We begin with the premise that Artificial Intelligence (AI) represents a change of era, even beyond the transformation brought about by the Digital Era and therefore it has a profound impact on society as a whole and, in particular, on Higher Education. After justifying this hypothesis, we first analyze the role of Artificial Intelligence, and in particular, Generative Artificial Intelligence, as a complement to educational materials, as an intelligent tutor, and as a participant in student forums. Indeed, AI can accelerate and support many teaching-learning processes. Secondly, we analyze how AI affects all aspects of universities, forcing a deep reflection on what the university of the future should be like, taking eight fundamental key factors as a reference. The creation of organizational structures within universities to support this change, by training faculty, students, and administrative staff, is essential, as is the adoption of innovative educational methodologies, the structuring of the educational offering, and the adaptation of spaces to carry them out. To analyze the impact on universities, we compare how these eight key factors have undergone transformation, first through the evolution and integration of digital technologies and subsequently through the evolution and integration of Artificial Intelligence.
The development of generative artificial intelligence (AI) has become one of the emerging fields in educational technology in recent years. Therefore, this study aims to design two generative AI tools, based on educational chatbots, that contribute to the improvement of teaching-learning processes in higher education, as well as to evaluate their impact from their users' perspective in the context of a MOOC offered by UNED, considering the effect of sociodemographic variables and previous experience with AI. To this end, after designing the two tools, an ex post facto, non-experimental study was conducted, in which the perceptions of 1,302 users of the MOOC were collected and analyzed. The results show a positive assessment of the potential of these tools in relation to overall satisfaction and their contribution to learning, along with a general tendency to reject the replacement of humans by educational chatbots. Moreover, the analysis of the role of sociodemographic and AI usage variables highlights a particular importance of both familiarity with AI and age in shaping these perceptions. In conclusion, both chatbots seem to be tools with great potential to enhance teaching-learning processes in higher education, as evidenced by the high levels of satisfaction reported by participants. However, it is essential to continue developing and updating these tools, taking into account users' feedback, in order to maximize their educational potential.
The irruption of artificial intelligence (AI) in education not only optimizes academic tasks but also transforms educational practices and university learning. The present study analyzes functional dependence on AI as a change in the way academic work is approached and conceptualized. The results make it possible to propose an interpretative conceptual structure of this construct. The study was conducted using a qualitative approach, supported by the foundations and procedures of Grounded Theory. The methodology included semi-structured interviews with 265 undergraduate students from the Universidad Nacional Jorge Basadre Grohmann (Tacna, Peru). The data were processed in ATLAS.ti and analyzed through open, axial, and selective coding. The results reveal five dimensions of the AI dependence construct: (1) preference for use over traditional sources, (2) academic functionality, (3) creative and intellectual support, (4) critical competencies and quality control, and (5) operational technological need. The latter assumes the role of the central category, showing that the practical usefulness of AI is perceived as an indispensable tool for planning, drafting, and validating academic tasks. The emergent conceptual structure describes a transition from instrumental efficiency toward cognitive delegation and reduced critical autonomy, evidencing a structural change in university academic practices. These findings invite reflection on the implications for critical literacy and autonomous regulation in the academic use of AI.
Este estudio exploratorio, de enfoque de caso, analiza comparativamente la calidad textual y el desempeño en actividades de evaluación en línea de distinto nivel cognitivo ante la disponibilidad de inteligencia artificial generativa (IAG) en educación superior. Se examinan seis producciones textuales: dos respuestas de un estudiante en línea (caso típico, rendimiento medio-alto) y cuatro textos generados por dos modelos de IAG (uno de acceso gratuito y otro de pago), todos elaborados a partir de dos enunciados reales de evaluación continua (uno más expositivo y otro más abierto y evaluativo). Se aplicó un diseño mixto que combina una matriz lingüístico-discursiva (cohesión, coherencia, contenido, claridad, concisión, corrección lingüística y adecuación al enunciado) con indicadores automatizados de complejidad y cohesión (p. ej., TAACO y MultiAzterTest), junto con valoración independiente y triangulación consensuada. Los resultados sugieren patrones de diferencia entre las producciones analizadas: el texto del estudiante muestra mayor contextualización, juicio crítico y agencia comunicativa, especialmente en la tarea de mayor demanda cognitiva, mientras que las IAG tienden a generar respuestas formalmente correctas pero más estandarizadas y menos situadas. El modelo de pago presenta un desempeño intermedio superior al gratuito. En conjunto, los hallazgos aportan indicios para revisar el diseño de tareas y criterios de evaluación en contextos mediados por IAG, y orientan futuras investigaciones con muestras ampliadas y procedimientos de fiabilidad reforzados.
