Toward an empathy-based trust in human-otheroid relations

AI & SOCIETY https://doi.org/10.1007/s00146-024-02155-z MAIN PAPER Toward an empathy‑based trust in human‑otheroid relations Abootaleb Safdari1 Received: 20 June 2024 / Accepted: 29 November 2024 © The Author(s) 2024 Abstract The primary aim of this paper is twofold: firstly, to argue that we can enter into relation of trust with robots and AI systems (automata); and secondly, to provide a comprehensive description of the underlying mechanisms responsible for this relation of trust. To achieve these objectives, the paper first undertakes a critical examination of the main arguments opposing the concept of a trust-based relation with automata. Showing that these arguments face significant challenges that render them untenable, it thereby prepares the ground for the subsequent positive analysis, proposing a framework in which these challenges can be addressed . According to this framework trust does not originate from mere reliability, but rather from an empathic relation with automata. This initial empathic relation elevates the automata to the status of what I will term "Otheroids." The paper then explores how this human-Otheroid relationship inherently possesses the seeds for the development of trust. Finally, it examines how these seeds can grow into a basic form of trust with Otheroids through the establishment of a rich history of interaction. Keywords Trust-based relation · Empathy · Otheroids · Human-AI/Robot-relation · Relational Attitude 1 Introduction Trust is a multifaceted concept, encompassing a range of varied cognitive, emotional, and behavioral components, and its conceptualization and definition are the subjects of an extensive literature. Consequently, this complexity has come to influence how we understand trust-based relations with automata. Despite this complexity, the mainstream perspective posits that the primary mechanism for establishing trust is the cognitive assessment of the automata's reliability. As we interact with them, we gather information about their capabilities, behaviors, and past interactions. This information shapes our expectations about how they will behave in future encounters, leading to a sense of trust based on our judgment of the automata’s reliability. This mechanism aligns closely with what has been termed reliability (Baier 1986; Hawley 2014; Holton 1994) or functional trust (Myskja 2008) (a more in-depth discussion is provided in the subsequent sections). It is along this line of thought that researchers argue it is nonsense to talk about entertaining * Abootaleb Safdari 1 Institute for Philosophy, Bremen Universität, Bremen, Germany trust-based relations with AIs and robots, claiming instead that the appropriate way to describe the relations we have with them is in terms of attributing reliability and predictability (and not trustworthiness) (Alvarado 2023; Bryson 2018; Hatherley 2020; Metzinger 2019; Ryan 2020). In the following sections I will discuss this line of thought in detail and argue why it fails to explain the character of our interactions with automata. The paper consists of two main steps, one negative and the other positive. In the negative step, I examine the most important arguments against the possibility of humans entering into relations of trust with automata. The first of these is a conceptual argument, that employing the concept of trust in the context of human-automata-relations (HARs) is conceptually nonsensical. The second argument emphasizes the phenomenological differences between trust-based and reliability-based relationships, arguing that relations with automata falls into the latter category rather than the former. Finally, there is the argument from harmfulness, which contends that to establish relations of trust with automata would be socially and morally harmful. Critically examining these arguments will pave the way for my positive framework. In the positive step, I propose an empathy-based framework for entering into relations of trust with automata. Initially, we build an empathic relation with them, leading us Vol.:(0123456789) AI & SOCIETY to perceive and interact with them not as mere technological artifacts but as minded others or, more precisely, pseudoothers. This is why I prefer to call them "otheroids." I then provide a clear understanding of how interaction with otheroids results in a relation of trust with them. Here, the process of building a history of interaction is at the heart of my proposed framework. 2 Automata: reliable tools, not trustworthy companions The debate surrounding trust in AI has gained significant traction in the wake of the publication of the Ethics guidelines for trustworthy AI by the High-level expert group on artificial intelligence established by the European Commission (Ethics Guidelines for Trustworthy AI 2019). A central question in this debate is whether the concept of trust is even applicable in the context of AI. Opponents argue it is not applicable for three main reasons: First, trust is inherently human and cannot be meaningfully applied to AI systems; thus its application in the context of HAR is a kind of category mistake. Second, trust-based relationships are distinguished by the unique phenomenological experience of betrayal, as opposed to mere disappointment, upon their violation. Consequently, the absence of such a feeling of betrayal when interacting with automata suggests that our relationship with them cannot be classified as one of trust. And third, that to have trusting relations with automata would be harmful or dangerous.1 Thus we have to prevent this misplaced trust from arising. In what follows I first reconstruct these lines of argument and then illustrate that all of them suffer challenging problems. I then proceed to my positive proposal. not trustworthy; only humans can be trustworthy (or untrustworthy) (2019). Joanna Bryson is another well-known expert in the field who emphasizes that no one actually can trust AI (2018). Lets have a closer look at their argument(s). This radical line of argument often rests upon a specific conception of trust known as agent-centered or agential trust2 (Nguyen 2022; Viehoff 2023). Accordingly, trust starts with reliability and evolves into trustworthiness based on the presence of agential capacities. In simpler terms, trust is seen as the sum of reliability plus agential capacities. Since technological artifacts like AI and robots lack these agential capacities, proponents of this view argue, our relationship with them can only be one of reliance, not trust. The overall structure of these arguments is thus as follows: i. Genuine trust necessitates the presence of component X (where reliability + X=trust); ii. Automata lack X; iii. Therefore, the notion of trust is not applicable in the context of HAR iv. Therefore, reliability is the more suitable concept, rather than trust, to employ in the context of HARs. Trust, in this context, may be applicable only indirectly— we plac (...truncated)