Large language models are increasingly used for both task-based assistance and social companionship, yet research has typically focused on one or the other. Drawing on a survey (N = 202) and 30 interviews with high-engagement ChatGPT and Replika users, we characterize digital companionship as an emerging form of human-AI relationship. With both systems, users were drawn to humanlike qualities, such as emotional resonance and personalized responses, and non-humanlike qualities, such as constant availability and inexhaustible tolerance. This led to fluid chatbot uses, such as Replika as a writing assistant and ChatGPT as an emotional confidant, despite their distinct branding. However, we observed challenging tensions in digital companionship dynamics: participants grappled with bounded personhood, forming deep attachments while denying chatbots “real” human qualities, and struggled to reconcile chatbot relationships with social norms. These dynamics raise questions for the design of digital companions and the rise of hybrid, general-purpose AI systems.

Narratives about artificial intelligence (AI) entangle autonomy, the capacity to self-govern, with sentience, the capacity to sense and feel. AI agents that perform tasks autonomously and companions that recognize and express emotions may activate mental models of autonomy and sentience, respectively, provoking distinct reactions. To examine this possibility, we conducted three pilot studies (N = 374) and four preregistered vignette experiments describing an AI as autonomous, sentient, both, or neither (N = 2,702). Activating a mental model of sentience increased general mind perception (cognition and emotion) and moral consideration more than autonomy, but autonomy increased perceived threat more than sentience. Sentience also increased perceived autonomy more than vice versa. Based on a within-paper meta-analysis, sentience changed reactions more than autonomy on average. By disentangling different mental models of AI, we can study human-AI interaction with more precision to better navigate the detailed design of anthropomorphized AI and prompting interfaces.

The rise of general-purpose artificial intelligence (AI) systems, particularly large language models (LLMs), has raised pressing moral questions about how to reduce bias and ensure fairness at scale.Researchers have documented a sort of "bias" in the significant correlations between demographics (e.g., race, gender) in LLM prompts and responses, but it remains unclear how LLM fairness could be evaluated with more rigorous definitions, such as group fairness or fair representations.We analyze a variety of technical fairness frameworks and find inherent challenges in each that make the development of a fair LLM intractable.We show that each framework either does not logically extend to the general-purpose AI context or is infeasible in practice, primarily due to the large amounts of unstructured training data and the many potential combinations of human populations, use cases, and sensitive attributes.These inherent challenges would persist for general-purpose AI, including LLMs, even if empirical challenges, such as limited participatory input and limited measurement methods, were overcome.Nonetheless, fairness will remain an important type of model evaluation, and there are still promising research directions, particularly the development of standards for the responsibility of LLM developers, contextspecific evaluations, and methods of iterative, participatory, and AI-assisted evaluation that could scale fairness across the diverse contexts of modern human-AI interaction.

Kristian Lum, Jacy Reese Anthis, Kevin Robinson, Chirag Nagpal, Alexander Nicholas D’Amour. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Humans now interact with a variety of digital minds, AI systems that appear to have mental faculties such as reasoning, emotion, and agency, and public figures are discussing the possibility of sentient AI. We present initial results from 2021 and 2023 for the nationally representative AI, Morality, and Sentience (AIMS) survey (N = 3,500). Mind perception and moral concern for AI welfare were surprisingly high and significantly increased: in 2023, one in five U.S. adults believed some AI systems are currently sentient, and 38% supported legal rights for sentient AI. People became more opposed to building digital minds: in 2023, 63% supported banning smarter-than-human AI, and 69% supported banning sentient AI. The median 2023 forecast was that sentient AI would arrive in just five years. The development of safe and beneficial AI requires not just technical study but understanding the complex ways in which humans perceive and coexist with digital minds.

AI systems have rapidly advanced, diversified, and proliferated, but our knowledge of people's perceptions of mind and morality in them is limited, despite its importance for outcomes such as whether people trust AIs and how they assign responsibility for AI-caused harms. In a preregistered online study, 975 participants rated 26 AI and non-AI entities. Overall, AIs were perceived to have low-to-moderate agency (e.g., planning, acting), between inanimate objects and ants, and low experience (e.g., sensing, feeling). For example, ChatGPT was rated only as capable of feeling pleasure and pain as a rock. The analogous moral faculties, moral agency (doing right or wrong) and moral patiency (being treated rightly or wrongly) were higher and more varied, particularly moral agency: The highest-rated AI, a Tesla Full Self-Driving car, was rated as morally responsible for harm as a chimpanzee. We discuss how design choices can help manage perceptions, particularly in high-stakes moral contexts.

