San Lorenzo teaches its students to distrust AI: the plan came from a proposal by the students themselves

By IRE Somos Región · September 26, 2026.

At Secondary School No. 549 "María Catalina Echevarría de Vidal", in San Lorenzo, fourth- and fifth-year students held the first session of a new training program: learning to use artificial intelligence (AI) critically, not just to operate it. The activity is led by Jaquelina Serrano, a graduate in Education and techno-pedagogical adviser, as reported by IRE Somos Región, and will be extended to other secondary schools in the city.

The program is not limited to teaching how to write better prompts (the text instructions used to converse with a chatbot) in order to get more out of these tools. It also devotes space to the risks: fact-checking, fake news, manipulated images, deepfakes (synthetic audiovisual content that impersonates real people), privacy, digital footprint and authorship of what is produced with these tools. As part of the activities, students take on a challenge that combines creating, verifying and reflecting on what they have learned.

The most interesting part of the case is not the syllabus -fairly standard within what is understood today as digital literacy- but its origin. The program grew out of a project submitted by students from Colegio Misericordia to the Presupuesto Participativo Joven, a municipal mechanism in which young people themselves decide how to spend part of the public budget. It was they, not an education ministry or a consultancy, who identified the need and turned it into public policy at city scale.

That genealogy matters. Broadly speaking, the arrival of critical AI literacy in classrooms is advancing unevenly and depends more on local initiatives -municipalities, individual schools, teachers acting on their own- than on uniform national mandates. That it is teenage demand pushing the curriculum, and not the other way around, is a healthy sign: it indicates that part of that generation does not use these tools uncritically, but rather perceives its own blind spots and asks for help covering them.

It is also the right answer to the wrong problem usually posed in public debate. The discussion about AI and education tends to boil down to whether or not students should use a chatbot to do their homework. The relevant question is a different one: who teaches them not to believe everything a model generates, to tell a real image from a fabricated one and to protect their digital footprint in an environment where producing fake content is ever cheaper. A program of this kind does not solve the underlying problem -the saturation of synthetic information on social networks and search engines- but it builds the only kind of defense that scales: people capable of doubting with judgment.

The scope of the news should not be overstated: it is a program that started at a single school and is beginning to be replicated, with one teacher in charge and no framework described beyond the municipal push. The real challenge is for it to survive the novelty and become a stable part of the curriculum, and for other teachers -not just the one running the program- to incorporate these criteria into their own practice. Playing out there, in miniature, is the same tension running through the teaching profession in the age of AI: those who use it to orchestrate better questions in the classroom win; those who limit themselves to transmitting content a model already generates as well or better lose. San Lorenzo, with a program born from its own students, is betting on the former.