Human-Centered Reflections on Care Robots: A Comparative Study of Caregiver Perspectives

Human-Centered Reflections on Care Robots: A Comparative Study of Caregiver Perspectives Laura Londoño1,2 , Klaus Baumann3 , Abhinav Valada2 , Markus Langer1 1

2

Department of Psychology, University of Freiburg, Germany. Department of Computer Science, University of Freiburg, Germany. 3 Department of Theology, University of Freiburg, Germany.

arXiv:2608.02411v1 [cs.RO] 3 Aug 2026

Corresponding author e-mail: londono@cs.uni-freiburg.de; Abstract Care robots are increasingly being introduced into healthcare settings, raising important questions about their acceptance and ethical implementation. To better understand these challenges, this study investigates caregivers’ perceptions of four categories of care robots: delivering supplies, helping patients into bed, monitoring vital signs, and assisting with mobility. We conducted a mixed-methods study employing a mixed-factorial design in which 298 caregivers from the United States, Mexico, and Chile evaluated all four robot categories. Quantitative measures integrated constructs from the Unified Theory of Acceptance and Use of Technology, the Cognitive–Affective–Normative model, and overall acceptance ratings. Qualitative data were collected through open-ended questions and analyzed using a literature-informed ethical framework. The results indicate that participants across countries generally evaluated care robots positively, particularly for logistical and physically demanding tasks rather than those requiring intensive interpersonal interaction. The qualitative findings provide further insight into stakeholders’ views of the ethical implications of care robot use. Participants emphasized potential benefits such as reduced workload, lower risk, and greater patient autonomy, while also expressing concerns about dependability, the need for human oversight, and potential job displacement. Although many ethical concerns were shared across countries, participants differed in how they interpreted and prioritized them. These findings advance a context-sensitive and socially informed understanding of responsible design and implementation of care robots. Keywords: Care Robots, Technology Acceptance, Human-Robot Interaction, Cross-Cultural Comparison, Robot Ethics, Responsible Robotics

1 Introduction

patients, and participate in practices traditionally grounded in human relationships [5–8]. Addressing these questions requires more than improving robots’ technical performance and usability. It also demands careful consideration of the ethical, social, cultural, and emotional dimensions of care. As part of broader efforts to promote the responsible development of care robots and foster their acceptance, scholars have emphasized

The use of robots has expanded beyond industrial applications into a range of diverse social domains, including healthcare and long-term care [1–4]. As robots become increasingly integrated into those settings characterized by vulnerability and dependence, their presence raises critical questions about how they should interact with caregivers, support

1

the importance of incorporating the perspectives of stakeholders directly involved in the design, deployment, and use [9–13]. For example, [8, 14] argues that meaningful acceptance requires a thorough understanding of why end users accept or reject care robots in social contexts. They argue that responsible and context-sensitive care robotics should align robots’ technical capabilities and forms of agency with the goals, values, and needs of professional care practices. Despite growing interest in the ethical design and acceptance of care robots, stakeholders from countries outside the so-called “developed world” remain underrepresented in the literature [15–17]. This imbalance may introduce systematic biases into the development of care technologies by privileging culturally narrow assumptions about care practices, technological expectations, and user needs. Moreover, acceptance and ethical concerns may depend not only on sociocultural context but also on the specific function assigned to a robot. A robot that transports supplies, for example, may be evaluated according to different expectations and ethical criteria than one that physically assists or monitors a patient. To address this gap, this study investigates (1) caregivers’ perceptions of four categories of care robots: robots that deliver supplies, help patients into bed, monitor vital signs, and assist with mobility. The study compares stakeholders from the United States, Mexico, and Chile, three countries with distinct socioeconomic, cultural, institutional, and technological contexts; and (2) it examines both the acceptance of these robot categories and the ethical considerations that stakeholders regard as important for their responsible design and deployment. Our analysis is guided by three premises: (1) acceptance cannot be understood independently of ethical considerations, as perceptions of a robot’s benefits, risks, and impact on care shape users’ willingness to adopt it; (2) a cross-country perspective is needed to understand how social, cultural, and institutional contexts influence these perceptions; and (3) acceptance is task-specific, meaning evaluations depend on the robot’s function and care context and cannot be readily generalized across applications. Accordingly, this study pursues two main objectives (see Figure 1):

Fig. 1: Overview of the comparative mixed-methods design. Quantitative acceptance analyses (UTAUT and CAN) and qualitative ethical evaluations were conducted in parallel and subsequently integrated to develop a comprehensive understanding of the acceptance of four types of care robots across sociocultural contexts.

