The game of artificial listening: when chatbots take the place of counselors and therapists

Abstract

This study problematizes the growing use of artificial intelligence (AI), more specifically Large Language Models (LLMs), in functions associated with the work of therapists and counselors. To critically examine this issue, the classic question formulated by Turing is revisited: “Can machines think?”, updating it within the contemporary paradigm, in which the discussion moves beyond a merely methodological dimension and enters the ontological realm. In this context, the study analyzes the ethical, conceptual, and political limits involved in attributing human subjectivity to machines within the context of psychotherapeutic and emotional support practices. This is a qualitative study with a theoretical and interdisciplinary approach. As its methodological procedure, a critical narrative review of Brazilian scientific literature published over the past five years on the use of LLMs in Psychology was conducted. The conceptual framework articulates a critique of mechanistic reductionism, computational imperialism, and technological determinism, understanding technology as a political artifact. Cultural-Historical Psychology is adopted as the epistemological foundation. The findings indicate that the functional emulation of language performed by LLMs does not constitute evidence of thought or consciousness. In contrast, human thought is understood as a relational, embodied, affective, and historically mediated activity, inseparable from lived experience and social relations. The growing tendency to anthropomorphize chatbots is understood as an expression of an instrumental rationality that reduces the psyche to measurable data. Finally, the study draws attention to the ethical risks involved in the use of these technologies and reaffirms the irreplaceability of clinical judgment and human praxis in psychological care.