Bakhtin and generative AI: a narrative review of Brazilian studies on dialogism, authorship and responsiveness

Abstract

Generative Artificial Intelligence (AI) has reshaped how discourses are produced, circulated, and interpreted, reopening questions about authorship, interaction, and meaning-making. This article offers a critical narrative review of three Brazilian studies that approach generative AI through the Dialogical Theory of Discourse of Bakhtin and the Circle: Gonçalves and Coitinho (2023), Conte, Devechi and Maciel (2025), and Duque-Pereira and Moura (2025). Adopting a qualitative, interpretive approach, the corpus, selected through searches on Consensus and SciSpace, was analyzed along four thematic axes: the currency of Bakhtinian concepts in the face of AI; dialogism, responsiveness, and the limits of human–AI interaction; authorship and alterity; and the studies' convergences, divergences, and contributions. The findings indicate that all three studies affirm the relevance of Bakhtinian categories for AI-mediated discourse, but through different strategies: applying established categories, rethinking authorship, or proposing new concepts such as "simulated dialogism" and "algorithmic monologism." The review argues that AI does not eliminate authorship or the dialogical, responsive nature of language, but reconfigures the conditions under which utterances are produced. It also identifies two gaps in the field: limited attention to the explanatory limits of the theory and a scarcity of empirical research on concrete uses of AI.