The expansion of artificial intelligence (AI) in education has generated new training demands for teachers that go beyond technical skills, requiring critical competencies capable of addressing the ethical, pedagogical, and social challenges posed by this technology. Understanding how teachers develop the dispositions and skills needed to integrate AI reflectively has become a priority for educational systems. This study analyzes the relationship between general self-efficacy, critical thinking awareness, and AI literacy among pre-service and in-service teachers in Chile. A non-experimental quantitative design with a correlational-comparative approach was applied to an intentional sample of 1,169 participants. Validated scales and multivariate analyses were used to examine group differences and predictive relationships. The results indicate significant differences between pre-service and in-service teachers across all studied variables. Critical thinking awareness emerged as the strongest predictor of teaching self-efficacy in both profiles, underscoring its central role in fostering autonomous and informed pedagogical decision-making. AI literacy showed differential effects: positive among in-service teachers and ambivalent among pre-service teachers, suggesting that professional experience conditions how technological knowledge contributes to perceived competence. Gender gaps were also identified. The findings highlight that strengthening critical thinking awareness is a strategic pathway for enhancing teaching self-efficacy and promoting more thoughtful, ethical, and context-sensitive integration of AI in education, especially in scenarios of accelerated digitalization.
Este trabajo presenta la implementación y evaluación del modelo "Ubícate, Transfórmate, Corónate", una propuesta metodológica centrada en el uso pedagógico de la inteligencia artificial generativa para personalizar el aprendizaje, generar contenidos, proporcionar retroalimentación formativa y reducir finalmente la carga docente. El modelo se estructura en tres etapas metodológicas: diagnóstico inicial (Ubícate), progresión personalizada (Transfórmate) y consolidación (Corónate), integrando las producciones individuales, los apuntes colectivos y el seguimiento automatizado mediante inteligencia artificial. Su implementación se ha realizado en una asignatura de matemáticas de primero de ingeniería cuya tasa de fracaso ha sido tradicionalmente muy elevada, pero su diseño es replicable a cualquier otro contexto educativo y materia. El diseño metodológico para evaluar el método, combina técnicas cuantitativas y cualitativas (aproximación mixta) incluyendo grupos de control y experimental, una evaluación motivacional inicial, el rendimiento de una prueba sorpresa y la percepción final del alumnado sobre el método. Los resultados indican que el grupo experimental mejoró significativamente en asistencia y calificaciones, con una diferencia de más de una desviación estándar respecto al grupo de control. Por otra parte, el alumnado identificó como puntos positivos la personalización del aprendizaje y retroalimentación formativa incluidas en el modelo y señaló como punto de mejora el incremento de los recursos de aprendizaje al comenzar su implementación. Se concluye que este modelo representa una arquitectura educativa viable y escalable, que permite integrar la inteligencia artificial generativa como mediadora pedagógica real.
La formación competencial en la universidad requiere alinear los modelos educativos con los resultados de aprendizaje que el alumnado debe alcanzar. La Inteligencia Artificial (IA) se presenta como un recurso con alto potencial para apoyar el rediseño curricular, orientar la docencia hacia dichos resultados y generar instrumentos coherentes con las competencias de cada asignatura. Sin embargo, persiste la falta de herramientas que favorezcan la reflexión metacognitiva del alumnado como parte del proceso evaluativo, lo que genera desajustes entre los modelos competenciales y las prácticas docentes. El objetivo de este estudio fue explorar el impacto de los cuestionarios de autoevaluación como innovación didáctica mediada por la IA para el análisis de los niveles de competencia autopercibida por el alumnado y la orientación de la enseñanza en estudios de posgrado. Se empleó un diseño mixto secuencial (CUAN–CUAL), con 1324 estudiantes y 93 docentes de una universidad privada online española. Los estudiantes completaron los cuestionarios ad hoc de autoevaluación competencial. Los docentes analizaron la utilidad del modelo mediante otro cuestionario. Se realizaron entrevistas a los coordinadores académicos y se observaron tres sesiones de clase de cada asignatura. Los resultados mostraron que los cuestionarios fortalecen la reflexión metacognitiva del alumnado y actúan como instrumento diagnóstico para el profesorado. El análisis observacional reveló una brecha entre su aceptación teórica y su aplicación práctica. Los hallazgos subrayan el potencial de la IA para orientar el diseño competencial de las asignaturas y mejorar la elaboración de instrumentos evaluativos alineados con los resultados de aprendizaje.
