Ethical Use of AI Tools in L2 Learning: Observations of Undergraduate Critical AI Literacy in an ESL Context

Abstract

The emergence of Artificial Intelligence (AI) in higher education has enhanced second language (L2) learning by solving language issues in text production, translation, grammar correction, summarisation, and academic research. However, the overuse of AI tools indicates an ethical or critical drawback among the learners. In this scenario, critical AI literacy means “the ability to understand, evaluate, and use AI tools with a perception of their limitations, biases, and risks”. It enables the learners to question AI-generated outputs and use AI responsibly as a support for learning rather than a replacement for human thinking. Against this backdrop, this study investigates undergraduate awareness of ethical AI use and their level of critical AI literacy, aiming to encourage responsible AI engagement. The data were gathered under a mixed-methods research design that combines a questionnaire administered to 120 second-year Arts undergraduates at a Sri Lankan state university and semi-structured interviews with nine ESL instructors teaching the same cohort. The study explores the students’ understanding of the ethical considerations in AI usage, their awareness of the risks of bias, misinformation, and plagiarism, and their critical evaluation of AI-generated outputs. The findings reveal high levels of AI adoption, with students frequently using multiple tools for academic tasks. The majority of the participants demonstrated awareness of ethical concerns, despite their unfamiliarity with the term “critical AI literacy”. However, this contrasts with classroom observations by the teacher respondents, which indicate that students often submit AI-generated responses for assignments and classroom activities. The results further show a variance in critical AI literacy levels. Overall, the study suggests that although the undergraduates display a notable level of ethical awareness, their evaluation of critical AI usage remains inconsistent. The paper argues in favour of a systematic introduction of critical AI literacy components into L2 curricula.