PhD Thesis

My PhD thesis was entitled « Construction of Model Behavior by Novices through Interaction with Machine Learning ». I was supervised by Jules Françoise, Baptiste Caramiaux, and Michèle Gouiffès. In this thesis, I approached ML models as sociotechnical artifacts whose behavior takes shape through users’ interactions, and I studied how novice users could surface, negotiate, and contest it in context. I explored this through two empirical studies: one focused on novices collaboratively teaching models during the development phase, and another on helping them make sense of model outputs after deployment. Rather than trying to expose a single hidden logic, the thesis invites deeper engagement with the social, political, and epistemic conditions that shape ML systems, toward a more participatory and reflexive vision of human-centered AI. I hope this work opens space for practices that make ML systems more accountable, interpretable, and open to change.