The emergence of Generative Artificial Intelligence (GAI)-particularly Large Language Models (LLMs) such as ChatGPT-is transforming the educational landscape, especially in the field of foreign language instruction. This article explores the potential of these technologies to automate the assessment of writing proficiency in Spanish as a Foreign Language (SFL), a task that is especially time-consuming at the beginning of university-level courses for Erasmus students. The study is based on three experiments conducted using the Spanish Learner Corpus compiled by the Instituto Cervantes. The first experiment applied a zero-shot learning approach by prompting the model with level descriptors from the Instituto Cervantes's Curriculum Plan. In the second and third experiments, the model was adjusted through finetuning using 90% and 80% of the corpus, respectively, with the remaining data reserved for testing and validation. The results indicate that the fine-tuned models significantly outperform the zero-shot configuration in identifying the correct proficiency levels of learner texts. These findings demonstrate that LLMs can be effectively employed to streamline the initial placement process in SFL courses, thus reducing the workload of instructors and improving efficiency. The study concludes that GAI can serve as a valuable complementary tool in multilingual and multicultural educational settings, provided its use is guided by sound pedagogical principles.
Academic leveling is one of the primary challenges in higher education, particularly in STEM disciplines-one that has intensified after the academic lockdowns of COVID-19 and for which traditional remedial courses have not yielded the expected results. This study evaluated the impact of an Adaptive Learning Strategy (ALS) designed to reinforce prior knowledge in 1,309 first-year students enrolled in the courses of Computational Thinking, Fundamental Mathematical Modeling, Mathematical Reasoning, and Mathematical Thinking at a private university in Mexico. Unlike conventional remedial approaches, the ALS offers a flexible, student-centered learning experience through brief adaptive modules integrated into regular courses. These modules, supported by an analytics, and facilitate data-driven instructional decision-making. The research adopted a mixedmethods approach (QUAN > QUAL) and a quasi-experimental design with matched samples to ensure group comparability. Results revealed statistically significant differences in academic performance in favor of the experimental groups compared to the control groups. Additionally, both students and instructors evaluated the ALS positively in terms of usefulness and learning experience. Findings suggest that an adaptive, integrated, and technology-mediated strategy is a promising alternative to address academic leveling in introductory university courses, offering substantial advantages over traditional remedial methods.
The emergence of Artificial Intelligence (AI) in higher education poses the challenge of integrating personalized learning with collaborative learning. Collective intelligence is presented in this study as a tool that allows both approaches to be articulated, transforming individual contributions into shared knowledge through AI-mediated technologies. Within this framework, the perception of the Kampal platform as a learning environment that simultaneously promotes student autonomy and collective knowledge construction is analyzed. A pretest-posttest quasi-experimental design was used with an initial sample of 399 digital distance higher education students and a final matched sample of 25 cases. Three dimensions were evaluated through a structured questionnaire: familiarity with the tool, perception of its effectiveness, and expectations of use or future willingness to use similar technologies. The results show a significant increase in familiarity with Kampal and willingness to use collective intelligence tools in the future, although no significant changes were found in expectations or perceptions of immediate effectiveness. Factor analysis revealed two main dimensions: familiarity and effectiveness/expectations. The findings suggest that these tools promote a favorable attitude toward their future use, although without significant changes in perceived effectiveness. This highlights the need for intentional teacher mediation. When properly supervised, AI empowers hybrid environments that integrate personalization and collaboration, where collective intelligence operates as a nexus, facilitating complex dynamics such as swarm intelligence in higher education.