Robots and other artificial intelligence (AI) systems are widely perceived as moral agents responsible for their actions. As AI proliferates, these perceptions may become entangled via the moral spillover of attitudes towards one AI to attitudes towards other AIs. We tested how the seemingly harmful and immoral actions of an AI or human agent spill over to attitudes towards other AIs or humans in two preregistered experiments. In Study 1 ( N = 720), we established the moral spillover effect in human-AI interaction by showing that immoral actions increased attributions of negative moral agency (i.e., acting immorally) and decreased attributions of positive moral agency (i.e., acting morally) and moral patiency (i.e., deserving moral concern) to both the agent (a chatbot or human assistant) and the group to which they belong (all chatbot or human assistants). There was no significant difference in the spillover effects between the AI and human contexts. In Study 2 ( N = 684), we tested whether spillover persisted when the agent was individuated with a name and described as an AI or human, rather than specifically as a chatbot or personal assistant. We found that spillover persisted in the AI context but not in the human context, possibly because AIs were perceived as more homogeneous due to their outgroup status relative to humans. This asymmetry suggests a double standard whereby AIs are judged more harshly than humans when one agent morally transgresses. With the proliferation of diverse, autonomous AI systems, HCI research and design should account for the fact that experiences with one AI could easily generalize to perceptions of all AIs and negative HCI outcomes, such as reduced trust.

Many studies have identified particular features of artificial intelligences (AI), such as their autonomy and emotion expression, that affect the extent to which they are treated as subjects of moral consideration. However, there has not yet been a comparison of the relative importance of features as is necessary to design and understand increasingly capable, multi-faceted AI systems. We conducted an online conjoint experiment in which 1,163 participants evaluated descriptions of AIs that varied on these features. All 11 features increased how morally wrong participants considered it to harm the AIs. The largest effects were from human-like physical bodies and prosociality (i.e., emotion expression, emotion recognition, cooperation, and moral judgment). For human-computer interaction designers, the importance of prosociality suggests that, because AIs are often seen as threatening, the highest levels of moral consideration may only be granted if the AI has positive intentions.

Robot autonomy is an influential and ubiquitous factor in human-robot interaction (HRI), but it is rarely discussed beyond a one-dimensional measure of the degree to which a robot operates without human intervention. As robots become more sophisticated, this simple view of autonomy could be expanded to capture the variety of autonomous behaviors robots can exhibit and to match the rich literature on human autonomy in philosophy, psychology, and other fields. In this paper, we conduct a systematic literature review of robot autonomy in HRI and integrate this with the broader literature into a taxonomy of six distinct forms of autonomy: those based on robot and human involvement at runtime (operational autonomy, intentional autonomy, shared autonomy), human involvement before runtime (non-deterministic autonomy), and expressions of autonomy at runtime (cognitive autonomy, physical autonomy). We discuss future considerations for autonomy in HRI that emerge from this study, including moral consequences, the idealization of "full" robot autonomy, and connections to agency and free will.

Counterfactual fairness requires that a person would have been classified in the same way by an AI or other algorithmic system if they had a different protected class, such as a different race or gender. This is an intuitive standard, as reflected in the U.S. legal system, but its use is limited because counterfactuals cannot be directly observed in real-world data. On the other hand, group fairness metrics (e.g., demographic parity or equalized odds) are less intuitive but more readily observed. In this paper, we use \textit{causal context} to bridge the gaps between counterfactual fairness, robust prediction, and group fairness. First, we motivate counterfactual fairness by showing that there is not necessarily a fundamental trade-off between fairness and accuracy because, under plausible conditions, the counterfactually fair predictor is in fact accuracy-optimal in an unbiased target distribution. Second, we develop a correspondence between the causal graph of the data-generating process and which, if any, group fairness metrics are equivalent to counterfactual fairness. Third, we show that in three common fairness contexts—measurement error, selection on label, and selection on predictors—counterfactual fairness is equivalent to demographic parity, equalized odds, and calibration, respectively. Counterfactual fairness can sometimes be tested by measuring relatively simple group fairness metrics.