Imagine a journalism professor. For years she’s taught her students how to write reports: how to find sources, how to structure the story, how to find the right tone. Now, her faculty is recommending that she incorporate chatbots into her classes. They offer her a two-hour workshop on how to build effective prompts . She leaves the workshop with a list of instructions. Has this helped her become a better teacher? Or have they simply taught her how to use a tool without asking her if that tool makes sense in her subject?
This scene, which is repeated with variations in universities around the world, summarizes the problem that most concerns those of us who investigate university teaching in the age of generative artificial intelligence: the temptation to reduce a pedagogical issue to a technical one.
A technology qualitatively different from previous ones
For decades, educational technology arrived in classrooms as inert resources: a projector, a computer, a content management platform. Teachers either used them or not. They adapted them to their style. They maintained control over what happened in the classroom.
AI is different. A generative model doesn’t passively wait for instructions: it identifies patterns, develops arguments, synthesizes information, evaluates responses, and simulates conversations. They are not passive intermediaries between teachers and students, but can play an active role in shaping educational processes . In a sense, they act as interlocutors within the classroom, which radically changes the nature of their integration into teaching.
The AI agent intervenes in processes that until now belonged exclusively to the teacher’s judgment: designing activities, interpreting student progress, generating explanations, and evaluating work. This means it can either expand the teacher’s capabilities or erode their role.
The teaching agency: something that is achieved
In educational research, teacher agency is the ability to act reflectively and intentionally within the conditions offered by the environment. It is not a skill that can be acquired in a course, but rather one that develops under specific conditions .
These conditions relate to three aspects:
- The baggage each teacher carries: their educational background, their experiences as a student and as a teacher, their beliefs about what it means to learn. Our journalism professor, for example, how did she learn to write a report? Does she consider writing a thought process or a technique to be executed? These answers will profoundly influence how she interprets the arrival of a chatbot in her class.
- The context and the decisions it allows you to make. Do you have access to the full versions of the tools or only to the free versions with their limitations? Do you have real time to critically review what artificial intelligence generates, or does your teaching load make that review impossible? Do your university’s regulations allow you room to experiment with assessment systems?
- The teaching objective: if your goal is for students to develop the ability to write a report—with all that entails: researching, structuring, empathizing, revising—you might see the chatbot as a dangerous shortcut. If, on the other hand, you interpret the tool as a first reader that helps students identify weaknesses in a draft before the teacher’s review, it can be a resource that enriches the process without replacing it.
Four conditions that make the difference
Accumulated experience is a determining factor in achieving teacher agency. Having established pedagogical references facilitates a more thoughtful integration of technologies.
Furthermore, critical literacy in artificial intelligence is essential . Understanding how these systems work—how they generate responses, what biases they may introduce, and what their limitations are—allows for informed pedagogical decisions regarding their use. A journalism teacher who knows that the chatbot doesn’t “understand” what she writes, but rather predicts likely word combinations, has a very different perspective on what she can and cannot ask of that tool.
Nor should we forget institutional support. When institutions offer clear guidance , time for pedagogical reflection, and training focused on didactic questions—and not just functionalities—teachers have more opportunities to experiment with and critically integrate technology.
Finally, communities of practice are fundamental when teachers face pedagogical innovation. These shared learning spaces allow for the exchange of experiences, the analysis of difficulties, and the construction of common interpretations of educational changes. Departments function as professional communities where teaching methods are defined—often implicitly—and where innovation in teaching occupies a place . When this culture penalizes experimentation, the space for exercising agency is considerably reduced.
Taking the spotlight away from AI
The debate surrounding artificial intelligence in universities often oscillates between those who see it as a transformative opportunity and those who see it as a threat to academic integrity. But both positions share a blind spot: they assume that technology is the protagonist of the story. Technodeterminism , in its various forms, has been operating with this logic for decades.
The ecological approach proposes another protagonist: the teaching staff. Not as a passive victim of an inevitable technological disruption, nor as a solitary hero who resists or embraces the new, but as a professional situated in a specific context, with a history, with real institutional conditions, and with educational objectives that give meaning to their work.
Therefore, the greatest threat is the uncritical delegation to these systems of processes that are at the core of a teacher’s role: interpreting the classroom context, making informed pedagogical decisions, and guiding learning with sound judgment. Systems based on this technology should not be seen as automatic solutions to educational problems ; they always require the pedagogical judgment of the teaching staff, ethical reflection, and human oversight.
Author Bios: Mari Mar Boillos Pereira is a Contracted PhD Professor at the Faculty of Education of Bilbao at the University of the Basque Country / Euskal Herriko Unibertsitatea
