This study examined differences associated with participation in a multimodal immersive instructional sequence designed to support artificial intelligence literacy and metacognitive self-regulation in pre-service teacher education. A quantitatively dominant quasi-experimental pretest–posttest design with an embedded qualitative descriptive component was used. Sixty fifth-year teacher education students from two intact academic sections participated: one section served as the active comparison group (n = 30) and the other received the multimodal intervention (n = 30). The intervention integrated generative AI tasks, immersive virtual reality, haptic interaction, guided reflection, and metacognitive scaffolding across eight sessions. The comparison section addressed the same content and learning objectives through readings, videos, platform activities, and non-immersive simulations. Individual-level ANCOVA yielded statistically significant adjusted differences between sections after controlling for pretest scores. However, treatment condition was completely confounded with section membership, only one section represented each condition, and the analyses could not separate intervention-related differences from section-level influences. Outcomes were measured primarily with study-specific self-report scales supported by expert content review and internal consistency evidence; therefore, the very large standardized differences should be treated as exploratory and potentially influenced by common-method bias, novelty, motivation, and demand characteristics. Reflective journals and brief interviews illustrated critical evaluation of AI outputs, bias identification, planning, monitoring, reflective evaluation, and pedagogical transfer. The findings document the feasibility and promise of an integrated pedagogical sequence, but they do not establish causal effects of immersive VR, haptic interaction, or any individual component.
With the increasing reliance on Information and Communication Technologies (ICTs) in education, the satisfactory access of all people, including those with various disabilities, needs to be ensured. Virtual Learning Environments (VLEs) are interactive learning applications, broadly used in all levels of education, but many of them do not embrace those people that depend on inclusive design of web applications or support for assistive technologies, leading to exclusion and distancing from digital interactive learning. By analyzing and interpreting the literature in this context and its associated artefacts, it is possible to delimit, in terms of accessibility and User Experience, what is currently being done, who is being reached and how it is being evaluated. A Systematic Mapping (SM) was carried out with the goal of gathering this information. 24 works were considered in the SM, and by their interpretation, it was found that cognitive and motor disabilities were less addressed in those VLEs that are intended to be accessible, few occurrences of studies related to resources corresponding to the "Operable" and "Understandable" principles of WCAG, as well as a frequent use of automatic accessibility evaluation software. This study contributes by identifying trends and gaps in the inclusiveness of VLE design, so that they can be used to further improve accessibility in learning applications.
Educational robotics and block-based programming are widely used in STEM education, yet their effects depend strongly on instructional design rather than on technological novelty alone. This paper presents the design and multicenter evaluation of an inclusive-oriented, barrier-reducing instructional intervention for educational robotics activities based on block-based programming. The intervention reduces unnecessary technical and coordination barriers while preserving the conceptual and procedural demands of system behaviour, control logic, sensor–actuator interaction, and iterative validation. A quasie-xperimental study was conducted in Spain across ten institutions and three formal educational contexts (pre-university education, vocational training, and higher education) with 4,700 learners distributed across 210 class- or laboratory-level activity cohorts working in small collaborative teams. Process indicators and task quality were analyzed at the activity-cohort level; conceptual learning gains and learner-reported usability were analyzed at the individual level and interpreted cautiously due to the nested data structure. Activity cohorts in the intervention condition were compared against cohorts using conventional educational robotics setups. Time to first functional behaviour was descriptively lower in the intervention condition in all three contexts (89.4 to 75.1, 111.4 to 89.5, and 137.0 to 121.8 min), with inferential support reaching the conventional significance threshold in vocational training. Task completion time decreased significantly in all contexts (largest reduction: 323.9 to 245.7 min in higher education). The intervention condition also showed fewer task-related difficulties, higher task quality, higher learning gains (raw post–pre on a 0–20 scale), and higher perceived usability (1–7 scale, 4.18–4.28 to 5.42–5.56). The findings support the intervention as a barrier-reducing instructional design that improves activity flow, perceived accessibility, and learning conditions, without constituting direct evidence of demographic equity effects.
