Do Artificial Agents Merely Simulate Human Sociality? An Empathy-Based Framework

Abstract This article critically examines simulation-based approaches to human–artificial agent interaction and proposes, in their place, an empathy-based framework. Simulationism holds that sociality with robots is nothing more than a simulation of human–human sociality: Robots do not introduce new forms of social relations but merely simulate the outward patterns of human interaction. This idea has found its most elaborated formulation in the Ontology of Asymmetric Social Interactions (OASIS). OASIS proposes that social interactions can be fundamentally asymmetrical, marked by unequal distributions of social capacities between interaction partners. Human–robot and human–animal interactions are typical instances of such asymmetrical sociality, while human–human interactions exemplify symmetrical sociality, where mutual agency and understanding are fully shared. While it presents a nuanced view by distinguishing between “socializing” and “humanizing,” it will be argued that its reliance on a simulationist foundation ultimately restricts its ability to move beyond an anthropocentric perspective. Drawing on recent phenomenology and enactivism, a more radical, non-anthropocentric framework will be proposed. This framework shifts the focus from a simulationist model—assessing how well robots simulate human behavior—to an examination of the systemic and holistic nature of embodied interactions themselves. The argument is that by focusing on the body and the structure of human–robot couplings, a more robust understanding of the new forms of sociality could be developed that are emerging with artificial agents, recognizing them not as human simulators but as other forms of otherness or otheroids.