Les limites épistémologiques des techniques numériques actuelles de l’intelligence artificielle en traduction

Abstract The aim of this article is to analyze in detail the concepts underlying the digital techniques of artificial intelligence currently used in various fields, including translation. Based on this analysis, the epistemological limits of these techniques are examined, especially those resulting from the difficulty of modeling mechanisms of human rationality such as abduction or selection.In addition, the article explores the risks induced by the mass use of these techniques in translation, including their nature, extent and effects, and illustrates them with practical examples.