Thèse le Problème « Novice - Expert » la Pensée Computationnelle et la Confiance dans les Nouveaux Développements en Matière d'IA en Météorologie H/F

Doctorat.Gouv.Fr

Établissement : Université Grenoble Alpes École doctorale : MSTII - Mathématiques, Sciences et technologies de l'information, Informatique Laboratoire de recherche : Laboratoire d'Informatique de Grenoble Direction de la thèse : Cassia TROJAHN DOS SANTOS ORCID 000000032840005X Début de la thèse : 2026-10-01 Date limite de candidature : 2026-07-31T23:59:59 AI systems are advancing rapidly, but their adoption and impact depend on more than technical capabilities. It requires balancing algorithmic performance with trust-building, encouraging computational thinking skills, and understanding the responses of users who interact with such systems.

Trust is central to AI development and deployment. Users need to believe that AI systems are reliable, safe, and aligned with human values. This includes trust in accuracy (does the AI give correct answers?), transparency (can its decisions be explained?), and fairness (does it avoid bias?). Without trust, even highly capable AI systems may face resistance or limited adoption. Trust may depend on knowledge of the domain and of AI. Domain experts do not necessarily have knowledge of AI.

The Novice/Expert problem refers to the paradox that evaluation of the answers of an AI system requires domain expertise, but the lack there off is the motive for using the AI system. AI systems lead to a four new situation in which users may be characterized as novices or experts both on the knowledge of the domain and on knowledge of AI. Computational thinking is an educational concept that emphasizes teaching and assessing computer science-related ways of thinking across all levels of schooling. These ways of thinking-including problem decomposition, abstraction, and algorithmic reasoning-are also relevant for understanding, designing, and interacting with AI system. It impacts the way engineers build models and helps users better understand how to interpret outputs and evaluate system behavior. It is increasingly seen as a core skill for working alongside AI.

A number of factors influence how people perceive and interact with AI. These include cognitive biases (such as over-trusting automated outputs or fearing machine decisions), emotional responses (such as anxiety about job displacement or excitement about productivity gains), and social perceptions (such as viewing AI as authoritative or human-like). These factors affect how much users rely on AI, how they judge its credibility, and whether they accept or reject its recommendations.
Advisor: The thesis will be advised by Cássia Trojahn dos Santos - LIG (****@****.**) and Erica De Vries - LaRAc (****@****.**).

Group: The work will be carried out in the mOeX team common to INRIA & LIG in cooperation with the LaRac laboratory.

Place of work: The position is located at INRIA Montbonnot (near Grenoble) a main computer science research lab, in a stimulating research environment.
L'objectif principal de cette thèse est d'examiner comment les facteurs humains et cognitifs influencent le développement, l'adoption et l'utilisation efficace des systèmes d'IA modernes, selon ces trois dimensions. Elle vise à explorer la relation entre les principes de conception technique et la perception humaine dans la mise en forme de l'interaction avec l'IA.