WHO THINKS WHEN AI THINKS? THE EROSION OF REAL KNOWLEDGE AMONG UNIVERSITY STUDENTS IN THE USE OF GENERATIVE ARTIFICIAL INTELLIGENCE

Abstract

This article aimed to investigate how the intensive use of generative artificial intelligence by university students relates to the erosion of real knowledge, that is, the progressive emptying of the cognitive competencies that higher education aims to develop. The methodology consisted of a qualitative and exploratory bibliographic review, surveying empirical studies published between 2021 and 2026 in scientific databases, prioritizing research with neural, behavioral, and academic performance measurements. The results converge toward a consistent pattern: artificial intelligence adoption among university students exceeds 90% in some countries and 70% in Brazil, while electroencephalography studies, large-scale surveys, and performance analyses associate dependent use of the tool with reduced neural connectivity, cognitive offloading, and declines in critical thinking and achievement indicators. It is concluded that the problem lies not in the tool, but in the mode of use: artificial intelligence employed as a substitute for cognitive effort, rather than as support, becomes a vector of epistemic erosion, which requires assessment redesign and training in critical literacy.