Abstract:

Artificial intelligence (AI) applied to linguistic generation establishes a reference context for interpreting human processing and its interaction with the physical and digital world. Natural Language Processing (NLP) faces the challenge of transcending linear modeling to achieve different forms in which metaknowledge, guided by algorithms that enable attentional supervision, self-regulation, and real-time updating of linguistic-textual and verbal representations, becomes a reference to clearly understand higher cognitive processes in humans.
This research aims to bring understanding to metalinguistic processing in humans facing a task of generating and interpreting narrative texts about value construction, based on patterns generated by AI, so that we can identify what elements are substantially significant in relation to the new learning framework that is presented today. The objective is to create a semantic-conceptual map of indicators of humanization of the narrative, through the extraction of entities of subjective relevance by human activity, which allows us to understand what meta-procedures must be taken into account in education to promote a construction of thought in accordance with the current needs of learning and development.
The study involved 253 students from university degrees in Early Childhood Education and Primary Education, corresponding to the total number of students enrolled in the second year. The established sequential design considers four phases: pre-training and generation of AI patterns, human creation of narratives, metalinguistic extraction, and final organization and categorization.
The obtained results allowed the discovery of the semantic and conceptual network underlying the initial interpretation of the narrative (first-level comprehension), defining in particular the nodes of meaning in the network and their impact on humans (metalinguistic level).
The implications of the study are several. On the one hand, they shed light on the organic nodes of a semantic-conceptual network and their potential applications in the development of AI-based networks. On the other hand, in their application to education, conceptual and meaning extraction enables real adjustments for pedagogical intervention in human value construction. The results in this sense support the need to aim in the teaching and learning process towards those elements of thought that transcend the linear reality of knowledge, drawing a school in which metacognitive construction and reflection on the management and manipulation of information must be prioritized over content-based learnings. Finally, the current educational framework must reconsider the pedagogical and methodological organization model, transferring the pedagogical discourse from the contents themselves to learning contexts, where the learners act on their contexts based on human qualities intrinsic to their biological and psychological properties. The determination of those indicators (metacognitive patterns of interpretation) that bring into play in the construction of the story is an extraordinary starting point for the creation of work proposals with the students, favoring the development of multiple processes of reasoning and understanding.

Keywords:

IA, artificial intelligence, innovative education, metacognition, learning, NLP.