The accelerated obsolescence of electrical and electronic equipment has led to a growing accumulation of E-waste or waste electrical and electronic equipment (WEEE), posing environmental challenges and educational opportunities. This study presents and implements a curriculum for higher education (HE) that integrates computational thinking (CT), sustainability education, and educational robotics (ER) through the creative reuse of WEEE. The curriculum engages students in hands-on learning with analog circuits, guiding them through the principles of electronics, physical robot construction, algorithmic modeling, and sustainable redesign using BEAM robotics, a platform based on biology, electronics, aesthetics, and mechanics. Instead of relying on software programming, students engage with CT-related practices through embedded learning, tangible interaction, and perceptual grounding, reasoning about system behavior, physical circuit debugging, and adaptation of recovered components. The intervention was carried out with students from diverse academic backgrounds and showed consistent improvements in self-reported competencies across all assessed domains, with particularly notable improvements among those with no prior technical experience. By integrating sustainability into the learning process and basing CT on physical construction and sensorimotor feedback, the curriculum also supports the development of essential 21st-century skills such as problem solving, creativity, systems thinking, and reflective decision making. This inclusive and cost-effective model contributes to a renewed vision of science, technology, engineering, and mathematics (STEM) education that connects learning processes with awareness of circular design and real-world relevance.
This paper describes the iterative evolution of a gamification system across three successive student cohorts, applied to a web technologies game programming course within a video game development bachelor’s program. The system was designed to address low classroom and course participation, challenges in forming working groups, and a lack of continuous project work. In the first year of gamifying the course, a badge system was implemented to highlight progress, encourage participation and completion of optional exercises, and recognize individual student achievements. The second iteration integrated deckbuilding mechanics, where the course project’s requirements were defined by initial cards, and badges were repurposed as currency to acquire new cards that modified the scope. Finally, the third iteration incorporated an “Effort” system based on health point mechanics to encourage work consistency and mitigate the accumulation of work before partial deliveries (crunch). Comparative results show a significant increase in participation in forums and optional assignments, suggesting the efficacy of linking symbolic rewards with functional utilities in the course design.
This paper presents a reproducible remote-laboratory activity, designed for satellite-communications education, that repurposes the remote-access SALSA (Such A Lovely Small Antenna) radio telescope as an authentic instrument for teaching satellite radiocommunications through Global Navigation Satellite Systems (GNSS). The activity is specified as a transferable instructional artifact—with defined learning objectives, target audience, prerequisite knowledge, estimated duration, student tasks, an assessment rubric, and instructor guidance—so that other educators can adopt and adapt it directly. Using SALSA’s web-based interface, representative satellites from GPS, Galileo, BeiDou, and GLONASS were tracked, and observation logs (azimuth, elevation, and time) were collected throughout each pass. A pairwise differential method is then applied: consecutive line-of-sight directions are converted into angular displacement and angular velocity, from which mean orbital velocity and orbital period are estimated under a circular-orbit approximation. The study reports the full observational datasets for each constellation and the resulting orbital estimates, including comparison with theoretical values and relative errors. Results show physically consistent orbital parameters across constellations and highlight the strong influence of pass geometry and sampling on accuracy, with one GLONASS pass yielding markedly smaller errors under favorable conditions. The proposed workflow supports an end-to-end “measurement–model–validation” learning cycle, enabling students to engage with real satellite signals and remote instrumentation while reinforcing core concepts in RF measurements and introductory orbital mechanics.
This study examined the contribution of an immersive virtual reality environment with haptic interaction to spatial understanding in secondary Biology education. A quasi-experimental pretest–posttest design with nonequivalent groups was conducted with 60 students, assigned to an experimental group using immersive VR with haptic feedback and a control group receiving traditional instruction with two-dimensional resources. Spatial understanding was assessed through a 20-item test covering spatial visualization, mental rotation, and depth perception. Additional data were collected through immersion and usability scales and structured observations of students’ interaction with three-dimensional biological models. Descriptive results showed different performance patterns between groups: the control group presented a decrease in posttest scores, whereas the experimental group showed stability and positive descriptive gains, particularly in mental rotation and depth perception. However, the ANCOVA did not identify statistically significant differences between groups after controlling for pretest scores (p = .832). The effect size was small and uncertain (d = 0.33, 95% CI [−0.12, 0.78]), and the post-hoc power analysis indicated limited sensitivity to detect small effects. Students reported high levels of immersion, multisensory congruence, haptic feedback, and usability, although correlations between perceived experience and learning gains were low and not statistically significant. Overall, the findings suggest that immersive VR with haptic feedback is a feasible pedagogical strategy with favorable descriptive trends, but its effects on spatial understanding require confirmation through larger, longer, and more controlled studies.
