The expansion of artificial intelligence in the classroom has created a fairly specific paradox: we expect teachers to learn how to use it pedagogically, critically, and responsibly, yet we still have few specific instruments for determining what they actually know and where they need further training.
The work of Andresa Sartor-Harada and Juliana Azevedo-Gomes addresses precisely this problem through the development of AI-ED-SAT, a self-assessment questionnaire that shifts the focus from generic digital competence to four dimensions specific to educational AI: conceptual understanding, pedagogical use, ethical reflection, and curricular integration. The distinction matters because being proficient with an AI tool does not necessarily mean knowing when to use it, how to incorporate it into the curriculum, or what the implications are of delegating certain decisions to an algorithmic system.
The instrument was developed through a fairly demanding process: a literature review, two Delphi rounds involving 12 experts, and a pilot study with 128 teachers from seven Spanish-speaking countries. Of the initial 40 items, 38 were retained, and the psychometric analyses yielded solid results: an overall internal consistency of 0.93, four factors explaining 67.8% of the variance, and good fit indices in the confirmatory analysis.
Perhaps the most suggestive finding, however, is not statistical but lies in the profile that emerges from the responses: teachers perceive themselves as more competent in the pedagogical use of AI than in its ethical dimension or curricular integration. This is a significant difference because it suggests that the practical adoption of these technologies may be progressing faster than reflection on how they should fit into broader educational decisions.
In any case, AI-ED-SAT should be interpreted for what it is: a self-assessment tool, not a direct measure of teaching performance. The sample was also selected by convenience, validation has so far been limited to Spanish-speaking contexts, and the authors themselves regard these findings as an initial validation that should be replicated with larger and more diverse samples.
That may be precisely where one of its most interesting uses lies. Rather than turning “AI competence” into a broad label, the questionnaire makes it possible to break the concept down and identify specific needs. Used in this way, it may be less useful for classifying teachers than for posing a far more productive question: what does each teacher need to learn in order to move from using artificial intelligence to making sound educational decisions about it?
---
How to Cite: Sartor-Harada, A., & Azevedo-Gomes, J. (2026). AI-ED-SAT: design and validation of a questionnaire for self-assessment of teaching skills in educational AI. RIED-Revista Iberoamericana de Educación a Distancia, 29(1), 79–110. https://doi.org/10.5944/ried.45413
