Sesgos discursivos e ideológicos en las respuestas producidas por tres modelos de inteligencia artificial (ChatGPT, Gemini y Copilot)
Un análisis desde la lingüística de corpus y el análisis crítico del discurso
Keywords:
Artificial Intelligence, AI, Spanish, algorithms, corpus linguistics, discourse analysisAbstract
This article analyzes repetitive patterns in responses generated by three artificial intelligence models. It begins with a selection of examples obtained through open ended instructions and prompts to identify lexical and syntactic structures that validate corpus linguistic approaches (Sinclair). These approaches are based on the two principles that govern this type of discourse production. The study also applies critical analysis (Fairclough) to determine if the responses obtained exhibit discursive biases that reinforce dominant ideological representations.
References
Barrios, Hernando, Vianney Pérez y Yolanda Guerra. «Subjetividades e inteligencia artificial: desafíos para ‘lo humano’». Veritas, 47, 2020, pp. 81-107.
Bucher, Tania. If... Then: Algorithmic Power and Politics. Oxford UP, 2018.
Fairclough, Norman. Analysing Discourse: Textual Analysis for Social Research. Routledge, 2003.
Fairclough, Norman. Critical Discourse Analysis: The Critical Study of Language. Longman, 1995.
Fairclough, Norman. Discourse and Social Change. Polity P, 1992.
Ferrante, Enzo. «Inteligencia artificial y sesgos algorítmicos: ¿por qué deberían importarnos?». Nueva Sociedad, 294, 2021, pp. 27-37.
Sinclair, John McHardy. Corpus, Concordance, Collocation. Oxford UP, 1991.
Sinclair, John McHardy. Trust the Text: Language, Corpus and Discourse. Routledge, 2004.
van Dijk, Teun A. Discourse and Power. Palgrave MacMillan, 2008.
