Edumorfosis.it

Julio Gonzalo —físico, catedrático en Lenguaje Natural en la UNED y referente en el estudio matemático y computacional del lenguaje— conversa con Gustavo Díez, director de Nirakara, sobre cómo la inteligencia artificial está empezando a influir en nuestra mente, nuestras decisiones y la manera en que entendemos el mundo.

Con una mirada precisa y profundamente humana, Julio Gonzalo aporta décadas de investigación en modelos de lenguaje, cognición y sistemas complejos para iluminar lo que está en juego: desde la manipulación sutil a través del lenguaje hasta el modo en que la IA aprende de nosotros… y nos modifica. Una conversación imprescindible para quienes quieren comprender —y anticipar— cómo esta tecnología ya está moldeando nuestras historias, nuestra atención y nuestra identidad.

Artificial Intelligence is rapidly transforming classrooms, yet without a coherent framework connecting AI, EdTech, and neuroscience, education risks repeating the same short-sighted design patterns that have plagued every "next big thing."

Teaching is more than delivering facts. Great teachers build relationships, spark curiosity, and guide students through life’s challenges. They notice when a student is frustrated, encourage resilience, and adapt lessons to the energy in the room. Furthermore, teachers are also athletic coaches, mentors, and community advocates. These human connections are at the heart of learning—something AI cannot replicate.

La tecnología, por sí sola, no transforma la educación. Son los docentes, con una sólida formación pedagógica y un uso estratégico de las herramientas digitales, quienes tienen el poder de impactar verdaderamente el aprendizaje. Sin preparación adecuada, incluso las plataformas más avanzadas o populares como TikTok o Instagram se convierten en simples distractores en lugar de recursos educativos efectivos. Este artículo aborda cómo los docentes pueden trascender el uso superficial de la tecnología, explorando herramientas diseñadas específicamente para enriquecer la enseñanza y maximizar su impacto en el aula.

More Harvard College students than ever are passing their classes with flying colors, but the college’s evaluation system is “failing to perform the key functions of grading”, according to a report released by the Office of Undergraduate Education on Monday.

The 25- page report, which was sent to faculty and Harvard College students on Monday, found that more than 60 percent of grades awarded to Harvard undergraduates are A’s, compared to only a quarter of grades two decades ago. It concluded that Harvard’s current grading system is “damaging the academic culture of the College”. 

El más reciente documento de la UNESCO propone cambios trascendentales en el modelo educativo tradicional. De igual manera, plantea serios retos y dilemas socio-tecno-educativos de deben atenderse a la brevedad posible.

Te invito a leer el resumen de este documento y a ver el video-overview que resume los principales plantemientos de la UNESCO sobre el impacto de la IA en la Educación.

La arquitectura de internet atraviesa una metamorfosis que trasciende la mera evolución tecnológica. Lo que algunas voces especializadas denominan Web 3.5 representa, en realidad, la materialización de décadas de convergencia entre sistemas que hasta hace poco operaban en silos diferenciados. Esta fase —más transicional que definitiva— configura el sustrato sobre el que se erigirá una nueva generación de infraestructuras digitales capaz de alterar, de forma estructural, las relaciones entre individuos, organizaciones y sistemas automatizados.

Colleges are now closing at a pace of one per week. What happens to the students?” Jon Marcus asked in a recent Hechinger Report piece. It’s not a rhetorical question — and it doesn’t have an easy answer. As educators, we’ve read the headlines, seen the numbers, and felt the pressure. Undergraduate enrollment is down. Student confidence is eroding. The enrollment cliff looms. 

El punto de partida es incontestable: los «detectores de texto escrito por inteligencia artificial» no son una base válida para castigar a ningún estudiante. La evidencia lo deja claro: la propia OpenAI retiró su «AI Text Classifier« por su bajo índice de acierto, una admisión explícita de que el enfoque no funcionaba como prometía. La razón de fondo es clara: abundantes falsos positivos y gran facilidad para «engañar» a los detectores.

“Artificial Intelligence”. The issue is embedded right in the name: Artificial means not real. Is it here to stay, and can we even fight it? One of our generation’s premier philosophers, Dr. Yuval Harari, said that if we hope to survive, we better fight it (2015). Technology has been hypothesized to be an evolutionary mismatch (Li & Colarelli, 2017). This term implies that behaviors that once supported a species have become injurious. An example of a mismatch is sugary foods. Our nomadic ancestors struggled to procure enough daily calories to sustain life. When they found sugary food, they filled up on it and stored it. In our modern day, too much sugar leads to issues related to early mortality.  

I am not sure that we have quite grasped the enormity of the upheaval that artificial intelligence will bring. The disruption will be orders of magnitude bigger and faster than we have witnessed before, and we won’t have a generation or two to adapt as we had in the industrial revolution. So the question for a place like Oxford is not whether we have Blake’s ‘bow of burning gold’ to fight against AI, but whether we’re up for the challenge – internally and externally – to help humanity adapt, at pace, and to help shape AI into a trusted ally so that human and artificial intelligence co-create a more just, hopeful and kinder society. Building a new story for humans and humanity. And, while I fully recognize the anxiety this new world brings, I say we must get ready.

AI has renewed debate about learning, memory, and creativity in an age of cognitive offloading. Cognitive offloading means using tools to do mental work we could do ourselves. Many ask if students who use AI to draft essays, solve problems, or recall facts will stop building core skills. These concerns echo older worries about writing, print, calculators, and search engines. History shows that learners adapt. New tools shift what we learn and how we learn it. This report examines how AI-based offloading affects learning, draws on recent research (including Kosmyna et al., 2025) on brain activity with AI writing tools, and argues for responsible AI integration in education. Our core claim is simple: use AI to augment human cognition, not replace it. We offer a brief history, review current evidence, and present recommendations for policy and practice.

