Étude de la compétence lexicographique de ChatGPT dans le domaine de l'architecture contemporaine française

Abstract

A Study of ChatGPT's Lexicographic Competence in the Field of Contemporary French Architecture. This article analyses the capabilities of ChatGPT as an assistive tool for the production of specialised lexicographic definitions in the field of contemporary French architecture. The study focuses on a set of architectural terms displaying different degrees of specialisation and complexity, submitted to ChatGPT in order to assess its ability to generate lexicographic definitions and organise the associated semantic information. The methodology adopted is based on a comparative analysis of the definitions generated by ChatGPT and those provided by reference general and specialised dictionaries. The analysis considers, in particular, the precision of the definitions, the selection and organisation of semantic information, and their adequacy to the specialised domain. This approach makes it possible to examine the role of generative artificial intelligence models in specific lexicographic compilation practices, including the production of draft definitions, the reformulation of information, and the initial organisation of lexical data. The aim is therefore to assess the extent to which ChatGPT can be used as an assistive tool in the design and development of specialised lexicographic resources, while also examining the methodological conditions required for the use of its outputs in this context. More broadly, the study contributes to the ongoing discussion on the integration of generative artificial intelligence technologies into contemporary digital lexicographic practices. Keywords: contemporary architecture, ChatGPT, specialised lexicography, artificial intelligence, lexicographic definition, terminology