• Aim 1: To experimentally evaluate caregivers’ acceptance of four categories of healthcare robots across three countries (the United States, Mexico, and Chile) with distinct socioeconomic and technological contexts. Acceptance is assessed using constructs derived from the Unified Theory of Acceptance and Use of Technology (UTAUT) [18] and Cognitive–Affective–Normative (CAN) [19] model, together with an overall acceptance rating. The analysis identifies factors associated with acceptance and compares acceptance profiles across robot categories and countries. • Aim 2: To develop and apply a literatureinformed ethical framework for examining the considerations that caregivers regard as important for the responsible design and deployment of the four categories of care robots across different sociocultural contexts. To address the objectives, we employ a mixedmethods study that combined quantitative and qualitative approaches within a mixed-factorial framework. Robot category served as the withinsubjects factor, as each participant evaluated all four categories of care robots. Sociocultural context served as the between-subjects factor and was operationalized through participants’ country of residence and language. Participants completed an online survey in which the four robot categories, each representing distinct clinical functions, were presented in randomized order. The quantitative component collected standardized measures of acceptance for each category. The qualitative

2

component used open-ended questions to elicit participants’ views on the ethical considerations relevant to the design and deployment of care robots. These responses enabled us to identify recurring ethical themes that standardized quantitative measures might not capture. Integrating the quantitative and qualitative findings allowed us to examine not only whether stakeholders accepted different types of care robots, but also how they understood the ethical considerations associated with care robots. Accordingly, this study makes three contributions. First, it provides a within-participant comparison of caregivers’ evaluations of four care-robot scenarios representing distinct clinical functions. Second, it integrates quantitative acceptance measures with qualitative ethical analysis to identify how concerns relating to safety, autonomy, relational care, accountability, and accessibility shape and qualify participants’ evaluations of these technologies. Third, it offers comparative evidence on similarities and differences between the sampled groups, comprising English-language participants in the United States and Spanish-language participants in Mexico and Chile.

UTAUT provides a comprehensive account of technology adoption by incorporating social, organizational, and contextual factors that influence both behavioral intention and actual technology use. It proposes four core constructs: performance expectancy, effort expectancy, social influence, and facilitating conditions. This model recognizes that their influence varies according to demographic and situational factors such as age, gender, prior experience, and voluntariness of use [21]. In addition to technology acceptance models, other frameworks can contribute to the evaluation of emerging technologies by capturing dimensions beyond instrumental acceptance. One example is the Cognitive–Affective–Normative (CAN) model proposed by Reinares-Lara et al. [19], which incorporates ethical considerations by assessing users’ perceptions of fairness, morality, and other normative aspects of technology evaluation. While the CAN model enables the measurement of ethical perceptions, it offers limited insight into the underlying values and forms of ethical reasoning that shape these judgments. This limitation motivates the use of complementary qualitative approaches to better understand stakeholders’ ethical evaluations of emerging technologies.

2 Related Work

2.2 Ethical Evaluation of Care Robots

This section provides a focused review of the literature on technology acceptance. It also examines key findings on ethical evaluations of robot design, as well as research highlighting the importance of robot type and task characteristics in shaping user acceptance. In addition, the section considers cross-cultural variations in the adoption and acceptance of robotics technology. Finally, it analyzes the healthcare contexts of the United States, Mexico, and Chile to situate the study’s findings within their respective sociocultural and structural conditions.

While ethical reflection surrounding care robots has received increasing attention within the fields of responsible robotics, much of the existing literature has focused on developing normative frameworks, ethical principles, and design guidelines [22–25]. Comparatively little attention has been devoted to understanding how these normative insights can be integrated with empirical investigations or how ethical considerations shape the acceptance, rejection, and evaluation of care robots in practice [26, 27, 27, 28]. As a result, a persistent divide remains between normative ethical theorizing and empirical research [29]. In response, there has been growing recognition of the need for research approaches that move beyond this traditional separation [14, 21, 24]. Among the most influential contributions in this area is the work of Van Wynsberghe [27], who develops an ethical framework for evaluating the design, implementation, and use of care robots.

