Based on original videos and texts related to science dissemination, educational materials for STEM subjects were created using a communicative approach, originally developed to teach languages. The methodology employed utilized Artificial Intelligence (AI) tools, specifically Natural Language Processing (NLP) algorithms, trained with the curricula of five subjects (Computer Networks, Technological Processes of Fruits and Vegetables, Computer Science, Teaching Practice 1, and Didacts of Mathematics) to help identify connections between the topics of these subjects and the videos from the "Ciencia, Conciencia y Café” series (https://ciencia-conciencia-cafe.tic.unam.mx ) and the articles from the magazine “PaCiencia Pa’Todos” (https://publicaciones.aragon.unam.mx/ojs/index.php/paciencia/index) . The original videos, which were one hour long, were edited to produce reels lasting between two and three minutes, focusing on the topics identified by the AI. Similarly, materials were extracted from the magazine texts, consisting of a small number of paragraphs for each topic selected, to be used in the classroom. In both cases—the videos and the magazine—the only editing done to the original materials was in their length, without altering their content. The materials were cataloged using the Dublin Core standard, including metadata such as "description," "format," and "educational level" to facilitate their search on the web platform created for its storage Additionally, teaching guides for teachers were created for classroom use. These products were supported through project PE104025, which is funded for two years by the Universidad Nacional de México (UNAM) through the Program for Support Projects on Education Innovation and Improvement (PAPIME). In the second year of the project, the effectiveness of the materials used in will be evaluated.

Keywords: Communicative Approach, Natural Language Processing, Computer Science, Dublin Core Standard, UNAM.