Welcome to the IA-Emotion project website!

Can generative artificial intelligences (AIs) such as ChatGPT make their users feel emotions during conversational exchanges? How do these emotions transpire in conversations? Can AIs automatically detect emotions such as joy, fear, disgust and others? The IA-EMOTION project focuses on conversational agents, to understand how human-machine dialogue can constitute a new space for the fabrication of emotions, i.e. become a space conducive to the creation and sharing of emotions. The project will combine information and communication sciences with computer science to analyze and understand how emotions are expressed in conversational exchanges. It aims to produce an AI model for automatic emotion detection based on a better understanding of human language practices with a machine.

The IA-EMOTION project has two scientific ambitions:

The first is a citizen-oriented approach to the relationship between science and society. The aim is to gain a better understanding of communication phenomena between humans and machines, at a time when socio-numerical networks are being criticized for their tendency to generate hatred and facilitate the circulation of bitter discourse. The project aims to study the conditions under which emotions emerge and how they are expressed in dialogues between conversational agents and users. The aim is to better qualify the expression of emotional experiences in language practices and their role in communication processes with a machine. The case studies will be chosen in relation to societal themes with regional implications, such as eco-anxiety (ecological challenges) or assisted creation (audiovisual production), which give rise to the expression of individual or collective emotions.

The second ambition is to produce an experimental interdisciplinary investigative device in Information and Communication Sciences (ICS) and Computer Science. The central research question of this double doctoral project is the qualification of emotions in conversations between humans and machines: qualification in its cultural and social dimension, and in its formal dimension for machines. Based on this qualification work, the project aims to produce a taxonomy of emotions and a detection model. This type of model has societal implications in sectors where conversational agents will be used to support and develop practices, such as in the healthcare and food sectors. Better guidance and orientation, better information and communication all depend on a good understanding of users' emotions and adaptation to their needs, particularly in public policies.