This article examines the integration of generative artificial intelligence in education from a critical, historical, and ethical perspective. It highlights growing concerns about the opacity of current artificial intelligence tools, particularly in learning systems. The study adopts a metaphor-based approach to explore how technological narratives influence the adoption of educational innovations. It reviews historical metaphors used to describe educational technologies, from Multivac and Matrix to the free software Bazaar and the App Store, and proposes new conceptual frameworks that may better reflect the current context in which artificial intelligence is entering the educational sphere. Based on this metaphorical analysis, the article outlines seven fundamental ethical principles for the safe adoption of generative artificial intelligence in education, focusing on privacy, pedagogical alignment, human oversight, and technological transparency. These principles are illustrated through a practical application: the LAMB (Learning Assistant Manager and Builder) environment, an open-source software framework that enables the ethical and contextualized design of artificial intelligence-based learning assistants. The article presents real-world cases of LAMB implementation in higher education, including a controlled experience with students that demonstrates significant improvements in student autonomy and pedagogical coherence. Finally, it emphasizes how LAMB embodies the proposed ethical principles and responds to the identified critical metaphors, offering a model for technology integration centered on teacher autonomy, alignment with institutional values and practices, and meaningful student learning that prioritizes pedagogical control over technological determinism.
The accelerated integration of Artificial Intelligence (AI) in education poses new challenges for teacher training and assessment. This study presents the design, validation, and psychometric analysis of the AI-ED-SAT questionnaire, a self-assessment tool designed for teachers to diagnose their level of preparedness in the pedagogical, ethical, and curricular use of AI. The AI-ED-SAT questionnaire was validated within the Design-Based Research approach. It comprises four dimensions aligned with UNESCO's current AI competency frameworks (2024a, 2024b): conceptual understanding of AI, pedagogical use of intelligent tools, ethical and critical reflection, and curricular integration. Its construction was based on an exhaustive theoretical review and an iterative validation process using the Delphi method with 12 experts in educational technology, AI, and teacher training. Subsequently, in a pilot test involving a sample of 128 teachers from various educational levels, the reliability (Cronbach's alpha = 0.93), content validity, and construct validity were analyzed. The results of the exploratory factor analysis (EFA) confirmed the grouping of items into four theoretical factors, while the confirmatory factor analysis (CFA) showed excellent fit indices (CFI = 0.96; RMSEA = 0.045). The AI-ED-SAT is a robust and up-to-date tool that is useful in both research and teacher training programs. Its self-reflective approach helps strengthen teachers' critical literacy and professional agency in the face of the challenges posed by AI in education.
This study analyzes the impact of the formative use of generative artificial intelligence (AI) on the development of digital competencies in university students. The intervention was implemented through a randomized controlled trial research design. The experimental group received training aimed at strategically using generative AI models to complete academic tasks, while the control group carried out the same activities without specific AI guidance. The impact was assessed using a difference-indifferences model with fixed effects, based on pre-and post-intervention questionnaires. Competences were analyzed according to the European DigComp 2.2 framework, covering four main competence areas: information and data literacy, communication and collaboration, safety, and problem solving. The results show statistically significant improvements in information and data literacy and in problem solving, both in their functional and metacognitive dimensions. Differential effects were also identified depending on the initial level of digital competence, with more pronounced gains among students with lower prior proficiency, who showed significant progress across all evaluated competencies. These findings suggest a compensatory effect of the didactic use of AI, capable of reducing gaps and promoting more equitable and inclusive learning processes. The study supports the guided integration of emerging technologies in higher education to strengthen digital competencies.
