Abstract
Humans are in perpetual movement, constantly traversing buildings, cities, waters, oceans, and countries. Mobility stands out as a major driving force shaping our modern societies. Capturing and explaining human behavior in a world of eight billion distinct mobility agendas is a complex challenge. With the rise of interconnected devices and platforms, such as smartphones, wearables, and point-of-interest data, largescale behavioral data has become more accessible, enabling rich insights into mobility patterns. However, the widespread availability of such data introduces significant ethical challenges. Detailed mobility data can inadvertently reveal sensitive personal information, including individuals' locations, habits, social interactions, and even political or religious affiliations. Beyond privacy breaches, the ethical implications of uncovering and potentially manipulating underlying behavioral patterns demand attention. Striking a balance between the utility of mobility models and the protection of individual privacy is therefore paramount. This paper explores the utility-privacy trade-offs in human mobility modeling, focusing on the impacts of introducing semantic and visit type ambiguities. By systematically examining how these ambiguities affect the fidelity of simulated trajectories and privacy risks, we provide a framework for evaluating ethical and privacy-conscious modeling practices. Our findings emphasize the need for methods that safeguard privacy without undermining the usefulness of mobility models, contributing to the responsible advancement of mobility science in alignment with ethical standards and societal expectations.
