Crédit Agricole S.A., École Polytechnique and the École Polytechnique Foundation are launching the International Chair for Teaching and Research in Trustworthy and Responsible Artificial Intelligence (AI), with the aim of contributing to the development of secure, ethical and robust AI-based systems. The new Chair was inaugurated on 26 March 2024 by Laura Chaubard, Director General and Acting President of the École Polytechnique, Olivier Gavalda, Deputy CEO of Crédit Agricole S.A. in charge of Universal Banking, and Jean-Paul Cottet, Delegate general of the École Polytechnique Foundation.
Supported by Crédit Agricole’s DataLab, the programme will aim to contribute to the development of systems based on trustworthy artificial intelligence* that are ethical, robust, secure, and sustainable. Sonia Vanier, researcher at the Computer Science Laboratory of the École Polytechnique (LIX**), is the head of the Chair.
The research project has two main focuses. On the one hand, it will focus on new ways of improvement for more reliable AI systems, using a hybrid approach between several disciplines and drawing on different research communities: symbolic AI, operations research, connectionist AI and machine learning. Integrating the reasoning and values derived from the trade knowledge of operational teams into these systems will enable the construction of models leading to accurate and coherent decision-making in relation to use cases. To achieve trustworthy AI*, it will be necessary to detect and correct the biases present in the training data and at risk of being reproduced by the AI. Another key point will be to develop models whose progress can be easily interpreted, with results that can be explained to users.
On the other hand, the project aims to achieve more environmentally responsible AI. The latest technologies, such as recent generative AI models and large language models, require more and more data, computing time, and IT and energy resources. In this context, the project aims to limit this environmental impact, through the implementation of AI with a lower environmental impact, and the design of AI algorithms for applications with a positive impact, for example to optimise the energy consumption of (connected) buildings.