The advent of Generative Artificial Intelligence (GenAI) in education presents opportunities, but it also raises ethical and pedagogical challenges. In this context, it is imperative to comprehend how pre-service teachers perceive this technology. The present study analysed the self-perception of 174 pre-service teachers regarding the application of GenAI in education. Seven dimensions (Familiarity, Relevance, Practical Skills, Barriers, Confidence, Ethical-Social Impact, and Expectations) were measured in relation to GenAI. In addition, the usefulness of ChatGPT as a tool for designing Learning Situations (LSs) was assessed after a training experience with this system. Descriptive statistics and Spearman correlations were calculated, and a network of correlations between the seven dimensions was visualised. Differences between degrees were also explored. The findings indicated medium-to-high levels of self-perception, suggesting a very positive evaluation of ChatGPT's usefulness and a high level of satisfaction with its use. Confidence emerged as a central node in the correlation network, exhibiting close associations with Relevance, Barriers, Ethical-social impact, and Expectations. This underscores its pivotal role in the adoption of these technologies. Similarly, most participants adopted a critical stance towards GenAI, checking the responses generated by ChatGPT rather than passively accepting them. In conclusion, while there is a favourable attitude towards integrating GenAI into education, future teachers demand specific training to use it pedagogically and express concern about the ethical implications of such integration.
Assessment for learning has become increasingly important in university teaching, particularly regarding the feedback process. However, there is still a perception of student dissatisfaction with the quality of feedback provided by faculty, highlighting the need to innovate in feedback strategies. This study aimed to explore the pedagogical and technological relevance of integrating Wilson's Feedback Ladder with generative artificial intelligence, specifically GPT-4o, to strengthen formative feedback in university students. The study was conducted using a qualitative and exploratory approach in two phases. First, a prompt was designed and validated using the Delphi method with the participation of eight experts in assessment and artificial intelligence, applying it to seven state-of-the-art language models. In the second phase, the validated prompt was implemented in two university courses of different nature, Assessment for Learning and Data Structures, integrating automatic feedback into the Moodle platform. The results showed that the experts agreed on the suitability of AI-mediated Wilson's Ladder and highlighted the superior performance of GPT-4o. At the classroom level, students valued the clarity, usefulness, and immediacy of the feedback, although they identified limitations in the tool's lack of contextualization and impersonal tone. It is concluded that the integration of Wilson's Ladder with generative artificial intelligence represents a promising innovation, but one that requires disciplinary adjustments, teacher supervision, and careful attention to the human dimension of feedback in elearning contexts.
Introduction: The study analyses the barriers perceived by university faculties to integrate immersive technologies (AR, VR and MRI) in higher education. Despite their pedagogical potential, these technologies face technical, pedagogical, economic, institutional and ethical/social barriers. Methodology: A quantitative, non-experimental approach was used by means of an online survey of 775 teachers from Spanish and Latin American universities with experience in XR technologies. The instrument included socio-demographic variables and 23 items on perceived difficulties, analysed by Multiple Correspondence Analysis (MCA). Results: The MCA identified four teacher profiles: (1) critical and experienced, (2) average or transitional, (3) technopositive or innovative, and (4) selective critical. Economic difficulties were the most prominent, followed by pedagogical and technical difficulties. Perceptions varied according to age, gender, subject area and institutional context. Discussion: Barriers are not homogeneous and respond to structural and cultural factors. Technical and economic difficulties affect older teachers or those from institutions with fewer resources. Pedagogical and ethical barriers are of particular concern to teachers in the humanities and social sciences. Institutional resistance to change also emerges as a key obstacle. Conclusions: The study evidences the need for differentiated training strategies and institutional support. It is recommended to move towards longitudinal and qualitative research that delves into the evolution of these perceptions and the impact of educational innovation policies.
In recent years, advances in information technology and the Internet have transformed education by creating new opportunities for learning and collaboration in the production of Digital Educational Resources (DERs). Both students and teachers are leveraging the potential of models, methods, and approaches derived from Generative Artificial Intelligence (GenAI). The study analyzes and describes a microlearning-based methodology to guide interaction between actors and support the co-creation of Microlearning Resources (abbreviated microDERs) using GenAI. Adopting a qualitative approach, the study integrates Participatory Action Research (PAR) and the Creative Social Problem Solving (CSPS) methodology. It unfolds in three phases: (1) recognizing the context and conducting exploratory, participatory, thematic delimitation; (2) collaboratively designing microDERs with high school and higher education students and teachers, including the evaluation of GenAI tools; and (3) validating the methodology through co-creation workshops, bootcamps, and qualitative analysis. Data are collected and analyzed through co-creation workshops with students and teachers from elementary, high school, and higher education. Findings on the contributions of the actors in each phase indicate that combining PAR and CSPS improves interaction and knowledge building in the co-creation of DERs using GenAI.
