PhD M/F - Informatique/Sociologie - Réduire les risques psycho-sociaux des interactions humain-IAg par l'analyse conversationnelle

Job Description

Thesis Subject

Generative AI (GenAI) tools, such as ChatGPT, are increasingly being used as companions (friendly, romantic, or sexual relationships) or personal advisors. The widespread adoption of everyday conversations between humans and GenAI systems has led to the emergence of public concerns regarding the risks associated with excessive use of these technologies, particularly addiction and behavioral manipulation, in some cases culminating in suicide.

Several research studies have explored the psychosocial risks associated with human relationships with GenAI systems in conversational settings, as well as the urgent need to develop tools capable of preventing risks linked to different forms of attachment.

This PhD project is interdisciplinary, combining Computer Science and Sociology. The social science component of the thesis has two main objectives. First, it seeks to understand the nature of the interactions that develop between users and GenAI systems, and how these interactions fit into the broader ecology of relationships that individuals maintain with both human and non-human actors. While there is an established body of research focusing on interactions with artificial beings (Velkovska et al., 2020), relatively little is known about the uses and implications of these new-generation tools. Second, the project aims to investigate how designers and developers respond to the moral panic surrounding these technologies by proposing technical solutions. From a Science and Technology Studies (STS) perspective, the research will analyze mechanisms of interdisciplinary translation: how psychological theories can be incorporated into technical systems.

The Computer Science component also has two main objectives. The first is the real-time automatic detection of the level of human attachment to AI through dialogue-based indicators (linguistic, contextual, and structural cues). The second is the development of real-time adaptation mechanisms for GenAI systems through conversational strategies aimed at preventing problematic forms of attachment (e.g., warnings, adjustments to conversational and relational alignment). One of the key challenges is to develop an innovative user-centered approach by involving a collective of users in the research process (ethics-by-design), enabling them to contribute to the design and evaluation of conversational strategies.

The sociological approach will make it possible to understand the nature of the interactions formed with these AI systems and the meanings they hold for users, as well as the challenges, risks, and discomforts they may experience. This approach relies on interviews and the analysis of interaction corpora. The goal will be to identify and measure indicators associated with these phenomena within the corpora (e.g., levels of attachment, signs of discomfort). The Computer Science approach will enable the development and evaluation of practical solutions for preventing risks in socio-emotional relationships. For example, automatic tools for detecting levels of attachment will be developed based on the indicators identified through sociological analysis and complemented by machine-learning exploration of interaction corpora.

Expected qualifications: interdisciplinary profile in Sociology and Computer Science.

Your Work Environment

TALEP Team, Laboratoire d'Informatique et des Systèmes (LIS), Campus de Luminy,

Constraints and risks

None

Compensation and benefits

Compensation

2300 € gross monthly

Annual leave and RTT

44 jours

Remote Working practice and compensation

Pratique et indemnisation du TT

Transport

Prise en charge à 75% du coût et forfait mobilité durable jusqu’à 300€

About the offer

About the CNRS

The CNRS is a major player in fundamental research on a global scale. The CNRS is the only French organization active in all scientific fields. Its unique position as a multi-specialist allows it to bring together different disciplines to address the most important challenges of the contemporary world, in connection with the actors of change.

CNRS

The research professions