AUTOMATED PRODUCTION OF ACADEMIC ABSTRACTS: PROMPT ENGINEERING FROM THE DIALOGIC PERSPECTIVE OF SPEECH GENRES

Abstract

This article investigates the automated production of academic abstracts using generative artificial intelligence (GenAI), exploring the tension between prompt engineering techniques and the discourse genre theory of Mikhail Bakhtin and the Circle. The main objective is to analyze how the formulation of dialogical commands influences the quality, epistemic density, and compositional stability of the abstract genre. The methodology, qualitative and exploratory in nature, is structured as an empirical test of textual confrontation, comparing the original abstract of a humanities article with a version generated by GenAI through the application of a parameterized prompt. The results show that the machine, when guided by a command that acts as a mediating utterance, not only satisfactorily emulates the normative structure and rhetorical moves of the genre, but also manages to mimic the textual marks of the ideological sign, retaining and deepening the specific theoretical jargon of the area. It is concluded that automation does not decree the end of academic authorship, but demands its displacement; the success of algorithmic generation lies in the researcher's competence to act as a "prompt architect", responsible for parameterizing GenAI to meet the discursive, ethical, and scientific constraints of their field of study.