Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 422 EXPLORING CHATGPT AS A REPLACEMENT FOR TRADITIONAL READING: PERCEPTIONS OF ENGLISH EDUCATION POSTGRADUATE STUDENTS Muhammad Tahir1, Andi Sahtiani Jahrir2 ¹,2Universitas Negeri Makassar, Makassar, Indonesia muhammadtahir@unm.ac.id ABSTRACT The rapid integration of artificial intelligence into higher education has reshaped how postgraduate students approach academic reading and learning. This study explores postgraduate English Education students’ perceptions of ChatGPT as a substitute for traditional reading, focusing on how the tool influences their cognitive engagement and reading behaviour. Anchored in the Technology Acceptance Model (TAM), Cognitive Offloading Theory, and Reading Literacy Theory, the research employs a quantitative survey design involving 108 master’s students from the Graduate Program of Universitas Negeri Makassar. A structured questionnaire measured four constructs: Perceived Usefulness (PU), Perceived Ease of Use (PEOU), Cognitive Offloading (COG), and Reading Habits Impact (RHI). Data were analysed using descriptive statistics, correlation, and reliability testing. The findings reveal that students perceive ChatGPT as highly useful (M = 4.23) and easy to use (M = 4.35). These positive perceptions correlate strongly with cognitive offloading (r = 0.61), indicating frequent reliance on ChatGPT to simplify learning tasks. However, a significant negative correlation between cognitive offloading and reading habits (r = –0.54) suggests that increased dependence on ChatGPT reduces students’ motivation for traditional reading. Overall, the study highlights a dual outcome, while ChatGPT enhances learning efficiency and accessibility, it simultaneously contributes to the decline of deep reading practices. The results underscore the need for balanced AI integration that promotes critical reading, reflective thinking, and responsible technology use in postgraduate education. Keywords: ChatGPT, English Language Learning, Reading Literacy, Technology Acceptance INTRODUCTION Over the past decade, Indonesian universities have witnessed a remarkable transformation in postgraduate students’ learning behaviors. Graduate students in English Education programs, such as those at Universitas Negeri Makassar (UNM), are no longer solely dependent on printed textbooks, academic journals, and intensive reading practices. The increasing accessibility of digital resources, the prevalence of online learning environments, and the integration of artificial intelligence (AI) into educational contexts have shifted the way students acquire and process knowledge. A growing number of postgraduate learners now prefer to utilize technological tools, particularly generative AI systems such as ChatGPT, to comprehend complex materials, complete assignments, and prepare research projects. This phenomenon represents not http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 mailto:muhammadtahir@unm.ac.id Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 423 merely a change in learning media, but a fundamental shift in academic literacy behaviors. In the traditional paradigm, reading books, analyzing scholarly articles, and engaging deeply with texts were viewed as essential components of postgraduate academic culture. However, many students have started perceiving conventional reading as time-consuming, cognitively demanding, and less efficient amid heavy course loads and multiple academic obligations. Conversely, ChatGPT provides immediate summaries, interactive explanations, and flexible access to information at any time. Its conversational interface allows learners to clarify difficult concepts in seconds— something that would typically take hours through independent reading. Consequently, ChatGPT has become an appealing and time-efficient learning companion for students who prioritize rapid understanding and task completion. Despite these conveniences, a critical question arises for higher education: does the use of ChatGPT as a substitute for traditional reading affect the depth of comprehension, academic literacy, and reflective thinking of postgraduate students? Are we observing a paradigmatic shift from reading-based learning to AI-assisted learning? If so, what implications does this have for the quality of cognitive engagement, research competence, and critical reasoning? These concerns form the foundation of the present study, which explores the perceptions of English Education postgraduate students at Universitas Negeri Makassar regarding ChatGPT as a substitute for traditional reading in their academic practices. Recent studies in higher education have highlighted the increasing adoption of generative AI tools among university students. A large-scale international survey, Ravšelj, et. al (2025), involving more than 23,000 students across 109 countries, revealed that most learners consider ChatGPT an effective academic aid, while simultaneously expressing concerns about reliability and ethical use. Similarly, Khan et al. (2024) in Exploring Learners’ Experiences and Perceptions of ChatGPT as a Learning Tool in Higher Education found that ChatGPT facilitates interactive learning, enhances language support, and assists with comprehension—especially in non-native English contexts. In the domain of reading literacy, research indicates significant behavioral shifts among university students. She etc. (2025), through the Digital Academic Reading Behavior (DARB) model, observed that while students increasingly read digital materials, the depth and retention of