Cognitive offloading and critical thinking

how LLM affect foundational competencies in EPCT

Abstract

Given the increasing ubiquity of artificial intelligence (AI), especially the large language models (LLM) such as ChatGPT, Gemini or Claude, a pedagogical dilemma emerges whose central issue lies in the tension between the proficiency in AI technologies for the world of work and the impact of their unmediated use on propaedeutic learning. This article aims to examine the implications of LLM use on the development of critical thinking during the formation of foundational competencies within the scope of Professional, Scientific and Technological Education (EPCT). The research is characterized as an exploratory and descriptive study, conducted through an integrative literature review, with a corpus comprising scientific articles indexed in the Scopus and Web of Science databases, selected through inclusion and exclusion criteria organized into five thematic axes. The results evidence that the absence of consolidated conceptual foundations leads students to passively accept AI outputs, compromising deep learning. Studies demonstrate that LLM suppress the desirable difficulties that constitute the engine of cognitive development, inhibiting the processes of assimilation, accommodation, and metacognition. It is further evidenced that, without intentional pedagogical mediation, AI ceases to operate as a tool and begins to function as a structuring technology, altering the very ends of the learning process. It is concluded that the decisive variable is not the presence or absence of technology in the educational pathway, but rather the timing of its integration and the quality of pedagogical mediation, such that AI incorporation should occur after the consolidation of the foundational competencies that underpin the intellectual autonomy of EPCT graduates.