Evaluación por Competencias en la Era de la Inteligencia Artificial: De la Prohibición Normativa a la Adaptación Pedagógica

Abstract

This action-research study addresses the knowledge validation crisis within the educational system caused by the widespread adoption of Generative Artificial Intelligence (GenAI). Given the obsolescence of traditional assessment instruments—which have become technologically "solvable" by Large Language Models (LLMs)—a teacher training mechanism was implemented in the format of a Didactic Athenaeum under the framework of Ministerial Resolution No. 70/2020 of the Province of Salta.

Using a sample of 277 participants from Higher, Technical, and Vocational Education, an experimental methodology was deployed that included phases of neurocognitive sensitization, diagnosis of algorithmic hallucinations, and curricular redesign. The study introduces the theoretical concept of Bio-Situated Anchoring Assessment (EABS, by its Spanish acronym) as a pedagogical resistance strategy, proposing assignment prompts anchored in physical experience and local context to guarantee algorithmic insolubility.

The results obtained demonstrate that, following the intervention, 67% of the teachers' productions successfully incorporated effective cognitive identity validation mechanisms. This shifted the focus from content retrieval (where AI demonstrates 72.7% effectiveness) to evaluative judgment and situated competence (where AI effectiveness drops to 4.5%). Furthermore, the study discusses the tension between jurisdictional hardware restriction regulations (device bans) and the technological inevitability projected toward the 6G horizon, concluding that the only sustainable defense against automated fraud is not physical surveillance, but the didactic immunization of the assignment.