2.1 Technology Acceptance Models Technology acceptance has been extensively studied across multiple disciplines, leading to the development of numerous theoretical frameworks and standardized measurement instruments. Among the most influential is the Unified Theory of Acceptance and Use of Technology (UTAUT) [18], which integrates and extends the Technology Acceptance Model (TAM) [20] and other earlier frameworks. 3

• Care Ethics: Originally developed by Gilligan [30] as a critique of justice-based moral theories, the ethics of care emphasizes the moral significance of relationships, interdependence, and contextual responsibilities. From this perspective, ethical judgment arises from attentiveness to concrete relational contexts and the needs of others, rather than from the application of abstract universal rules. Core values associated with this perspective include care, responsibility, relationality, empathy, interdependence, and sensitivity to context. • Capability Approach: Developed by Nussbaum [32], this framework emphasizes enabling individuals to achieve fundamental capabilities, such as health, bodily integrity, autonomy, emotional development, and control over their environment and decisions. Ensuring these capabilities is considered an ethical obligation, as they constitute the minimum conditions required for a dignified human life. Key values associated with this perspective include human dignity, agency, freedom of choice, equality of capabilities, attention to diversity, and social justice. • Consequentialism: Consequentialist theories evaluate the moral rightness of actions primarily in terms of their outcomes [33, 34]. From this perspective, actions are assessed according to the extent to which they produce desirable consequences or minimize harm. Core values associated with this perspective include the moral primacy of consequences, the maximization of overall good, impartiality, aggregation of outcomes, forward-looking moral reasoning, and sensitivity to cost–benefit considerations.

Starting from the question of how robots can support care without undermining the ethical values that make care meaningful, the author argues that care should not be reduced to the automation of care-related tasks. Drawing on the ethics of care tradition developed by Gilligan [30] and Tronto [31], she argues that care is fundamentally a relational moral practice rather than a specific activity or set of tasks. Consequently, the ethical evaluation of care robots must consider not only their effectiveness and precision in performing care functions but also their impact on the relationships, responsibilities, and values that constitute caring practices. While Van Wynsberghe [27] focus primarily on normative guidelines for the ethical design of care robots, Vandemeulebroucke et al. [29] argue that ethical evaluations must also consider the socio-historical, organizational, and institutional contexts in which these technologies are implemented. From this perspective, care robots should be evaluated not only in terms of their technical capabilities but also in relation to the broader care systems and social conditions in which they operate. Building on this work, the present study proposes a theoretical framework that integrates multiple ethical perspectives and applies them to the analysis of empirical data, thereby responding to calls for approaches that bridge normative ethics and empirical research.

2.2.1 Ethical Foundations of the Analytical Framework The theoretical framework proposed by Van Wynsberghe [27] provides an important foundation for understanding the relational dimensions of care through the lens of care ethics. However, in developing the ethical framework guiding the analysis in the present study, we considered that relying solely on this perspective would be insufficient to capture the broader range of ethical concerns that participants might articulate. To address this limitation, we incorporated additional ethical theories to enable a more comprehensive analysis of the diverse values that shape perceptions of care robots. Following, we present a short description of the ethical theories that guide our framework:

The theoretical framework was developed by drawing on these ethical theories to address three important limitations in the existing literature. First, while technology acceptance models are effective in capturing perceptions of usefulness, attitudes, and adoption intentions, they provide only limited insight into the ethical reasoning that underlies these evaluations. Second, many normative approaches to care robotics rely on a single ethical perspective, thereby overlooking the diversity of values, principles, and concerns that stakeholders may express. Third, there is a need for an ethical framework that can be applied to

4

the analysis of empirical data, enabling a more systematic integration of normative ethical analysis with empirical research.