This study examines whether open large language models (OLLMs) can be optimized for equitable and efficient use in higher education under resource-constrained conditions. A quantitative experimental design was implemented to evaluate five OLLMs —Falcon, Bloom, GPT-NeoX, T5, and Flan-T5—under four conditions: baseline unoptimized inference (C0), pruning only (C1), retrieval-augmented generation (RAG) only (C2), and pruning combined with RAG (C3). Each condition was tested using 50 educational queries per model across four repetitions. To justify the pruning configuration, an ablation study compared sparsity levels of 10%, 20%, 30%, and 40%, identifying 20% as the best trade-off between efficiency and response quality. The results show that pruning reduced response time by 10.7%, lowered RAM/VRAM usage by 18.9%, and increased throughput by 33.2%. Retrieval augmentation improved educational response accuracy by 1.3 percentage points, with the strongest gains observed for factual queries. Although pruning-only achieved the best efficiency results and RAG-only produced the highest accuracy, the combined condition provided the most balanced profile for realistic educational deployment. These findings suggest that moderate pruning and retrieval augmentation can jointly support lighter, faster, and more contextually grounded language-model deployment in higher education, particularly in institutions with limited computational infrastructure. The study contributes empirical evidence of technical and pedagogical feasibility under simulated deployment conditions.
This study analyzes the scientific evidence (2016–2026) regarding the integration of Generative Artificial Intelligence (GenAI) into undergraduate education, framed within a model of socio-technical tensions and multilevel governance. This systematic review followed the PRISMA 2020 protocol across five databases (IEEE Xplore, Scopus, WoS, SciELO, and Redalyc) using a Five-dimensional Boolean strategy (technological, cognitive-pedagogical, ethical, academic integrity, and governance). Following the screening of 134 records and the application of CASP criteria (threshold > 7/10), 38 articles were selected. The results reveal a hegemony of Scopus (79%) and an exponential acceleration of publications in 2025 (74%). The ecosystem is dominated by Large Language Models (45%), with primarily instrumental use in academic writing (27%) and learning support (22%). A critical trade-off was identified between operational efficiency and risks of cognitive atrophy, algorithmic bias, and privacy concerns. The research concludes that GenAI integration is a phenomenon of mediated contingency, where educational success depends on systemic alignment. A critical governance gap is identified due to the lack of operationalization of human oversight, compromising the integrity of the ecosystem. In response, a Responsible Integration Framework is proposed, underpinned by multilevel validation mechanisms. Within this framework, national regulation (Macro) is translated by the institution (Meso) so that human oversight in the classroom (Micro) transcends ethical ideals to consolidate as a functional, sovereign, and profoundly ethical pedagogical praxis.
The rapid integration of Artificial Intelligence (AI) into engineering demands a shift away from traditional competency models. This study introduces and validates the Hybrid Competency Integration Theoretical Framework (HTICH), a five-dimensional model encompassing augmented cognitive, metacognitive, interpersonal, collaborative-innovative, and ethical competencies. Using a concurrent mixed-methods design, the research included 35 interviews with Colombian engineers, analysis of national datasets from 1,247 organizations and 45,000 graduates, and psychometric testing of 312 professionals. The results confirm a five-factor structure that explains 67.3% of the total variance. Multiple regression analysis shows that the model explains 69% of AI effectiveness (R² = 0.69, p < 0.001), with strong correlations between metacognitive skills and continuous learning (r = 0.85). Significant skills gaps (24%–44%) were identified between industry requirements and academic curricula. The results support the amplification hypothesis—that AI enhances human capabilities—and provide an evidence-based framework for redesigning engineering curricula to align with the demands of Industry 4.0 in Latin America.
Hospital classrooms represent complex educational environments characterized by high variability, emotional sensitivity, and diverse learner needs. Despite increasing interest in integrating computational thinking and digital technologies in education, there remains a lack of conceptual frameworks that address the specific pedagogical and contextual constraints of hospital education. This paper proposes the CAL-HC (Coding as Another Language for Hospital Classrooms) framework, a human-centered model that integrates computational literacy, emotional learning, inclusive education, and contextual adaptation. The framework extends the Coding as Another Language (CAL) approach by positioning programming as a form of expression and meaning-making, while incorporating Universal Design for Learning (UDL) principles and coeducational perspectives to support equitable participation. The framework is illustrated through its application in a hospital classroom in Lanzarote (Spain), using a narrative-based learning design that combines unplugged activities, ScratchJr programming, and educational robotics. The study follows a practice- based research approach, drawing on teacher observations, student artifacts, and video-based reflection to analyze how the framework supports engagement, emotional expression, inclusion, and adaptability. Findings suggest that framing computational literacy as an expressive and human-centered practice enables meaningful participation in constrained educational environments. Beyond hospital classrooms, the framework may inform the design of human-centered computational learning experiences in other educational contexts characterized by vulnerability, discontinuity, or high learner variability.
