BUAP trains its professors as 'Docentes GuIA': AI forces a redesign of assessment, not a ban

By Zendoric · September 11, 2026.

The Benemérita Universidad Autónoma de Puebla (BUAP) has completed its Diploma in Artificial Intelligence and Teaching Transformation, a 120-hour programme aimed at its own faculty. The stated goal is ambitious: to train a thousand 'Docentes GuIA' —a new teaching profile designed for the era of generative AI— within one year, according to the university's own announcement.

The diploma, organised by BUAP's Agency for Artificial Intelligence and Algorithmic Ethics and endorsed by its School of Teacher Training and Academic Development, was structured around four micro-credentials —training blocks that can be certified separately—: the evolution of AI, 'GuIA Docente', the transformation of assessment, and wellbeing and development. The underlying idea is not to teach lecturers how to operate a tool, but to rethink the design of their classes and, above all, how to assess a student who now has permanent access to a generative assistant.

By BUAP's own definition, the Docente GuIA combines technical understanding of AI, pedagogical transformation, new assessment criteria, critical thinking, ethical responsibility and human development. Their added role is to teach students to distinguish when it makes sense to use AI, when to question it and when it is better to do without it altogether.

The first results, measured by the university itself at the close of the course, show notable improvements in teachers' responses: those who say they design activities incorporating AI rose from 50 to 91.7%; those who say they have defined criteria for assessing AI-assisted work, from 33.4 to 83.4%; and those who guide their students to critically analyse AI-generated content, from 33.3 to 100%. A further 83.3% say their students already verify data and cross-check sources. These figures should be taken for what they are: a self-assessment by the participants themselves immediately upon finishing the training, with BUAP specifying neither the sample size nor whether the effect holds months later, in a real classroom and without the momentum of the course itself.

As sector context, BUAP is merely formalising a problem that has run through higher education worldwide since generative AI became widespread: homework and the traditional essay have ceased to be a reliable test of what a student knows, because it is almost impossible to tell how much the machine wrote. The most common response so far has been reactive —unreliable AI detectors, bans that are hard to police—; what this programme proposes is the opposite route: equipping teachers to redesign assessment from the design of the activity itself, rather than chasing cheating after the fact.

Our reading is that this diploma is a miniature portrait of the thesis we have been advancing about AI's impact on education: the teacher who comes out ahead is not the one competing with the machine on speed or memory, but the one who learns to orchestrate it —designing with it, assessing through it, teaching students to question it—, while the profile limited to transmitting content is precisely the one these 120 hours of training aim to make disappear. It is a transition with a real cost: it demands time, budget and an institutional structure capable of sustaining an agency dedicated to algorithmic ethics, something not every university —starting with those competing for resources with BUAP within Mexico's own public system— can afford as quickly. That gap in capacity between institutions is one of the least discussed short-term risks: not only which students learn to use AI with judgement, but which universities have the means to train the teachers who will show them how.

If the effect holds outside the controlled environment of the diploma itself, the underlying direction points to something we share: an education system that stops rewarding memorisation and starts assessing judgement, verification and critical thinking in the face of a machine that already handles more data than any teacher. It is the kind of reallocation of tasks —from the routine to the specifically human— that underpins our fundamental bet on an AI that, once the friction of adaptation is overcome, ends up freeing teachers' time for what truly makes the difference in a classroom.