In the current context of digital education and open science, university professors are not only knowledge creators through their scientific and professional output but are also expected to disseminate it to both academic and non-specialist audiences. While related to digital competence, media literacy provides a critical-communicative perspective essential for effective science dissemination, particularly via digital social networks (DSNs). This study designed and validated two constructs to assess teachers' media competence (TMC): one focused on general DSN use and the other specifically on LinkedIn. Both models integrate the Common Framework for Digital Teaching Competence 2.2 and the media literacy model by Ferr & eacute;s and Piscitelli (2012). The validation process combined theoretical review, expert judgment (n = 30), and exploratory factor analysis (EFA). Data suitability was confirmed through KMO (> 0.80) and Bartlett's test of sphericity, with both instruments showing high reliability (alpha > 0.85; omega> 0.87). The EFA identified six theoretical dimensions, explaining 78.2% of the variance for DSN and 78.8% for LinkedIn. These findings provide an initial approximation of the TMC structure and its potential for diagnostic use in teacher training, highlighting LinkedIn's strategic role as a professional environment for academic dissemination. The incorporation of communicative and media strategies in teacher education is underscored. Future research should include confirmatory factor analysis with larger samples to consolidate these initial results.
This study presents the implementation of the PathRAG model within an adaptive, hybrid, and inclusive learning environment, supported by generative artificial intelligence. Aligned with the Sustainable Development Goals (SDGs), the proposal aims to personalize university-level learning through dynamic and equitable educational pathways. The objective is to address student diversity while reducing access and participation gaps through advanced educational technology. A quasi-experimental design was applied to a sample of 52 students enrolled in a Master's program in Inclusive Education at a Spanish university. The intervention was developed in a hybrid format, combining the PathRAG algorithm with generative AI tools (GPT-3.5 turbo). Key indicators such as active participation, competence development, perceived inclusion and equity, and overall student satisfaction were assessed. Findings show significant improvements in active engagement, skill acquisition, and inclusive perception, especially among students with special educational needs or limited technological access. Overall satisfaction was high, particularly regarding the usefulness of personalized learning paths. The study concludes that PathRAG fosters more equitable and adaptive learning processes. Nevertheless, limitations such as the absence of a control group, short duration, and lack of validated instruments are acknowledged. Future research should involve controlled designs, broader samples, and longitudinal approaches. This work highlights the transformative potential of generative AI in promoting sustainable and inclusive educational models.
Introducción: El estudio analiza las barreras percibidas por el profesorado universitario para integrar tecnologías inmersivas (RA, RV y RM) en la educación superior. A pesar de su potencial pedagógico, estas tecnologías enfrentan obstáculos técnicos, pedagógicos, económicos, institucionales y éticos/sociales. Metodología: Se empleó un enfoque cuantitativo, no experimental, mediante encuesta online a 775 docentes de universidades españolas e iberoamericanas con experiencia en tecnologías XR. El instrumento incluyó variables sociodemográficas y 23 ítems sobre dificultades percibidas, analizados mediante Análisis de Correspondencias Múltiples (ACM). Resultados: El ACM permitió identificar cuatro perfiles docentes: (1) crítico y experimentado, (2) promedio o de transición, (3) tecnopositivo o innovador, y (4) crítico selectivo. Las dificultades económicas fueron las más destacadas, seguidas de las pedagógicas y técnicas. Las percepciones variaron según edad, género, área disciplinar y contexto institucional. Discusión: Las barreras no son homogéneas y responden a factores estructurales y culturales. Las dificultades técnicas y económicas afectan más a docentes mayores o de instituciones con menos recursos. Las pedagógicas y éticas preocupan especialmente a docentes de Humanidades y Ciencias Sociales. La resistencia institucional al cambio también emerge como un obstáculo clave. Conclusiones: El estudio evidencia la necesidad de estrategias diferenciadas de formación y apoyo institucional. Se recomienda avanzar hacia investigaciones longitudinales y cualitativas que profundicen en la evolución de estas percepciones y en el impacto de las políticas de innovación educativa.
