Deception in social robotics: problematic profiles of human-robot interaction and the universality of human vulnerability

This thesis examines the theoretical and procedural aspects of assessing deception in social robotics, with the aim of addressing the potential harmful effects and benefits for users in a holistic manner. In order to achieve this, we begin by elucidating the intended meaning of the term 'social robot' and the characteristics of the phenomenon of deception within the context of human-robot interaction (HRI). The aim is to identify a conceptualisation of this term that can bridge the critical vision of social sciences with the functional aspect attributed within robotics and computer science. In particular, the ability of these robots to evoke an appearance of sociality, intelligence, and emotionality, with the aim of facilitating user acceptance and co-operation, is analysed. Subsequently, European legislation that has predominantly focused on regulating practices that can manipulate individuals is delineated: the DSA, UCDP, and AI Act. Then, it is proved how the approaches there employed there are inadequate in safeguarding the actual 'average user', who does not adhere to the principles of perfect rationality assumed by traditional legal models. It is therefore proposed that the most appropriate theoretical framework for analysing the bounded rationality of individuals in HRI is Vulnerability Theory. This is then used as the basis for developing a procedural analysis, which aims to become a tool to guide legal and technical experts in a more verosimilar and all-encompassing analysis of HRI dynamics, so as to ultimately facilitate the identification of solutions in regulatory and design terms for the development of social robotics that are truly human-centric.