Computational thinking (CT) has become a key competency in secondary education; however, identifying instructional approaches that effectively support its development remains a challenge. This study investigates the effects of structured collaborative learning on the development of CT, geometry learning, and collaborative skills in secondary school students. A quasi-experimental pretest-posttest design was employed, including an experimental group that engaged in structured collaborative activities and a control group that worked individually on equivalent tasks (n = 36 per group). Student performance was assessed through a multimodal evaluation approach that combined the Computational Thinking Test (CTT), a geometry test aligned with the instructional intervention, and a collaborative skills instrument based on the OECD framework for collaborative problem solving. Additionally, analytical rubrics were used throughout the intervention to monitor students' performance across six Scratch-based instructional guides that integrated geometry content with fundamental programming concepts. The results indicate that students who participated in the collaborative intervention achieved significantly greater gains in CT compared to those who worked individually, particularly in pattern recognition and debugging skills. Similarly, the experimental group showed greater improvements in geometry performance, especially in tasks associated with pattern recognition and debugging, compared to the control group. In addition, significant improvements were observed in all evaluated dimensions of collaborative skills.
As the demand for data-driven strategies in education grows, learning analytics dashboards have proven to be essential tools for enhancing transparency, monitoring, and decision-making at various administrative levels. This study examines the acceptance of a prototype control panel developed to provide relevant information to municipal education departments. Utilizing the Technology Acceptance Model (TAM), the study evaluates dashboard experience satisfaction, perceived ease of use, and behavioral intention to use the implemented dashboards. Data were collected through a survey that included secretaries of education, technical agents, administrative directors, pedagogical coordinators, school principals, and secretaries, after testing and analyzing a functional prototype. The results indicate a high level of satisfaction with the dashboards during phase 3 of the experiment, with an overall dashboard experience satisfaction score of 5.13 on a 7-point scale and a perceived ease of use score of 4.89. Additionally, the behavioral intention to use the dashboards was significant, with a score of 4.89. These findings indicate that participants viewed the dashboards positively as support tools for municipal educational management. The study discusses the practical implications of these results.
This article presents a comparative analysis of automatic error classification in Python programming tasks involving loops, with a focus on machine learning approaches for identifying error types. Three experiments were conducted: the first evaluated traditional machine learning algorithms; the second explored alternative kernel functions in SVM and MLP algorithms; and the third involved two multi-input deep learning models. Following the CRISP-DM methodology, we compiled a dataset of 3000 looped programming tasks, each including a problem description in English, a Python solution, and a classification based on state errors formulated by Gries theory (initial S, final E, state transformation T, and their combinations, totaling seven labels). In preprocessing, the code was partitioned using its AST to align it with Gries states, and both the problem descriptions and code snippets were embedded with Bert, Code-bert, and Graph-code-bert. Ablations were performed to approximate the optimal solutions for each algorithm as closely as possible. The results showed that traditional machine learning algorithms achieved up to 90% accuracy and an MCC of 83%, while alternative kernels marginally improved performance but did not surpass that of traditional algorithms. Deep learning models (ANN-GCB) achieved the best balance, with 94% accuracy and 91% MCC, respectively, demonstrating their superiority for this classification task. The model was validated with external sources from different repositories, yielding an 85% success rate, and with a group of 138 students, achieving a 96.7% correct or partially correct classification rate.
Over three years, the DevTech Research Group convened an international network of educational institutions across seven countries to support the localization, implementation, and evaluation of the Coding as Another Language (CAL)-ScratchJr curriculum. This collaboration culminated in the 2025 international symposium "A Palette of Virtues: A Humanistic Education through Computer Science," where practitioners and researchers reflected on their pedagogical practices using the Palette of Virtues Reflection Tool and the continuum of playpen to playground learning environments. Across diverse cultural contexts, participants examined how coding playgrounds can cultivate virtues such as curiosity, perseverance, generosity, and gratitude through creative, collaborative engagement with technology. Emerging insights suggest that when early CS education intentionally integrates technical instruction with opportunities for reflection and moral choice, it can serve as a powerful context for human formation. Framed within the broader dialogue between STEM and the liberal arts, this work positions the teaching of coding as a humanistic endeavor focused on digital formation by providing opportunities for self-directed learning, critical thinking and relational processes that lead to empowerment and self-actualization. Drawing on the tradition of virtue ethics, participants in the symposium explored the potential of teaching computer science as a path to human flourishing rather than merely job preparation or utility.