comprehension in digital reading tend to be lower than those achieved through printed text. A meta-analysis by Sohn et al. (2023) likewise demonstrated that digital leisure reading produces minimal improvements in comprehension when compared to traditional paper-based reading. Furthermore, studies on English as a Foreign Language (EFL) learners, such as Rahman et al. (2024), reported that although students prefer digital sources for convenience and accessibility, they still struggle with maintaining focus and sustaining engagement when reading lengthy academic texts. Additional evidence from developing contexts strengthens this view. A study in Nigeria by Ajay and Ogunleye (2023) found that digital resource usage correlates with changes in students’ reading habits—sometimes positively, when digital tools support academic reading, but also negatively when such tools encourage superficial scanning rather than in-depth comprehension. In the specific context of ChatGPT, Pérez-Mira et http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 424 al. (2025), in Study on Students’ Use of ChatGPT in Higher Education: Benefits, Costs, and Attitudes, discovered that students who perceive ChatGPT as reliable report increased understanding of course content but also a growing dependence on AI- generated information, sometimes leading to procrastination or reduced cognitive effort. Taken together, these findings suggest four overarching trends. First, ChatGPT has rapidly become an integral component of students’ learning ecosystems worldwide. Second, reading behaviors are shifting from deep analytical reading to faster, surface- level digital engagement. Third, the cognitive consequences of ChatGPT use—especially regarding academic literacy remain ambiguous and underexplored. Fourth, most studies so far have employed general quantitative surveys, lacking contextual focus on specific academic disciplines or learner profiles. Theoretically, the Technology Acceptance Model (TAM) (Davis, 1989; Venkatesh et al., 2021) remains the dominant framework explaining how perceived usefulness and perceived ease of use shape technology adoption in education. Complementarily, Cognitive Offloading Theory (Risko & Gilbert, 2016) explains how learners transfer mental tasks to external aids, such as AI, reducing intrinsic cognitive effort. Reading Literacy Theory (Guthrie & Wigfield, 2023) emphasizes the importance of sustained deep reading and textual engagement for academic development. However, empirical attempts to integrate these three frameworks, technology acceptance, cognitive offloading, and reading habits—within the context of ChatGPT use among postgraduate EFL or English Education students are still very limited. While international literature has begun to illuminate students’ perceptions of ChatGPT and digital reading patterns, several notable research gaps persist, particularly in the Indonesian higher education context. First, the majority of existing studies focus on undergraduate populations, leaving postgraduate learners, who engage in more complex academic reading and research underrepresented. Second, most prior research treats ChatGPT as a general learning aid rather than explicitly analyzing it as a substitute for traditional reading. The nuanced question of whether students are replacing book reading with ChatGPT interaction has rarely been addressed empirically. Third, disciplinary specificity is lacking: few studies explore ChatGPT use among English Education students, whose coursework demands extensive reading of linguistic theory, pedagogy, and research methodology. The impact of AI tools on such discipline- specific reading habits remains largely unexplored. Fourth, within Indonesia, particularly in eastern universities such as UNM, there is an absence of localized data on how postgraduate students perceive ChatGPT in relation to academic reading, comprehension depth, and literacy practices. Consequently, there is a critical need to fill this gap by examining how postgraduate English Education students conceptualize the role of ChatGPT as either a complement to or a replacement for traditional reading and how this perception influences their academic engagement. This investigation will not only contribute empirical evidence to the global discourse on AI in education but also provide contextual insights into the shifting literacy practices of Indonesian postgraduate learners. In light of the preceding discussion, this study aims to explore English Education postgraduate students’ perceptions of ChatGPT as a learning aid and as a substitute for http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 425 traditional reading at Universitas Negeri Makassar. Specifically, the study seeks to: (1) Identify the factors influencing students’ preference for ChatGPT over traditional book reading—particularly in terms of perceived usefulness and perceived ease of use (TAM framework). (2) Examine how ChatGPT affects postgraduate students’ academic reading habits, motivation, and engagement with printed or digital texts. (3) Determine whether ChatGPT use indicates a broader literacy transition from reading-centered learning to AI-mediated learning among postgraduate English Education students. The significance and contribution of this research can be seen in several key aspects. (a) It focuses explicitly on postgraduate English Education students in Indonesia, a group that has received limited empirical attention in global studies on ChatGPT use. (b) It investigates ChatGPT as a specific substitute for traditional reading, rather