What is generative in generative artificial intelligence? A design-based perspective - Archive ouverte HAL
Article Dans Une Revue Research in Engineering Design Année : 2024

What is generative in generative artificial intelligence? A design-based perspective

Antoine Bordas
Pascal Le Masson
Benoit Weil

Résumé

Abstract Generative artificial intelligence (GenAI) models have attracted tremendous interest since the advent of ChatGPT, raising numerous opportunities and challenges. However, their generative power has not yet been studied, leaving open the question of what is truly generated by these tools. This paper addresses this question and precisely characterizes the generativity behind GenAI models. Owing to the latest advancements in engineering design, we first propose a framework for uncovering the various types of generativity. Then, we consider the main families of GenAI models and systematically analyze them to characterize their generativity within this framework. By doing so, we highlight the existence of two distinct generative levels in GenAI: one leading to the generation of new artifacts and the other leading to the generation of GenAI models themselves. We are also able to characterize the generativity of both of these levels, thus specifically confirming the generative power of GenAI and opening research avenues toward human-GenAI collaboration.
Fichier non déposé

Dates et versions

hal-04718137 , version 1 (02-10-2024)

Identifiants

Citer

Antoine Bordas, Pascal Le Masson, Maxime Thomas, Benoit Weil. What is generative in generative artificial intelligence? A design-based perspective. Research in Engineering Design, 2024, ⟨10.1007/s00163-024-00441-x⟩. ⟨hal-04718137⟩
0 Consultations
0 Téléchargements

Altmetric

Partager

More