Abstract
Although ChatGPT, a state-of-the-art, large language model, seems to be a disruptive technology in higher education, it is unclear to what extent students rely on this tool for completing different tasks. To address this gap, we asked university students (N = 490) recruited via CloudResearch to rate the extent to which they rely on ChatGPT for completing 13 tasks identified in a previous pilot study. Five distinct profiles emerged: ‘Versatile low reliers’ (38.2%) were characterised by low overall self-reported reliance across the tasks, while ‘all-rounders’ (10.4%) had high overall self-reported reliance. The ‘knowledge seekers’ (16.5%) scored particularly high on tasks such as content acquisition, information retrieval and summarising of texts, while the ‘proactive learners’ (11.8%) on tasks such as obtaining feedback, planning and quizzing. Finally, the ‘assignment delegators’ (23.1%) relied on ChatGPT for drafting assignments, writing homework and having ChatGPT write their assignment for them. The findings provide a nuanced understanding of how students rely on ChatGPT for learning.
•Latent profile analysis with 490 university students revealed five distinct reliance profiles on ChatGPT.•Nearly 40% of students were ‘versatile low reliers’, showing minimal reliance on ChatGPT across different tasks.•16.5% were “knowledge seekers” exhibiting reliance for content acquisition, information retrieval, and summarization.•11.8% of participants relied on ChatGPT for obtaining feedback, planning, and quizzing (“proactive learners”).•“Assignment Delegators” (23.1%) relied on ChatGPT for drafting assignments and writing homework.