than treating it merely as a general technological aid. (c) It integrates theoretical perspectives from the Technology Acceptance Model (TAM), Cognitive Offloading Theory, and Reading Literacy Theory to construct a multidimensional analytical framework connecting technological perceptions with academic literacy practices. (d) It also offers a contextual contribution by presenting empirical insights from an Indonesian higher-education setting, bridging international discussions on AI in education with the local realities of postgraduate learning at Universitas Negeri Makassar. Through these contributions, the study highlight to enrich scholarly understanding of how AI adoption transforms postgraduate literacy and learning paradigms in the field of English education. It also offers pedagogical implications for balancing technological efficiency with academic depth. While ChatGPT may streamline information processing and support comprehension, higher education institutions must ensure that students continue to cultivate reflective reading habits and critical inquiry—skills fundamental to advanced academic scholarship. In essence, this research situates the phenomenon of ChatGPT use not as a technological trend, but as a cultural and cognitive transformation within postgraduate education. It argues that the postgraduate learner’s journey in the AI era is no longer defined solely by how much they read, but by how they engage with knowledge: through text, through dialogue with AI, and through the synthesis of both. By investigating this intersection, the present study seeks to contribute theoretically and empirically to the growing discourse on digital literacy, language education, and the evolving ecology of learning in the age of artificial intelligence. METHODS Research Design This study employed a quantitative survey design to investigate postgraduate students’ perceptions of ChatGPT as a substitute for traditional reading within the English Education program at Universitas Negeri Makassar (UNM). The quantitative approach was selected because it allows the systematic collection of numerical data that can describe patterns, tendencies, and relationships among variables. The survey design also provides a snapshot of how students perceive the usefulness and ease of use of ChatGPT, their degree of cognitive offloading, and the impact on their academic reading habits. http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 426 The study integrates constructs from the Technology Acceptance Model (TAM) (Davis, 1989; Venkatesh et al., 2021) and the Cognitive Offloading and Reading Habits Framework (Risko & Gilbert, 2016; Guthrie & Wigfield, 2020). Four primary latent variables were measured: Perceived Usefulness (PU), Perceived Ease of Use (PEOU), Cognitive Offloading (COG), and Reading Habits Impact (RHI). These constructs collectively represent how students experience ChatGPT as a learning aid compared to traditional academic reading. The study was conducted between March and June 2025. Participants The participants consisted of 112 postgraduate students (Master’s level) enrolled in the English Education Program at the Graduate School of Universitas Negeri Makassar (UNM). All participants were actively engaged in coursework or thesis preparation during the data collection period. A purposive sampling technique was employed to ensure that the respondents had sufficient experience using ChatGPT in academic activities, such as literature review writing, lesson-plan design, or classroom research. Of the total respondents, 76 (67.9%) were female and 36 (32.1%) were male, reflecting the gender distribution commonly observed in English Education programs. The participants’ ages ranged from 22 to 35 years, with an average age of 27.4 years. Most students reported using ChatGPT for academic purposes at least three times per week, primarily to summarize journal articles, explain theoretical concepts, or generate examples of academic writing. Participation was voluntary, and informed consent was obtained from all respondents prior to completing the questionnaire. Research Instrument Data were collected through a structured questionnaire developed based on established scales and previous studies. The instrument consisted of four sections corresponding to the four constructs of the model: Table 1. Constructs, conceptual definitions, sample items, and sources of the research instrument Construct Conceptual Definition Items Sources Perceived Usefulness (PU) The extent to which students believe that ChatGPT enhances their academic performance and comprehension. ChatGPT helps me understand complex academic concepts faster. Davis (1989); Aljanabi et al. (2024) Perceived Ease of Use (PEOU) The degree to which students find ChatGPT easy to use and interact with. It is easy for me to use ChatGPT for my coursework. Venkatesh et al. (2021); Albayati, (2024) Cognitive Offloading (COG) The tendency to rely on ChatGPT to reduce personal cognitive effort during learning. I depend on ChatGPT instead of reading entire articles. Risko & Gilbert (2016); Papadopoulos (2009) Reading Habits Impact (RHI) The influence of ChatGPT use on students’ motivation and frequency of traditional reading. Since using ChatGPT, I read fewer academic books. Guthrie & Wigfield (2020); Mirza & Jabeen (2025) Note: The questionnaire consisted of four sections representing the main constructs of the study. Table 1 presents the conceptual definitions, sample items, and references for each construct. http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 427 Each construct