Tryfon Sivenas

This mixed-methods study examined secondary students' interactions with the generative AI chatbot ChatGPT-5 in a formal classroom setting, focusing on attitudes, anxiety, and responses to hallucinated outputs. Participants were 109 16-year-old students from three Greek high schools who used ChatGPT-5 during an eight-hour intervention in the course "Technology." Students engaged in information seeking, CV generation, document and video summarization, image generation, quiz creation, and age-appropriate explanations, including tasks deliberately designed to elicit hallucinations. Quantitative data were collected with the Student Attitudes Toward Artificial Intelligence scale (SATAI) and the Artificial Intelligence Anxiety Scale (AIAS); qualitative data came from semi-structured interviews with 36 students. SATAI results showed moderately positive attitudes toward AI, with stronger cognitive evaluations than behavioral intentions, whereas AIAS scores indicated moderate learning-related anxiety and higher concern about AI-driven job replacement. Gender differences in AI anxiety were small and non-significant, while female students reported more positive cognitive attitudes than males. AI attitudes and AI anxiety were essentially uncorrelated. Thematic analysis identified four pedagogical affordances (knowledge expansion, immediate feedback, familiar interface, perceived skill development) and three constraints (uncertainty about accuracy, anxiety about AI feedback, privacy concerns). After encountering hallucinations, many students reported restricting AI use to domains where they already possessed knowledge and could verify answers, a strategy termed "epistemic safeguarding." The study discusses implications for critical AI literacy in secondary education.

This study evaluates the predictive performance of advanced machine learning models, including DeepBoost, XGBoost, CatBoost, RF, and MLP, in estimating the Ω₂, Ω₄, and Ω₆ parameters based on a comprehensive set of input variables. Among the models, DeepBoost consistently demonstrated the best performance across the training and testing phases. For the Ω₂ prediction, DeepBoost achieved an R² of 0.974 and accuracy of 99.895% in the training phase, with corresponding values of 0.971 and 99.902% in the testing phase. In comparison, XGBoost ranked second with an R² of 0.929 and accuracy of 99.870% during testing. For Ω₄, DeepBoost achieved a training phase R² of 0.955 and accuracy of 99.846%, while the testing phase results included an R² of 0.945 and accuracy of 99.951%. Similar trends were observed for Ω₆, where DeepBoost obtained near-perfect training phase results (R² = 0.997, accuracy = 99.968%) and testing phase performance (R² = 0.994, accuracy = 99.946%). These findings are further supported by violin plots and correlation analyses, underscoring DeepBoost’s superior predictive reliability and generalization capabilities. This work highlights the importance of model selection in predictive tasks and demonstrates the potential of machine learning for capturing complex relationships in data.

This study presents a novel approach to predicting the factor of safety (FOS) for infinite slopes stabilized with nano-silica (NS), leveraging machine learning (ML) models to address limitations of traditional geotechnical assessments. A unique dataset, consisting of 1053 samples, was compiled to capture critical parameters, including NS content and curing time, thereby enhancing model accuracy across diverse conditions. Six ML models were evaluated, with the gradient boosting (GB) model emerging as the most robust, achieving an R ² value of 0.99, mean absolute error of 0.03 and root mean squared error of 0.04. A reliability analysis quantified a failure probability of 29.66% and a reliability index of 1.64 for NS-treated soils, underscoring the model’s applicability in real-world geotechnical design. Additionally, a parabolic regression equation was derived, offering practitioners a reliable tool for FOS estimation. To facilitate practical implementation, an intuitive graphical user interface (GUI) was developed, allowing for accurate FOS predictions based on user-defined inputs. This study provides a comprehensive, data-driven model for NS-stabilized slopes, advancing the field with a user-friendly predictive tool that supports sustainable engineering practices in challenging geotechnical environments.

The aim of this study was the examination of the perceived affordances and constraints of drones for teaching and learning. The sample consisted of 44 in-service teachers who attended an introductory presentation on drones, their technology and control mechanisms and afterwards assembled, flew and programmed four multicopter drones. Data was collected through anonymous online questionnaires. Results from qualitative data analysis revealed seven affordances, namely programming through block-based languages, recording and bird’s-eye view, viewing places that are unseen from ground level and real-time photo and video streaming, drone assembly, data collection and processing, gamification, and development of several student skills. Additionally, they revealed four constraints, namely the necessity of teacher training, time restrictions regarding the drone’s battery life, infrastructural and personal data restrictions. These findings will contribute to a better understanding of the educational value of drones for teaching and learning and, at the same time, provide a base-layer for future research in using drones for educational purposes.

This study examined in-service teachers’ perceptions regarding spatial presence, simulator sickness and usability of First-Person view drones through Head-Mounted Displays in order to determine the suitability of their use in teaching and learning processes. The sample consisted of 60 in-service teachers of primary education. Data was collected via the Temple Presence Inventory Scale, Simulator Sickness Scale and the System Usability Scale. Results showed that the teachers rated an increased level of spatial presence. Additionally, the simulator sickness was relatively low, and the drone’s usability was rated as excellent. These findings contribute to the better understanding of the potential of First-Person view drones as learning tools.