In EDS ansehen

Of Humans and AI: A Critical Discourse Analysis of Their Encounters and Interactions

Gespeichert in:
Bibliographische Detailangaben
Titel: Of Humans and AI: A Critical Discourse Analysis of Their Encounters and Interactions
Autoren: Angela Zottola, Michelangelo Conoscenti
Verlagsinformationen: Palgrave Macmillan.
Schlagwörter: AI in academic discourse, Discursive practices, ChatGPT, Gemini, HuggingChat, Reverse Language Engineering, NLP
Beschreibung: This chapter explores the interplay between AI’s representation in academic literature and its discursive construction by three AI platforms, namely ChatGPT, Gemini, and HuggingChat. These representations influence public perceptions, expectations, and the development and regulation of AI. Employing the methodological frameworks of Critical Discourse Analysis and Corpus Linguistics, we analyse two datasets: a corpus of 25 academic publications and conference proceedings that explore the conceptualization of AI in research and education, and 16 dialogic exchanges (each consisting of a question and its corresponding answer) conducted with the AI platforms previously mentioned. These questions were specifically crafted to elicit the platforms’ self-perceptions and discursive patterns. The academic corpus reveals a shift from framing AI as a hazard to exploring its potential for independent thought and reasoning. Conversely, the AI-generated corpus demonstrates improvements in logical-abstract reasoning and increasingly natural interactions, with platforms exhibiting critical, balanced self-descriptions influenced by human interlocutors’ tone and phrasing. In conclusion, the discourse generated by academics has recently adopted a more technical and cautious tone and AI is discursively framed as a moral and social agent requiring governance. These findings underscore the need for developers and institutions to control and regulate the empowerment of AI systems before their generative capacities outperform those of their creators.
Publikationsart: Part of book or chapter of book
Dateibeschreibung: application/pdf
Sprache: English
Zugangs-URL: https://hdl.handle.net/2318/2092360
Dokumentencode: edsair.od.......970..c704d7a7f0ae2f8f4d0b09a5fff2ed15
Datenbank: OpenAIRE
Beschreibung
Abstract:This chapter explores the interplay between AI’s representation in academic literature and its discursive construction by three AI platforms, namely ChatGPT, Gemini, and HuggingChat. These representations influence public perceptions, expectations, and the development and regulation of AI. Employing the methodological frameworks of Critical Discourse Analysis and Corpus Linguistics, we analyse two datasets: a corpus of 25 academic publications and conference proceedings that explore the conceptualization of AI in research and education, and 16 dialogic exchanges (each consisting of a question and its corresponding answer) conducted with the AI platforms previously mentioned. These questions were specifically crafted to elicit the platforms’ self-perceptions and discursive patterns. The academic corpus reveals a shift from framing AI as a hazard to exploring its potential for independent thought and reasoning. Conversely, the AI-generated corpus demonstrates improvements in logical-abstract reasoning and increasingly natural interactions, with platforms exhibiting critical, balanced self-descriptions influenced by human interlocutors’ tone and phrasing. In conclusion, the discourse generated by academics has recently adopted a more technical and cautious tone and AI is discursively framed as a moral and social agent requiring governance. These findings underscore the need for developers and institutions to control and regulate the empowerment of AI systems before their generative capacities outperform those of their creators.