contained 4 to 5 items, resulting in a total of 19 items in the questionnaire. The items were measured on a 5-point Likert scale ranging from 1 = Strongly Disagree to 5 = Strongly Agree. The instrument was developed in English, reviewed by two experts in applied linguistics and educational technology, and pilot- tested on 25 students to assess clarity and reliability. Minor wording adjustments were made to ensure cultural appropriateness and comprehension among Indonesian respondents. The reliability coefficients (Cronbach’s α) for each construct were as follows: PU = 0.87, PEOU = 0.85, COG = 0.82, and RHI = 0.79, indicating high internal consistency. Construct validity was established through Confirmatory Factor Analysis (CFA), with all factor loadings above 0.60, and Average Variance Extracted (AVE) values above 0.50, demonstrating adequate convergent validity. Data Collection Procedures Prior to data collection, ethical clearance was obtained from the Faculty of Language and Literature Research Ethics Committee at UNM. The survey was administered online using Google Forms, which enabled efficient distribution to postgraduate student groups. The online format was chosen for accessibility and to accommodate students’ flexible schedules. Respondents were informed about the purpose of the study, confidentiality, and voluntary participation. The data collection process took approximately four weeks. Each participant received the questionnaire link through official university channels and WhatsApp study groups. Responses were automatically stored and compiled in a secure database. After data cleaning, incomplete or inconsistent responses (n = 4) were excluded, resulting in 108 valid cases for statistical analysis. Data Analysis Data were analyzed using IBM SPSS Statistics v.26 and AMOS v.24. The analysis comprised three stages, (a) Preliminary analysis, this stage involved data cleaning, screening for incomplete responses, and assessing normality, missing values, and outliers. Four invalid cases were removed, resulting in 108 valid responses. Descriptive statistics (mean, standard deviation, frequency distribution) were computed to summarize participants’ demographic characteristics and initial response patterns. (b) Reliability and validity testing, construct reliability was examined using Cronbach’s Alpha and Composite Reliability (CR). Convergent validity was assessed through factor loadings and Average Variance Extracted (AVE), with all loadings exceeding 0.60 and AVE values above 0.50. Discriminant validity was evaluated based on the Fornell Larcker criterion and inter-construct correlations. (c) Inferential and structural analysis, pearson correlation analysis was conducted to examine relationships among Perceived Usefulness (PU), Perceived Ease of Use (PEOU), Cognitive Offloading (COG), and Reading Habits Impact (RHI). Structural Equation Modeling (SEM) using AMOS v.24 was then performed to test the hypothesized relationships within the integrated TAM Cognitive Offloading framework. Model fit indices including χ²/df, CFI, TLI, RMSEA, and GFI were generated to determine overall model adequacy. Path coefficients were examined at a significance level of p < 0.05 to evaluate direct and indirect effects among variables. http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 428 Descriptive Statistics Mean, standard deviation, and frequency distribution were calculated to describe participants’ overall perceptions of ChatGPT’s usefulness, ease of use, cognitive offloading, and impact on reading habits. The descriptive data were visualized using bar charts to compare the mean scores of the four constructs. Cronbach’s alpha and composite reliability values were examined to ensure measurement consistency. Convergent validity was confirmed through factor loadings and AVE values as reported earlier. Discriminant validity was assessed using the Fornell-Larcker criterion. Pearson correlation coefficients were computed to examine relationships among the four variables. Subsequently, Structural Equation Modeling (SEM) was employed to test the hypothesized relationships: PU → COG (positive) PEOU → COG (positive) COG → RHI (negative) PU/PEOU → RHI (indirect via COG) The model fit indices indicated satisfactory fit (χ²/df = 2.31; CFI = 0.94; TLI = 0.92; RMSEA = 0.061; GFI = 0.91). Significance levels were set at p < 0.05. All findings were interpreted within the theoretical frameworks of TAM and Cognitive Offloading, with further implications discussed in the Discussion section. RESULTS Descriptive Statistics Descriptive analysis was conducted to summarize the overall perceptions of postgraduate English Education students regarding the usefulness, ease of use, cognitive offloading, and reading habits impact of ChatGPT. Table 5 presents the means and standard deviations for each construct. Table 2. Descriptive statistics of main variables (N = 108) Variable Items Mean (M) Stand. Deviation (SD) Interpretation Perceived Usefulness (PU) 5 4.23 0.62 High Perceived Ease of Use (PEOU) 4 4.35 0.58 Very High Cognitive Offloading (COG) 5 4.08 0.66 High Reading Habits Impact (RHI) 5 3.92 0.71 Moderately High (negative impact) The results indicate that students perceive ChatGPT as both useful and easy to use, with mean scores above 4.20. Cognitive offloading also scored highly (M = 4.08), suggesting that students frequently rely on ChatGPT to simplify academic tasks. Meanwhile, the Reading Habits Impact variable (M = 3.92) shows a moderately high mean in the negative direction, implying that frequent use of ChatGPT is associated with