Carina Ines Hausladen

I am an Assistant Professor of Computational Social Science. My research focuses on understanding and measuring the societal impacts of AI. My work combines causal inference, computational social science, machine learning, and behavioral modeling to study how AI systems shape social perception, decision-making, and collective outcomes. In one line of research, I developed methods to causally measure social perception in vision-language models; we thus provide rigorous tools for evaluating social biases of AI systems (https://doi.org/10.1145/3715275.3732041). This work forms the basis of a graduate-level course on fairness, bias, and collective behavior in AI systems: https://github.com/carinahausladen/konstanz-fairness-collective-ai. I have also studied the use of LLM-based digital twins in democratic settings; we show that current models fail to adequately capture the diversity of human preferences; these are important limitations for AI-mediated representation and decision-making (https://doi.org/10.1609/aies.v7i1.31758). In another line of research, I use inverse Q-learning to model cooperation and social dilemmas, with a focus on understanding strategic behavior in environments where AI agents increasingly interact with humans and with one another (https://arxiv.org/pdf/2601.08803). This research is closely connected to a graduate-level course on computational models of social behavior; we cover computational game theory, reinforcement learning, theory of mind, cooperation, LLM-based agents, and LLM world simulations: https://github.com/carinahausladen/konstanz-dynamic-social-behavior.