Chair Reliable and Confident Use of LLMs

The emergence of Large Language Models (LLMs) represents a transformative milestone in

AI, with applications ranging from natural language processing (NLP) to content generation

and decision support systems in many areas such as software development, healthcare,

company management, education, etc. As these models become increasingly integrated into

critical domains such as healthcare and industry, their reliability and the confidence with which

they are used have become significant concerns. The Chair Reliable and Confident Use of

LLMs addresses these challenges through multidisciplinary research aimed at assessing and

enhancing the trustworthiness of LLMs in real-world settings. The chair, currently being

developed with the UCA Foundation, is set to launch in the beginning of 2025 with the following

partners: CGI Inc., Agaetis, Cantal Department.

In collaboration with these partners, the chair aims at bringing together researchers from

several laboratories (LIMOS, CLERMA, LMBP) to explore this topic through different but

complementary research axes, in order to study and foster LLM usages in real industrial

scenarios with reliability, confidence and confidentiality. The chair is organized into thematic

axes having their own scientific objectives, currently comprising two, namely the ‘Computer

Science / Software Engineering Axis’ and the ‘Sustainable Management of Companies

and Public Organizations Axis’ with the flexibility to incorporate additional axes in the future.

This structure, completed with periodic meetings and brainstorming will allow the sharing ofthe results obtained from different scientific disciplines to establish more complete and robust approaches.

Keywords: LLM, Reliability, Confidence, Validation, Confidentiality