGPT functioned as a translator, knowledge facilitator, and strategic partner, reducing reading anxiety while supporting comprehension across key components of the DIME model, such as vocabulary knowledge, background knowledge, inference-making, and reading strategies. Notably, the learner demonstrated a shift in reading approach, moving from an initial dependence on translation toward more strategic engagement with academic texts, including the use of summarization and inference. However, perceived AI limitations, such as questionable reliability and excessively detailed responses, occasionally caused fatigue for the learner. By providing an in-depth exploration of human-AI interaction in academic reading, the study contributes to the growing body of literature on GenAI's role in L2 education, laying the groundwork for future studies on its long-term impact on L2 reading.