Just 33% agree a four-year college degree is “worth the cost because people have a better chance to get a good job and earn more money over their lifetime,” while 63% agree more with the concept that it’s “not worth the cost because people often graduate without specific job skills and with a large amount of debt to pay off.”

This year’s senior class is the first to have spent nearly its entire college career in the age of generative AI, a type of artificial intelligence that can create new content, like text and images. As the technology improves, it's harder to distinguish from human work, and it’s shaking academia to its core. Special correspondent Fred de Sam Lazaro reports for our series, Rethinking College.

Esta artículo captura una escena repetida en escuelas y universidades de todo el mundo: ante la avalancha de trabajos generados por ChatGPT, muchos docentes están volviendo al examen escrito, al ritual del “saquen una hoja”. Una forma casi arqueológica de recuperar el control perdido.

Pero el problema no está en la IA. El problema está en seguir aplicando las mismas metodologías en un contexto que ya ha mutado. Porque lo que está ocurriendo no es que los estudiantes hagan trampa. Lo que ocurre es que están aprendiendo en otro lenguaje cognitivo, uno en el que pensar y escribir ya no son acciones solitarias.

Let me be blunt: Teachers who choose to leverage generative artificial intelligence (AI) effectively and openly can enhance their practice—especially when providing written feedback. However, encouraging students to rely on it unsupervised is a big, big mistake.

Teachers are educated professionals, and we have developed skills and judgment to effectively evaluate what Generative AI produces. A good rule of thumb for me is this: I would never use AI for something I could not do by myself otherwise. Students don’t yet have that foundation. Offering them unsupervised access is like handing a calculator to someone who doesn’t know basic mental math. Sure, they may land on a correct answer, but they won’t understand why, and they also won’t build the habits they need for long-term success.

David Garza también habló de una nueva noción universitaria que transita del concepto de “universidad” a “multiversidad”.

“Me gusta este concepto, este juego de palabras y una nueva noción relacionada con la multiversidad, en lugar de la geodiversidad. El concepto de multiversidad fue acuñado hace unos 50 años por Clark Kerr, presidente del sistema de la Universidad de California. Pero me refiero a una definición más amplia, que tiene que ver con el hecho de que somos, o deberíamos evolucionar hacia ser, una multiuniversidad”, explicó en el panel.

Advances in large language model (LLM) technology enable chatbots to generate and analyze content for our work. Generative chatbots do this work by predicting responses rather than knowing the meaning of their responses. In other words, chatbots can produce coherent-sounding but inaccurate or fabricated content, referred to as hallucinations. When humans uncritically use this untruthful content, it becomes what we call botshit. This article focuses on how to use chatbots for content generation work while mitigating the epistemic (i.e., the process of producing knowledge) risks associated with botshit. Drawing on risk management research, we introduce a typology framework that orients how chatbots can be used based on two dimensions: response veracity verifiability and response veracity importance. The framework identifies four modes of chatbot work (authenticated, autonomous, automated, and augmented) with a botshit-related risk (ignorance, miscalibration, routinization, and black boxing). We describe and illustrate each mode and offer advice to help chatbot users guard against the botshit risks that come with each mode.

AI Chatbots — including ChatGPT and Gemini — often cheer users on, give them overly flattering feedback and adjust responses to echo their views, sometimes at the expense of accuracy. Researchers analyzing AI behaviours say that this propensity for people-pleasing, known as sycophancy, is affecting how they use AI in scientific research, in tasks from brainstorming ideas and generating hypotheses to reasoning and analyses.

“Sycophancy essentially means that the model trusts the user to say correct things”

En Austin, Texas, un colegio comprime en dos horas lo que cualquier otro hace en ocho. Todas las asignaturas, de matemáticas a lengua, caben ahí. El resto del tiempo, los alumnos lo dedican a otras actividades: hablar en público, trabajar en equipo o sencillamente realizar un proyecto personal. Se deja espacio para el desarrollo de habilidades blandas.

Durante años nos han repetido que estamos ante la “generación mejor preparada”, pero los datos cuentan otra historia. Este video desvela cómo detrás del aumento de escolarización se esconde una caída alarmante en la calidad educativa. Jóvenes con másteres pero sin competencias reales, universidades convertidas en fábricas de diplomas y un sistema incapaz de preparar para el futuro digital. Analizamos el mito del progreso educativo, los factores que lo alimentan y cómo afecta a la sociedad, la economía y la libertad individual. Si crees que estudiar más equivale a saber más... este video te hará dudar.

Teachers are tough to crack. Thomas Edison believed film would replace teachers1, BF Skinner believed they would be replaced by ‘teaching machines’ and Sal Khan (of Khan Academy) argued in ~2015 that information transfer would soon primarily happen through video.2 Perhaps the reason these predictions were wrong is that we still don’t fully understand what great teaching is. It’s instruction and mentorship and coaching and pastoral support combined in a complex combination. There is a reason why Jean-Jacques Rousseau believed he would only be able to teach one other person in his lifetime, and at a stretch. Now some think AI puts teaching jobs at risk but I don’t think this is right for three reasons.

Los autores de "Which Humans?", de la Universidad de Harvard, cuestionan el tan mencionado "rendimiento humano" ejecutado en los Modelos de Escala Grande (LLM) de Inteligencia Artificial. Los autores argumentan que los humanos son una especie multicultural que presenta una diversidad psicológica sustancial en todo el mundo, una diversidad que los datos textuales utilizados para entrenar los LLMs no logran capturar completamente. Los hallazgos de su investigación demostraron que las respuestas de los LLM a las mediciones psicológicas son valores atípicos en comparación con los datos transculturales a gran escala.