<![CDATA[This study evaluates the implementation of an Artificial Intelligence (AI)-based adaptive reading system that uses ChatGPT and examines reading comprehension gains and the predictors of self-reported deep learning among Generation Z senior high school students in Indonesia. A key feature of the system is its flexible and interactive discussion assistant, particularly its ability to paraphrase and adapt language use based on user needs. The study used a quantitative approach with a one-group pre-test–post-test design without a control group. The participants were 45 students from three senior high schools in Barru Regency, selected through purposive sampling. Data were collected using a reading comprehension test (pre-test and post-test) and a 30-item questionnaire on the adaptive reading system, reading activities, deep learning, and Generation Z characteristics. Data were analyzed using descriptive statistics, a paired-sample t-test, and multiple linear regression. Students responded positively to the system, with mean scores of 3.92 (adaptive reading system), 3.98 (reading activities), and 4.05 (deep learning). Reading comprehension scores improved significantly following implementation, from 80.93 to 88.02 (mean gain = 7.09, t(44) = −10.809, p < .001, Cohen’s d = 1.61). In a separate regression analysis, the adaptive reading system and reading activities jointly accounted for 77.6% of the variance in self-reported deep learning (R² = .776), and both were significant predictors; this indicates association, not causation. Because there was no control group, the gains are interpreted as improvement following implementation and not as an effect of the system alone. The study offers preliminary evidence for AI-supported reading instruction among Indonesian secondary-school learners and indicates that personalization combined with structured reading activity may create conditions for deeper learning.]]>
Copyrights © 2026
