Responsibility in Hybrid Societies: concepts and terms

Abstract

With increased digitalization and new technologies, societies are expected to no longer only include human actors, but artificial actors as well. Such a future of societies raises new questions concerning the coexistence, tasks and responsibilities of different actors. Manifold disciplines are involved in the creation of these future societies. This requires a common understanding of responsibility, and of definitions of actors in Hybrid Societies. This review aims at clarifying aforementioned terms from a legal and psychological perspective. Building from this common ground, we identified seven capacities in total which need to be met by actors in societies to be considered fully responsible, in both a legal and moral sense. From a legal perspective, actors need to be autonomous, have capacity to act, legal capacity, and the ability to be held liable. From a psychological perspective, actors need to possess moral agency and can be trusted. Both disciplines agree that explainability is a pivotal capacity to be considered fully responsible. As of now, human beings are the only actors who can, with regard to these capacities, be considered morally and legally responsible. It is unclear whether and to which extent artificial entities will have these capacities, and subsequently, whether they can be responsible in the same sense as human beings are. However, on the basis of the conceptual clarification, further steps can now be taken to develop a concept of responsibility in Hybrid Societies.

Artificial Morality: Differences in Responses to Moral Choices by Human and Artificial Agents

“…Overall, human agents received considerably more blame but were also more trusted. Blame requires attribution of moral responsibility which in turn depends on various legal and psychological capacities (e.g., the capacity to act, legal capacity, autonomy, liability, explainability, and moral agency) most of which are usually not ascribed to artificial agents (Meyer et al, 2023). Our finding of lower blame ratings for artificial agents (for making the same decisions as their human counterparts) indicate that they are held less responsible.…”

“…Although studies investigating responses to artificial moral agents cannot sufficiently address the question of whether these agents are actually capable of making moral choices (cf. Meyer et al, 2023), findings nevertheless yield important information. Understanding human perception of and responses to artificial agents in situations with moral implications is essential for optimizing these interactions with due consideration of their psychological and legal limitations.…”

“…However, the ongoing rapid development of the capabilities of artificial intelligence (AI) might give rise to a new type of moral agent. Thus, for future societies of humans coexisting with various types of artificial and hybrid agents to function, questions need to be addressed regarding their respective tasks and responsibilities (Meyer et al, 2023). Given their current rapid development, these new agents will likely gain in autonomy and their decisions might (to varying degrees) be dissimilar from those that humans might take (Gesmann-Nuissl, 2018).…”

“…Although there is no complete agreement among all humans even on key moral issues and considerable individual and cultural differences exist in human moral decision-making (Friesdorf et al, 2015;Graham et al, 2016), disagreements between humans and artificial agents on moral questions represent a new type of challenge. Since such disagreements also touch on the more fundamental issue of roles, rights, and responsibilities of humans and the emerging different artificial societal actors (Meyer et al, 2023), this challenge needs to be timely met. Thus, the question of how humans judge and respond to the decisions of artificial moral agents warrants further investigation, even iffor nowthis concerns decisions in fictional scenarios.…”

“…In sum, there is growing interest into how artificial moral agents are evaluated by human observers accompanied by an ongoing debate about whether and to what extent such agents are actually capable of moral decisions and whether they should be allowed to make them (Malle, 2016;Meyer et al, 2023;Misselhorn, 2018).…”