Mathieu Chevrier

I am a behavioral and experimental economist specialising in research on human–algorithm interaction. After completing my PhD at Côte d'Azur University under the supervision of Eric Guerci and Brice Corgnet on algorithm perception, I am currently pursuing Postdoctoral research on the impact of AI on human, under the supervision of Agnès Festré.

Together, we are conducting the Learn-AI project, funded by the IDEX programme of Université Côte d’Azur.
Since 2022, I have also served as co-manager of the Laboratory of Experimental Economics in Nice.

My research focuses on human–algorithm interaction and is structured around 4 axes.

• First, together with Brice Corgnet, Eric Guerci, and Julie Rosaz, I study the impact of algorithmic complexity on people’s credulity toward algorithms. We show that the more an algorithm is perceived as complex, the more individuals tend to lower their epistemic vigilance and become willing to follow erroneous algorithmic advice.
• A second research axis, initiated with Sébastien Massoni, explores the role of metacognition in the acceptance of AI-based recommendations by decision-makers. We show that, beyond the competence of the algorithm itself, the decision-maker’s metacognitive ability is crucial for appropriately following algorithmic recommendations.
• A third axis focuses on perceptions of algorithmic responsibility. Together with Vincent Teixeira, I examine who is perceived as responsible when an algorithm makes an unequal decision: the programmer or the user of the algorithm. We show that, depending on the model used, individuals may perceive an AI user as more responsible than a user relying on a rule-based algorithm.
• Finally, as part of my postdoctoral research, my current focus is on the impact of AI on human learning. When individuals delegate a task to AI, they may cognitively disengage, which can reduce learning. Through a series of laboratory experiments, we are currently investigating how AI affects human learning and how the negative consequences of this relationship can be mitigated.

Keywords: Algorithmic Acceptance, Advice, Delegation, Experimental Economics, Responsibility, Trust.