Self-Referential Discourses of LLMs on Artificial Intelligence: An Analysis in Light of Social Representations Theory and Bakhtinian Dialogism
Bakhtiniana: Revista de Estudos do Discurso · 2026 · 21 (4)
DOI: 10.1590/2176-4573e76414 ↗Kurz gefasst
Fachartikel in der Zeitschrift Bakhtiniana: Revista de Estudos do Discurso, 2026. Verfasst von Davi Gomes Sousa und 1 weiteren Autor:innen. Die Arbeit nennt 22 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
This article investigates, in light of Social Representations Theory and Bakhtinian dialogism, the self-referential discourses produced by Large Language Models (LLMs) about artificial intelligence. It assumes that these systems do not produce self-perceptions in the subjective sense but rather reproduce and reconfigure human social representations present in their training data. In a qualitative, descriptive and exploratory study, evocations were collected from 21 freely accessible LLMs through a prompt inspired by the Free Word Association Technique, adapted to the computational context. The corpus was analyzed with the IRAMUTEQ and Tri-Deux-Mots software. The models shared a technical-functional, positively toned repertoire of evocations, with recurrent use of automation, innovation and learning, and an absence of ethical or critical dimensions. From a Bakhtinian perspective, these utterances respond to dominant social voices about AI and produce a monologic effect by erasing dissonant voices. The study concludes that LLMs take part in the discursive reconfiguration of the social representations of AI, reinforcing its techno-scientific and progressive image.
Zitieren
APA
Sousa, D. G., & Moreira, L. S. (2026). Self-Referential Discourses of LLMs on Artificial Intelligence: An Analysis in Light of Social Representations Theory and Bakhtinian Dialogism. Bakhtiniana: Revista de Estudos do Discurso, 21(4). https://doi.org/10.1590/2176-4573e76414
Vancouver
Sousa DG, Moreira LS. Self-Referential Discourses of LLMs on Artificial Intelligence: An Analysis in Light of Social Representations Theory and Bakhtinian Dialogism. Bakhtiniana: Revista de Estudos do Discurso. 2026;21(4). doi:10.1590/2176-4573e76414
BibTeX
@article{Sousa2026,
author = {Sousa, Davi Gomes and Moreira, Lucas Sales},
title = {Self-Referential Discourses of LLMs on Artificial Intelligence: An Analysis in Light of Social Representations Theory and Bakhtinian Dialogism},
journal = {Bakhtiniana: Revista de Estudos do Discurso},
year = {2026},
volume = {21},
number = {4},
doi = {10.1590/2176-4573e76414}
}
EUVIMED