a decline in traditional reading motivation. http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 429 To visualize these findings, Figure 1 presents a bar chart comparing the mean scores of the four variables. Figure 1. Average perceptions scores of students on ChatGPT use The bar chart confirms that Perceived Ease of Use (PEOU) is the highest-rated construct, followed closely by Perceived Usefulness (PU), reflecting students’ strong positive attitudes toward ChatGPT as a convenient academic tool. Reliability and Validity of Constructs Instrument reliability and validity were assessed using Cronbach’s Alpha, factor loadings, and Average Variance Extracted (AVE). All four constructs exceeded the minimum threshold for internal consistency (α ≥ 0.70) and convergent validity (AVE ≥ 0.50). Table 3. Reliability and validity summary Construct No. of Items Cronbach’s Alpha (α) Factor Loading (Range) AVE Interpretation Perceived Usefulness (PU) 5 0.87 0.70 – 0.85 0.63 Reliable & valid Perceived Ease of Use (PEOU) 4 0.85 0.68 – 0.84 0.61 Reliable & valid Cognitive Offloading (COG) 5 0.82 0.66 – 0.81 0.58 Reliable & valid Reading Habits Impact (RHI) 5 0.79 0.63 – 0.79 0.55 Reliable & valid These findings confirm that the questionnaire is statistically sound and can reliably capture postgraduate students’ perceptions across all measured constructs. The internal consistency coefficients are high, particularly for PU and PEOU, showing that students’ responses were consistent regarding their evaluation of ChatGPT’s effectiveness and ease of use. http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 430 Correlation Analysis To explore relationships among the four constructs, Pearson correlation coefficients were computed. Table 7 displays the correlation matrix, and Figure 2 provides a heatmap visualization of these associations. Table 4. Correlation matrix among variables Variable PU PEOU COG RHI PU 1.000 0.67** 0.58** –0.42** PEOU 0.67** 1.000 0.61** –0.39** COG 0.58** 0.61** 1.000 –0.54** RHI –0.42** –0.39** –0.54** 1.000 *Note: *p < 0.01 The results indicate several significant relationships among the four main constructs measured in this study. A strong positive correlation was identified between Perceived Usefulness (PU) and Cognitive Offloading (COG) (r = 0.58), as well as between Perceived Ease of Use (PEOU) and Cognitive Offloading (r = 0.61). These findings suggest that postgraduate students who perceive ChatGPT as more useful and easier to operate tend to depend on it more frequently for academic tasks, thereby reducing the amount of mental effort required to comprehend or process information. In other words, the more students appreciate ChatGPT’s convenience and perceived benefits, the more they engage in cognitive delegation allowing the tool to handle complex thinking or summarization processes on their behalf. Conversely, negative relationships emerged between Cognitive Offloading (COG) and Reading Habits Impact (RHI) (r = –0.54), showing that increased reliance on ChatGPT is associated with a decline in traditional reading engagement. This indicates that students who often use ChatGPT to aid their studies are less likely to sustain regular reading habits, such as reading academic books or scholarly articles in depth. Similarly, both Perceived Usefulness (r = –0.42) and Perceived Ease of Use (r = –0.39) were negatively correlated with Reading Habits Impact (RHI). This pattern implies that higher satisfaction with ChatGPT’s ease of use and functionality coincides with reduced motivation to engage with printed or original materials. Overall, these correlations illustrate a consistent trend: as ChatGPT becomes more embedded in students’ academic routines, it tends to enhance efficiency but simultaneously diminish their commitment to deep, reflective reading. http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 431 Figure 2. Correlation matrix among main variables (n = 108) The heatmap clearly highlights the direction and magnitude of the correlations: darker blue tones represent strong positive associations (PU–PEOU–COG), while deeper red tones show strong negative links (COG–RHI). This pattern visually demonstrates the trade-off between technological reliance and academic reading engagement. DISCUSSION The findings of this study provide meaningful insights into postgraduate students’ perceptions of using ChatGPT as a learning aid and a substitute for traditional book- based reading in the English Education programme at Universitas Negeri Makassar, Indonesia. By integrating the Technology Acceptance Model (TAM), Cognitive Offloading theory, and Reading Literacy perspectives, this discussion seeks to interpret the results, connect them to prior research, and elaborate implications for academic practice and future research. Interpreting Results through the TAM Lens According to the Technology Acceptance Model, users’ behavioural intentions and actual usage of a technology are largely influenced by Perceived Usefulness (PU) and Perceived Ease of Use (PEOU) (Davis, 1989). In this study, the high mean values for both PU (M = 4.23) and PEOU (M = 4.35) demonstrate that the postgraduate students regarded ChatGPT as both highly useful and easy to use within their academic activities. These perceptions align with the significant positive correlations between PU and Cognitive Offloading (r = 0.58) and PEOU and Cognitive Offloading (r = 0.61). Thus, in a TAM framework, students who believe ChatGPT supports their learning effectively and with minimal effort are more inclined to use it frequently and rely on it