Abstract

This article examines ChatGPT’s discursive production from a linguistic-philosophical perspective based on the theory of Eugenio Coseriu (1981, 1992), who distinguishes three levels of language: the universal level (speaking), the idiomatic level (the norm), and the individual level (discourse). The research focuses on comparing the argumentation of ChatGPT 3.5 with that of human speakers when faced with a moral dilemma, using the D.I.T. Questionnaire (Rest, 2013) as a methodological tool. The findings show that, while 100% of humans express a point of view (either for or against), ChatGPT does so only in 23.4% of cases. It is therefore proposed that this avoidance of taking a position reflects a lack of discursive freedom due to a lack of responsibility and self-involvement, which places artificial intelligence outside of full linguistic activity, in accordance with the Coseriu’ framework. This paper discusses the relationship between norm, freedom, and responsibility in the production of meaning in large-scale language models, highlighting the structural, pragmatic, and moral differences in language use between humans and machines. © Los autores. Este artículo es publicado por la revista Lengua y Sociedad de la Facultad de Letras y Ciencias Humanas, Universidad Nacional Mayor de San Marcos.