Generate erroneous reasoning to enhance explainable AI H/F

This internship aims to improve human-AI collaboration, where accuracy alone is insufficient as trust requires understanding the AI's reasoning. Explainable AI (XAI) focuses on generating clear, faithful, concise, and user-adapted explanations, since a poor explanation can harm cooperation. The work concerns ExpressIF®, a symbolic fuzzy-logic AI from CEA. Though interpretable, it may be hard to select the most relevant rules for users.

The project addresses human-AI disagreement: the AI must detect possible human reasoning errors (lack of knowledge, confirmation or attention bias, illusory correlation, or uncertainty differences) to adapt its explanations. Goals: study fuzzy logic, ExpressIF®, and cognitive biases; design a model of user reasoning errors; integrate and evaluate it within ExpressIF®.

Durée du contrat (en mois)

We are looking for a second-year Master's student or third-year engineering school student specializing inComputer Science and/or Mathematics, with the following profile :

• A strong interest in artificial intelligence topics, particularly fuzzy logic ;
• A strong interest in scientific and technological research ;
• A genuine interest in cognitive sciences ;
• Solid object-oriented programming skills, ideally in C# ;
• Strong English skills., Diplôme préparé

Bac+5 - Master 2

Possibilité de poursuite en thèse

Entité de rattachement Le CEA est un acteur majeur de la recherche, au service des citoyens, de l'économie et de l'Etat. Il apporte des solutions concrètes à leurs besoins dans quatre domaines principaux : transition énergétique, transition numérique, technologies pour la médecine du futur, défense et sécurité sur un socle de recherche fondamentale. Le CEA s'engage depuis plus de 75 ans au service de la souveraineté scientifique, technologique et industrielle de la France et de l'Europe pour un présent et un avenir mieux maîtrisés et plus sûrs. Implanté au cœur des territoires équipés de très grandes infrastructures de recherche, le CEA dispose d'un large éventail de partenaires académiques et industriels en France, en Europe et à l'international. Les 20 000 collaboratrices et collaborateurs du CEA partagent trois valeurs fondamentales, Located in Saclay, in the southern Île-de-France region, CEA LIST is a scientific and technological research center dedicated to the development of software, embedded systems, and sensors for applications in defense, security, energy, nuclear, environment, and healthcare. CEA LIST has more than 700 researchers focusing on intelligent digital systems, with core areas including artificial intelligence, the factory of the future, innovative instrumentation, cyber-physical systems, and digital health. Within the institute, the laboratories of the Digital Instrumentation Department work on developing and transferring cutting-edge AI technologies to industry. The technical scope of our engineers and researchers covers signal analysis (i.e., time series as well as spectra) produced by equipment developed internally by CEA teams or by external companies. The exploitation of these data relies on a broad range of machine learning methods, including both numerical AI (deep neural networks, random forests, SVMs) and symbolic AI (knowledge-based systems). You will join a team of five permanent researchers working on the ExpressIF platform, a fully in-house developed symbolic artificial intelligence system designed to be interpretable. The goal is not to build a "black box" AI-whose reasoning process cannot be understood by humans-that simply provides decisions to an operator, but rather an AI that can be understood: experts can therefore extract insights for their own research, correct knowledge if it has been misled by data, or simply enrich it when certain elements are missing.