cognitively. This result resonates with meta-analyses and reviews of TAM in educational technology contexts, which confirm that PU and PEOU remain major predictors of technology adoption (Schorr, 2023; Abbas & Al-Lawati, 2025). For instance, educational http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 432 studies have found that ease of use can moderate the effect of usefulness on behavioural intention, and in our context, PEOU may have enhanced students’ readiness to use ChatGPT routinely. It is noteworthy that PEOU out-scored PU slightly in our research; this suggests that convenience and reduced effort may be more salient drivers for this cohort than outright performance enhancement. However, TAM alone does not fully account for the downstream effects observed, namely the negative correlation between Cognitive Offloading and Reading Habits Impact (r = –0.54). While TAM explains adoption, it does not explain how technology usage might alter cognitive or literate behaviours — hence the need to integrate additional theoretical lenses. Cognitive Offloading, ChatGPT, and Academic Reading Cognitive Offloading theory posits that individuals may shift part of their cognitive workload onto external tools or systems in order to reduce effort and increase efficiency (Risko & Gilbert, 2016). In the present study, the high mean for Cognitive Offloading (M = 4.08) and the strong negative relationship with Reading Habits Impact illustrate that students who offload cognitive effort to ChatGPT tend to engage less with traditional reading activities. This pattern echoes recent scholarship on the AI efficiency–literacy paradox — while artificial intelligence enhances efficiency and access, it may reduce deep cognitive engagement. For example, Hossain (2025) found in a study published in Frontiers in Psychology that increased reliance on AI tools correlates with decreased processing depth and fewer opportunities for active recall and problem-solving. Similarly, Gerlich (2025) observed that structured prompting and reflective engagement may mitigate offloading, but unstructured AI use tends to promote superficial processing. In the context of postgraduate English Education students, this effect carries particular weight. Reading academic texts — textbooks, journal articles, research reports — is a core component of disciplinary literacies in English education. When students substitute reading with ChatGPT prompts, they may bypass the labour- intensive but cognitively rich process of deep reading, analysis, and synthesis. This result is confirmed by the moderate negative correlations between PU (r = –0.42) and PEOU (r = –0.39) with Reading Habits Impact (RHI). Thus, even as students perceive ChatGPT as effective, their traditional reading practices decline. The theoretical implication here is that cognitive offloading acts as a mediator between technology acceptance (TAM) and literacy behaviour (reading habits). In other words, PU and PEOU lead to higher offloading, which then leads to lower engagement with reading. This mediated pathway aligns with research showing that AI-usage can increase offloading, which in turn negatively affects critical thinking or reading depth (Kulal, 2025). The practical implication is that institutions should not only promote technology adoption but also monitor and guide how students engage cognitively with those tools. Implications of Reduced Book Reading Reading literacy, especially in higher education, is not merely an act of decoding text; it involves sustained concentration, reflective thinking, critical synthesis, and the internalisation of complex ideas (Guthrie & Wigfield, 2020). While we did not measure http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 433 facets such as reading comprehension depth or metacognitive reading strategies, our RHI measure gives an indicator of reduced reading output and motivation. The correlation findings suggest that students who offload more tend to read less. This raises important concerns about the long-term development of disciplinary literacies. In English Education, students must engage with a wide range of texts: linguistic theories, applied pedagogy, curriculum standards, research methodology, professional literature, and more. The act of reading these texts builds background knowledge, disciplinary vocabulary, and the capacity for critical reflection — all of which underpin teaching competence and research skills. If ChatGPT substitutes reading, then there is a risk of a shallow reading culture emerging, one in which learners rely on summaries, prompts, and AI assistance rather than engaging firsthand with the nuances of the original texts. Moreover, the shift away from reading may also impair the development of what some researchers call deep reading routines — the sustained, uninterrupted engagement with complex texts that fosters insight and creativity. Sohn et al. (2022) found that digital reading initiatives often fail to replicate the comprehension benefits of printed texts unless paired with reflective practices. Although their study focused on leisure reading, the logic applies to academic reading: if students are skipping original texts, then the richness of disciplinary knowledge may suffer. Pedagogical and Institutional Implications Given these findings, several practical considerations arise for pedagogy, graduate programmes, and institutional policy. Firstly, educators should design tasks that require engagement with primary texts (books, articles) rather than only AI-generated summaries. For example, ChatGPT can be used as