Mapping GenAI Literacy: Disciplinary Differences, Latent Profiles, and Perceptions Among EMI Undergraduates
This study aims to explore the Generative AI (GenAI) literacy of English-medium instruction (EMI) undergraduates, with a particular focus on disciplinary variations across engineering, mathematics, and humanities and social sciences. Specifically, it seeks to a) examine differences in GenAI literacy across disciplines, b) identify distinct literacy profiles among disciplinary groups, and c) understand students’ self-perceptions of their GenAI literacy.
The advent of GenAI tools has reshaped higher education, offering personalised academic support and aiding non-native English speakers. While traditionally focused on STEM fields, the widespread adoption of GenAI, alongside concerns about accuracy and ethics, emphasises the urgent need for GenAI literacy education across all disciplines. However, gaps remain in the literature regarding how students’ GenAI literacy varies across disciplines and how their underlying literacy profiles are shaped, particularly in EMI contexts, where non-native English-speaking students face additional linguistic challenges.
Utilising a mixed-methods approach, the research assesses five critical dimensions of GenAI literacy: basic technical proficiency, communication proficiency, creative application, critical evaluation, and ethical competence. The quantitative phase, which included 347 questionnaire participants recruited via convenience sampling, employed the Kruskal-Wallis test to examine disciplinary differences in GenAI competencies. Further, a multigroup latent profile analysis was conducted to identify distinct literacy profiles. To complement the quantitative findings, follow-up semi-structured interviews were carried out with 24 students to collect in-depth qualitative data. These interviewees were drawn from the questionnaire participants using a nested sampling strategy. Reflexive thematic analysis was then applied to uncover key themes related to students’ perceived GenAI literacy.
This study highlights notable variations in GenAI literacy among students across different disciplines and identifies distinct learner profiles within an EMI university context. The findings underscore the importance of considering both disciplinary and learner-profile factors when developing educational strategies. This work offers a foundation for designing equitable and targeted strategies to develop students’ AI literacy across all disciplines, a pressing need in EMI contexts where learners navigate additional linguistic challenges.
The Kruskal-Wallis test results indicated that, with the exception of ethical competence, engineering students outperformed their peers in mathematics, humanities, and social sciences across four dimensions of GenAI literacy: basic technical proficiency, communication proficiency, creative application, and critical evaluation. Additionally, the multigroup latent profile analysis identified three distinct literacy profiles across disciplines: Foundational Learners, Balanced Practitioners, and Proficient Achievers. Complementary qualitative insights from interviews corroborated these findings and provided nuanced explanations of the underlying patterns.
Synthesising these insights, evidence-based pedagogical recommendations are proposed: the integration of AI literacy courses across all disciplines to foster foundational competencies and equitable access, and the implementation of profile-specific educational strategies to enhance personalised learning.
Refining methods for assessing GenAI literacy is recommended for future studies to enhance their validity and reliability. This includes employing more representative sampling, integrating observation-based or task-based measures alongside self-reported literacy levels, and further refining the theoretical frameworks underpinning AI literacy in response to evolving technologies.
The insights from this study into GenAI literacy and its disciplinary variations will enable universities to develop more responsive and inclusive educational strategies, ultimately fostering a more AI-literate society. This will ensure that graduates across all disciplines are better prepared to effectively and critically engage with AI technologies in their future careers and daily lives.
Future research should consider stratified sampling across multiple institutions, regions, and cross-national contexts to capture a more representative picture of GenAI engagement and facilitate meaningful comparisons across educational systems. In addition, while the study focused on undergraduates, postgraduate students, and academic staff are also critical stakeholders in the AI literacy agenda. Investigating how these groups engage with GenAI could provide valuable comparative insights. More importantly, the identification of learner profiles also raises new questions about movement between profiles over time and the kinds of interventions that support such transitions. Longitudinal studies and action research involving instructional design experiments could help clarify how GenAI literacy evolves and what pedagogical strategies are most effective for supporting growth.
Back