For a Critical Review of Artificial Intelligence in Education

Abstract

This article reflects critically upon the implications of introducing Large Language Models, popularly called Artificial Intelligence, in Brazilian Education. This essay begins with a discussion of how these models are created. It starts with a definition of data and their role in such models, pointing to the myth of its objectivity and how its selection and organisation might lead to ideological and subjective assumptions. The need for human intervention in the training process is also discussed to demythologise the idea that these models are superior and autonomous intelligences, as they are in fact based on statistical training models. The role of the different Artificial Intelligence models in our society is then discussed. The discussion relies on the concepts of technological invisibility and technological singularity, which imply technological assimilation in everyday life over momentary wonder without the necessary criticality. Finally, the article discusses LLMs as part of our techno-social system, emphasising their ideological significance and practical impacts. The final section discusses some paths to be considered if we think about Large Language Models presence in education situations.