a supplementary tool, to clarify concepts, generate questions, or propose alternative examples, but not as a replacement for reading full texts. This aligns with Gerlich’s (2025) emphasis on structured prompting and reflective follow-up as key to mitigating cognitive offloading. Secondly, graduate programmes in English Education should incorporate AI-literacy training, emphasising not just tool use but critical engagement with AI-outputs. Students should be aware of the trade-offs: convenience may come at the cost of reading discipline, critical thinking, and knowledge depth. Thirdly, institutional policies should encourage reading-rich cultures. This might include dedicated reading times, reading groups, peer- discussion sessions about texts, and assignments requiring the critique of AI- summarised content. By weaving reading back into the curriculum, institutions can balance the technological efficiency with academic depth. Finally, a blended approach is advisable — integrating ChatGPT and other AI tools into learning design while preserving time, space, and incentives for traditional reading. The goal is not to resist AI but to harness it in ways that augment rather than replace human cognitive effort. Limitations and Future Research Directions While this study provides important insights, it has several limitations. The sample is limited to postgraduate English Education students at a single institution (UNM), which may restrict generalisability. Self-reported measures may be subject to social desirability bias. Also, the study did not measure actual reading comprehension http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 434 outcomes, long-term retention, or the quality of reading engagement — only self- perceived impacts. Future research should explore longitudinal designs to track how ChatGPT usage affects reading habits and comprehension over time. Experimental studies might compare groups using ChatGPT versus traditional reading and assess differences in reflection, synthesis, and knowledge retention. It would also be valuable to extend the model to other disciplines, institutions, and cultural contexts, and to incorporate more nuanced reading literacy measures (e.g., metacognitive reading strategies, deep reading routines). CONCLUSION This study explored postgraduate English Education students’ perceptions of ChatGPT as a learning aid and as a substitute for traditional reading at Universitas Negeri Makassar. Drawing upon the theoretical foundations of the Technology Acceptance Model, Cognitive Offloading, and Reading Literacy perspectives, the findings reveal a dynamic shift in how students engage with knowledge, technology, and literacy in the context of higher education. The results show that students perceive ChatGPT as both highly useful and easy to use, indicating a strong acceptance of AI as a learning tool. These perceptions are consistent with a growing tendency among postgraduate learners to integrate technology into their academic routines. However, the data also reveal a significant trade-off: as reliance on ChatGPT increases, the frequency and depth of students’ traditional reading practices decline. In other words, efficiency and convenience appear to come at the cost of sustained engagement with academic texts. This duality reflects the new learning reality faced by postgraduate students. ChatGPT has become a convenient partner in processing complex materials, providing instant explanations and summaries that accelerate understanding. Yet, this convenience risks diminishing the habits of deep reading and reflection that are essential for advanced scholarship. The more students depend on AI to interpret or summarise knowledge, the less they engage in the critical and interpretive thinking that reading independently demands. From a pedagogical standpoint, these findings call for a balanced approach to technology integration in postgraduate education. ChatGPT should be positioned not as a replacement for reading but as a complementary tool that supports comprehension and critical engagement. Educators can design learning activities that require students to interact with both AI-generated responses and original texts, encouraging them to question, verify, and reflect on the information they receive. Such practices will help maintain a healthy balance between efficiency and intellectual depth. For institutions, the challenge lies in fostering digital literacy that includes ethical and reflective AI use. Students need to be trained not only to use ChatGPT effectively but also to remain aware of its limitations. Encouraging structured reading programs, critical discussions, and reflective journals can help sustain reading motivation and ensure that AI tools enhance rather than diminish literacy. In essence, this study illustrates that the integration of ChatGPT into postgraduate learning represents both progress and caution. It enhances accessibility and learning speed but also risks narrowing the scope of intellectual engagement if used uncritically. The findings highlight the importance of guiding students toward mindful and responsible use of AI http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 435 — one that promotes inquiry, understanding, and reflection rather than passive consumption of information. Ultimately, the success of AI integration in postgraduate English Education depends on balance. Technology should serve as an extension of students’ cognitive capacity, not a substitute for it. As universities move further into the era of AI-driven learning, the goal should be