This study aims to test the serial mediation pathway based on the expanded Affective–Behavioural–Cognitive (ABC) model to explain Reliance on ChatGPT (RC) and Critical Use of ChatGPT (CU) in an academic context. A cross-sectional quantitative survey study was conducted among multi-ethnic students in the coastal region of Southeast Sulawesi, Indonesia, in two stages: a pilot study (n = 72) to test the instruments and a main study (n = 287) to test the model. The analysis used confirmatory factor analysis (CFA) based on CB-SEM to validate the measurement model, and partial least squares structural equation modelling (PLS-SEM) to test the structural model through 5,000 bootstrap resamples (CI 95%). The results showed that the indirect serial effect from PCU to CU via EA/AIS-ST-RC was significant, whereas the direct path from ST to CU was not. These findings confirm that attachment to ChatGPT is not directly related to a decrease in CU. CU practices are primarily formed when User Stickiness (ST) has a positive and significant effect on RC, which is functional in completing academic tasks. Thus, RC acts as a proximal mediator in the relationship between ST and CU.
This study aimed to explicitly examine cognitive patterns of prompt engineering, the role of pre-service teachers' artificial intelligence literacy (AIL) and online self-regulated learning (SRL) abilities in their use of prompting skills to develop lesson plans. The study was designed as a convergent parallel exploratory mixed-methods study involving quantitative analysis of students' data on achievement in lesson planning, SRL and AIL, and qualitative interpretative exploration of students' prompt sets based on cognitive domain and argumentation levels. The quantitative data was collected from 63 sophomore college students in a Faculty of Education through a quasi-experimental study with pre and post achievement tests, an AIL scale and an online learning self-regulation scale. The participants attended the experimental procedure with AI-PRO.ORG in a computer laboratory within two hours. Two raters analyzed a total of 1070 prompts on the basis of hierarchical level of taxonomy of cognitive domain, and prompt sets on the basis of argumentative characteristics in three levels. The thematic analysis of the prompts revealed that the prompts were categorized into seven different themes. Sets of statistical tests, including general linear model 2X2 ANOVA tests and t tests, were conducted with the participants' quantitative data revealed that the study was effective in helping participants, regardless of level of SRL and AIL, to develop knowledge of lesson planning. The study then merged quantitative and qualitative results revealing implications for generative AI based college learning environments, and perspectives to inform new studies of learning and teaching with AI applications.
Proctored online exams have become widespread, especially after COVID-19, increasing exam security and efficiency. In recent years, artificial intelligence developments have increased these systems’ importance by expanding their use. Accordingly, this study aims to provide an in-depth understanding of the conceptual structure of artificial intelligence-supported proctored online examination (AI-SPOE) research. Systematic literature review and text mining were used as methods. Thirty-two studies in Scopus and Web of Science databases were analyzed according to the PRISMA technique. While all the studies were examined in the context of content analysis, abstracts and keywords of the relevant studies were analyzed by text mining. The current study found it effective in increasing academic integrity and exam security, but problems such as privacy violations, exam anxiety, and algorithmic biases were encountered. Although methods such as biometric verification and behaviour analysis successfully detect cheating, technical difficulties, and false positives negatively affect the user experience. In the future, more inclusive designs, transparent algorithms, and alternative evaluation methods are suggested. In conclusion, the study emphasizes the potential of AISPOE systems and highlights the importance of ethical and technical improvements.
The analysis of students’ learning traces in digital environments enables a better understanding of the factors influencing their academic performance. This paper proposes a hybrid predictive model combining Machine Learning (ML), Deep Learning (DL), and Generative Artificial Intelligence (GenAI) to predict students’ academic performance (GPA) by leveraging traces from six dimensions: demographic, cognitive, social, emotional, contextual, and normative. First, the Random Forest algorithm is used to select the most relevant features. Then, a combined model based on MLP, LSTM, and XGBoost is trained to optimize prediction accuracy. Generative AI (DeepSeek) is integrated to enrich contextual data and provide personalized recommendations. Model decisions are interpreted using SHAP, allowing for explainability of the predictions. The model is evaluated on a real dataset of 1,002,393 records, built from learning traces collected from student interactions with courses hosted on a Moodle platform. The results highlight a significant improvement in predictive performance (a coefficient of determination of 87%, an average precision of 89.2%, an overall recall of 87.5%, an F1-score of 88.3%, and an accuracy of 90.4%) compared to traditional approaches. This work underscores the value of hybrid techniques for advanced educational data analysis. The results open promising perspectives for integrating intelligent learning support systems, as well as for adapting and extending the model to other educational contexts, learning platforms, and academic levels.
Feedback plays a critical role in learning; however, its structure and implementation in distance education differ significantly from traditional educational practices due to the unique nature of online environments. This necessitates a systematic investigation into feedback mechanisms in distance education. This systematic review aims to explore the literature on feedback in distance education, focusing on the types, purposes, and timing of feedback, as well as their implementation and impact on students. We conducted a systematic search of relevant databases on May 5, 2024, without applying a publication year restriction, thereby including all eligible studies published up to that date in accordance with PRISMA guidelines. After applying inclusion and exclusion criteria, we included 91 studies and subjected them to content analysis. Existing literature guided the categorization of feedback, examining three dimensions: feedback types, purposes, and timing. Automated systems emerged as the most common source of feedback. Formative feedback was the predominant type, evaluation was the primary purpose, and academic performance was the most frequently measured outcome. Findings highlight the emphasis on feedback’s role in 1 improving student learning outcomes, engagement, and instructional practices. This review synthesizes key findings on feedback mechanisms in distance education, focusing on their critical role in enhancing learning outcomes. It underscores the need for innovative feedback strategies that balance personalization and scalability to address diverse learner needs across contexts. Future research should explore adaptive feedback approaches to ensure inclusivity and effectiveness in distance education settings.
