ESL STUDENTS’ PERCEPTIONS OF AI-ASSISTED LEARNING OF ENGLISH LITERATURE: BENEFITS, CHALLENGES, AND ACADEMIC IMPLICATIONS IN SOUTH PUNJAB UNIVERSITIES

Abstract

Academic environment in undergraduate education in resource-constrained universities has transformed significantly due to the incorporation of the use of artificial intelligence (AI) tools that are specially adjusted to the process of reading, writing, and analyzing literature by students. The study will focus on how English as a Second Language (ESL) students in three state universities in South Punjab, Pakistan, feel regarding the usage pattern of AI tools, preference in platforms, and the academic implication of these tools. The study conducted a survey of 100 ESL students based on a quantitative descriptive-correlational design, and a 25-item Likert-scale questionnaire that was developed by the researcher and has a high internal consistency (Cronbach’s α = .87) was used to collect data. The findings showed that the preference was towards ChatGPT (80%), then Meta AI (15%), and the Google Gemini (5%). Despite the fact that 70 percent of students referred to themselves as frequent users, there is still a substantial number of 10 percent who were completely unaware of these tools, signaling the presence of a digital divide that still prevails. The descriptive analysis showed that the means of strong reliance (M = 4.10, SD = 0.65) and alarming rates of over-reliance (M = 3.85, SD = 0.72) are high. The multiple regression determination found the use of ChatGPT to be the most positive predictor of academic outcomes (β = .62, p = .001, R² = .52), and over-reliance was a significant negative predictor (β = −0.31, p = .002). The study expands on the concept of the Performance-Dependency Paradox, according to which, even though AI tools can potentially enhance academic performance, they are also a threat to the development of the ability to independently think critically. Such results are placed in the Technology Acceptance Model (TAM) and are explained in connection with the shortcomings of convenience sampling with suggestions of further stratified, mixed-method studies.