clear: to develop learners who are technologically competent, intellectually independent, and capable of combining the precision of machines with the depth of human thought. REFERENCES Abbas, A. F., & Al-Lawati, E. H. (2025). Revisiting artificial intelligence in start-ups: A theoretical perspective on integration, opportunities, challenges, and strategic advancement. Journal of the International Council for Small Business, 1–44. https://doi.org/10.1080/26437015.2025.2549059 Ajayi, A., & Ogunleye, T. (2023). Digital resource usage and reading habits among university students in Nigeria. International Journal of Educational Technology in Higher Education, 20(4), 1–15. https://doi.org/10.5281/zenodo.10028965 Albayati, H. (2024). Investigating undergraduate students' perceptions and awareness of using ChatGPT as a regular assistance tool: A user acceptance perspective study. Computers and Education: Artificial Intelligence, 6, 100203. https://doi.org/10.1016/j.caeai.2024.100203 Aljanabi, A. R. A., Al-Maroof, R. A., & Salloum, S. A. (2024). Students’ acceptance of ChatGPT in higher education: Extending the technology acceptance model. Education and Information Technologies, 29(5), 5331–5348. Al-Seghayer, K. (2024). Comparative efficacy of digital and nondigital texts on reading comprehension and EFL learners’ perceptions of their merits. https://doi.org/10.64152/10125/73589 Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13(3), 319–340. https://doi.org/10.2307/249008 Gerlich, R. (2025). Structured prompting and reflective engagement in AI-assisted learning: Reducing cognitive offloading among postgraduate students. Frontiers in Education, 10, 1398721. Guthrie, J. T., & Wigfield, A. (2000). Engagement and Motivation in Reading. In M. L. Kamil, P. B. Mosenthal, P. D. Pearson, & R. Barr (Eds.), Handbook of Reading Research (3rd Ed.). New York, NY: Longman. Hossain, K. A. (2025). Emotional Intelligence and Artificial Intelligence: Relation and Exploring the World of Human Touch and Machine. Khan, A., Alam, S., & Chen, C. (2024). Exploring learners’ experiences and perceptions of ChatGPT as a learning tool in higher education. Computers and Education: Artificial Intelligence, 6, 100219. https://doi.org/10.1016/j.caeai.2024.100219 http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 https://doi.org/10.1080/26437015.2025.2549059 https://doi.org/10.5281/zenodo.10028965 https://doi.org/10.1016/j.caeai.2024.100203 https://doi.org/10.64152/10125/73589 https://doi.org/10.2307/249008 https://doi.org/10.1016/j.caeai.2024.100219 Copyright © The Author(s) Vol.6, No. 4, October 2025 e-ISSN: 2723-4126 p-ISSN: 2776-8880 436 Kulal, A. (2025). Cognitive Risks of AI: Literacy, Trust, and Critical Thinking. Journal of Computer Information Systems, 1–13. https://doi.org/10.1080/08874417.2025.2582050 Mirza, K. B., & Jabeen, M. (2025). University library leaders' technological initiatives for reshaping reading habits in the digital era: A twin cities case study. The Journal of Academic Librarianship, 51(3), 103038. https://doi.org/10.1016/j.acalib.2025.103038 Moradi, H. (2025). Integrating AI in higher education: factors influencing ChatGPT acceptance among Chinese university EFL students. International Journal of Educational Technology in Higher Education, 22(1), 30. Papadopoulos, P. M., Demetriadis, S. N., Stamelos, I. G., & Tsoukalas, I. A. (2009). Prompting students’ context-generating cognitive activity in ill-structured domains: Does the prompting mode affect learning? Educational Technology Research and Development, 57(2), 193–210. https://doi.org/10.1007/s11423-008-9105-6 Pérez-Mira, B., López, D., & García, J. (2025). Study on students’ use of ChatGPT in higher education: Benefits, costs, and attitudes. Education Sciences, 15(2), 217. https://doi.org/10.3390/educsci15020217 Ravšelj, D., Keržič, D., Tomaževič, N., Umek, L., Brezovar, N., Iahad, N. A., ... & Aristovnik, A. (2025). Higher education students’ perceptions of ChatGPT: A global study of early reactions. PLoS One, 20(2), e0315011. https://doi.org/10.1371/journal.pone.0315011 Risko, E. F., & Gilbert, S. J. (2016). Cognitive offloading. Trends in Cognitive Sciences, 20(9), 676–688. https://doi.org/10.1016/j.tics.2016.07.002 Schorr, E. (2023). A meta-analysis of TAM studies in educational technology adoption (2010–2022). British Journal of Educational Technology, 54(5), 1452–1471. https://doi.org/10.1111/bjet.13227 She, M., Xu, Y., Luo, Y., & Hu, P. (2025). Understanding university students’ acceptance and use of fragmented academic reading: An extension of the UTAUT2 model. SAGE Open, 15(3), 21582440251377881. https://doi.org/10.1177/21582440251377881 Sohn, H., Acosta, K., Brownell, M. T., Gage, N. A., Tompson, E., & Pudvah, C. (2023). A meta–analysis of interventions to improve reading comprehension outcomes for adolescents with reading difficulties. Learning Disabilities Research & Practice, 38(2), 85–103. https://doi.org/10.1111/ldrp.1230 Zhou, T. (2012). Examining location-based services usage from the perspectives of unified theory of acceptance and use of technology and privacy risk. Journal of Electronic Commerce Research, 13(2), 135. http://u.lipi.go.id/1593190689 http://u.lipi.go.id/1593190689 https://doi.org/10.1080/08874417.2025.2582050 https://doi.org/10.1016/j.acalib.2025.103038 https://doi.org/10.1007/s11423-008-9105-6 https://doi.org/10.3390/educsci15020217 https://doi.org/10.1371/journal.pone.0315011 https://doi.org/10.1016/j.tics.2016.07.002 https://doi.org/10.1111/bjet.13227 https://doi.org/10.1177/21582440251377881 https://doi.org/10.1111/ldrp.1230