Electromagnetics courses are frequently perceived by undergraduate students as abstract and mathematically demanding, often resulting in low conceptual confidence and limited engagement. This paper reports a multi-course teaching experience conducted across three undergraduate electromagnetics-related subjects delivered at two campuses of the Universitat Polit & egrave;cnica de Val & egrave;ncia, Spain. A coherent set of student-centered instructional strategies was progressively integrated into regular teaching sessions without modifying curricular content or increasing contact hours. Pre-course questionnaires were used to characterize students' initial attitudes, expectations, and learning pReferences, revealing consistently low conceptual confidence, high expected difficulty, and a strong demand for interactive and applied learning approaches. These findings directly informed the design of the instructional intervention. Post-course results indicate systematic improvements in students' self-reported conceptual confidence, perceived applicability of electromagnetics content, and engagement across all courses and academic levels. The findings suggest that embedding lightweight but coherent student-centered strategies into standard electromagnetics instruction can significantly enhance students' learning experience without disrupting existing curricular structures.
The digital transformation of higher education has driven the incorporation of intelligent systems aimed at academic support and competency development, particularly in courses related to entrepreneurial training. In this context, this article aims to design and evaluate a prototype of an AI-based educational virtual agent built upon a Retrieval-Augmented Generation (RAG) architecture, intended to support the learning process in the Entrepreneurial Culture course. The prototype was designed and developed using an Agile methodology based on SCRUM, enabling an iterative and incremental construction process centered on the pedagogical and technological requirements of the institutional context. The virtual agent was implemented in a web environment and integrates semantic retrieval mechanisms, natural language generation, and contextual knowledge management to provide personalized guidance in business plan development and academic inquiry resolution. To evaluate the prototype's quality, functional testing and the System Usability Scale (SUS) were applied in accordance with the criteria established in the ISO/IEC 25010 standard. This analysis enabled assessment of the system's performance in terms of usability, functionality, efficiency, and reliability from the end-user perspective. The results indicate an average SUS score of 72.96, reflecting an acceptable level of usability according to international software quality standards for educational systems.
The Final Degree Project (FDP) represents the culmination of higher education studies and marks a key transition from academic training to professional practice. This paper aims to analyze the integration of sustainability, ethics, and gender perspectives in the design, development, and evaluation of FDPs, in alignment with the principles of the European Higher Education Area (EHEA). These dimensions are essential for fostering transversal competences that prepare students to address complex social and professional challenges. We propose a general framework for addressing these cross-cutting aspects throughout the different stages of the FDP process. This involves embedding them in course guides, applying them during project development, and integrating them into the final evaluation criteria. Additionally, we present two case studies that illustrate the current state of integration of these dimensions in real FDP contexts. The analysis reveals that, while some sustainability, ethical, and gender-related aspects are already considered, their implementation remains partial and inconsistent. This highlights the need for clearer guidelines, structured methodologies, and greater institutional support to ensure their effective incorporation.
Generative AI (GenAI) is rapidly transforming higher education, requiring a consequent shift in teaching methodologies. This article describes a two-year teaching experience (2024/2025 and 2025/2026) integrating the free, responsible, and critical use of GenAI into a coordinated, complex software development project for fourth-year Computer Science students. The study aimed to explore GenAI’s potential as a learning support tool across the software lifecycle and analyze student interaction patterns. The methodology evolved over two academic years: an initial exploratory phase (2024/2025) was followed by a revised implementation (2025/2026) that incorporated active training seminars focused on prompt engineering and critical analysis. Results show that GenAI is highly effective as a “companion” for automating low-level tasks such as code implementation, debugging, and test scenario specification, allowing students to focus on higher-level design and decision making. Initial findings revealed student frustration when GenAI struggled to reason about advanced project contexts, often due to a lack of prompt expertise. However, the corrective training actions successfully mitigated this frustration, significantly improving the critical interpretation of GenAI-generated outputs and the control of “hallucinations”. Usage patterns were asymmetric, with general-purpose tools dominating early stages and specialized tools gaining relevance as project complexity increased. This experience confirms the value of GenAI in promoting critical and responsible use within software engineering education, supporting a shift from a basic skill focus to one centered on analysis, integration, and high-level problem-solving.
This research proposes a guideline that outlines the redesign process of a web system by compiling a set of tools and techniques aimed at improving accessibility for users with visual functional diversity (blindness). It is worth mentioning that the study adheres to the W3C (World Wide Web Consortium) standards using automatic evaluation programs and screen readers. The paper presents the selection process for these tools and techniques, which constitute the guideline composed of five stages: (1) training on accessibility issues, (2) initial system exploration, (3) evaluation with users and experts, (4) requirements analysis, and (5) recoding. It also specifies the artifacts to be produced at each stage and how they are used in subsequent phases. The guideline was applied to the institutional system “MiUV” which is used by the Universidad Veracruzana community, including faculty members, administrative staff and students—some of whom have visual functional diversity. The results showed that the proposed process was effective in guiding the redesign to improve accessibility. However, opportunities for improvement were also identified in each phase of the redesign process.
