AI Literacy Does Not Eliminate Challenges: Evolving Complexity among English Education Students in AI Integrated ELT Classrooms

Shofia Mardhiyah (Islamic State University of Sunan Gunung Djati Bandung)
Tedi Rohadi (Islamic State University of Sunan Gunung Djati Bandung)

IDEAS: Journal on English Language Teaching and Learning, Linguistics and Literature
Vol. 14 No. 1 (2026): IDEAS: Journal on English Language Teaching and Learning, Linguistics and Lite

The increasing integration of artificial intelligence (AI) in education has generated a widespread assumption that higher AI literacy reduces the challenges students face. This study challenges that assumption by examining how challenges persist and evolve alongside students’ developing AI literacy. Focusing on 47 English Education students enrolled in an AI-integrated ELT classroom at an Islamic State University in Java, Indonesia, this research investigates: (1) the level and distribution of AI literacy, and (2) how challenges shift across different levels of competence. A sequential explanatory mixed-method design was employed, combining questionnaire data from 47 respondents, adapted from the Meta AI Literacy Scale (MAILS), with semi-structured interviews involving three purposively selected participants representing contrasting literacy levels. Grounded in socio-technical systems theory and critical AI literacy frameworks, the findings reveal that students demonstrate a generally high level of AI literacy (M = 2.97), with the strongest performance in operational use (M = 3.07) and the weakest in generative competence (M = 2.53). More importantly, challenges do not diminish as literacy increases; instead, they become more complex. Lower-level students struggle with basic operational issues such as prompt formulation and output interpretation, while intermediate-level students encounter cognitive difficulties in refining and aligning AI responses. Higher-level students face critical challenges related to accuracy evaluation, bias detection, and uncertainty management. These findings suggest that AI literacy development is not a linear progression toward reduced difficulty, but a layered socio-technical process in which higher competence exposes deeper limitations. This study contributes to AI literacy research by reframing challenges as evolving rather than diminishing and highlights the need for instructional approaches that emphasize critical, reflective, and ethical engagement with AI in educational contexts.