2.4 Cross-Cultural Differences in Care Robot Adoption and Implementation Several scholars argue that, despite the transformative impact of digital technologies, many people, particularly in countries with limited participation in technological development, remain excluded from the debates shaping digital societies, a phenomenon referred to as digital coloniality [15, 41]. Because digital societies reflect the social structures in which they are embedded, they often reproduce existing historical inequalities [42]. As digital innovation is concentrated in the Global North, the priorities, values, and assumptions embedded in digital technologies predominantly reflect the perspectives of these regions, reinforcing structural imbalances and marginalizing voices from elsewhere. One way to address this challenge is to design robots that can adapt to diverse socio-cultural contexts. Research in human–robot interaction has shown that robots developed within specific sociocultural settings often embed implicit assumptions about interpersonal distance, communication styles, and social roles [43–46]. When treated as universal interaction norms, these assumptions may conflict with the social practices of other cultural contexts [44, 47]. Moreover, Lawrence [47] argue that social norms in human–robot interaction should be understood not only at the societal level but also at the individual level, where personal expectations shape how robot behavior is interpreted and evaluated. Consistent with this view, Marchesi et al. [48] found that individual cultural values, rather than nationality, better predict the social inclusion of robots. In contrast, Li et al. [49] found that expectations regarding appropriate robot behavior vary across cultural contexts. In their cross-cultural experiment with participants from the United States and China, preferences differed with respect to robot autonomy: while participants in some contexts preferred more autonomous and proactive robots, others expected robots to behave more passively. Together, these findings highlight the importance of examining social robot acceptance across different cultural contexts, as acceptance is a key prerequisite for the successful deployment of robots and their contribution to stakeholder well-being.

2.3 Robot Type and Task Characteristics In the context of social robots, acceptance also depends on robot-related characteristics, including embodiment, social capabilities, perceived agency, and emotional expressiveness [35]. In addition, both the type of robot and the tasks it performs influence how it is perceived and accepted across contexts [36–39]. Consistent with this, Chatzoglou et al. [36] and de Graaf and Allouch [40] found that robot-related characteristics, such as physical attractiveness and task importance, exert a stronger influence on acceptance than sociocultural factors. Their findings further suggest that users evaluate robots according to the tasks they perform and the value these tasks provide in everyday life. Similar evidence has been reported in the healthcare robotics literature. In their review of healthcare robots for older adults, Broadbent et al. [11] identified the robot’s intended role and function as important determinants of acceptance. The authors argued that users may evaluate robots differently depending on the tasks they are designed to perform, ranging from health monitoring and reminders to physical assistance and social companionship. Although these conclusions were based on a literature review rather than direct empirical comparisons across robot types, the review provided early evidence that task characteristics influence robot acceptance and highlighted the need for systematic empirical research. Taken together, this body of research suggests that social robot acceptance is task-dependent. Users evaluate robots according to the functions they perform and the value they provide in a given context. Consequently, findings from one robotic application cannot necessarily be generalized to others, highlighting the need to examine multiple robot types performing different care-related tasks.

5

may contribute to the reduction of care to a set of technical tasks, potentially undermining the relational and emotional dimensions that are central to caregiving [55].

2.5 Healthcare and Sociocultural Contexts in the United States, Mexico, and Chile Healthcare environments across nations are shaped by a range of structural factors that influence how innovations are developed, adopted, and integrated [50]. Given that the acceptance and implementation of healthcare robots are likely to be influenced by these contextual conditions, this study focuses on three distinct national settings—the United States, Mexico, and Chile. The following outlines key characteristics of these healthcare environments, with particular attention to technology adoption and implementation.

2.5.2 Technology Adoption in Mexican and Chilean Healthcare Systems Mexico and Chile were grouped to represent a broader Latin American context characterized by comparatively lower levels of healthcare robot implementation, greater resource constraints, and sociocultural values surrounding care that may differ from those commonly described in the US. These countries also remain largely dependent on technological innovations developed in countries like the US [56, 57]. This situation is primarily driven by comparatively low levels of investment in technological development, limited scientific infrastructure, and a continued reliance on imported technologies [58]. Such asymmetries in technological production reinforce existing patterns of dependency. Within this structural context, inequality remains a defining feature of the region, with wealth and social resources concentrated among a small segment of the population [59]. These disparities contribute to stratified healthcare systems in which access to infrastructure, specialized personnel, and advanced medical technologies is unevenly distributed across social classes and geographic regions, systematically disadvantaging rural and marginalized communities. Moreover, institutional arrangements, financing mechanisms, and patterns of technological diffusion often reinforce these inequalities by favoring already well-resourced sectors [59]. Consequently, technological innovation can have ambivalent effects: while new medical technologies may improve healthcare access, efficiency, and reduce costs, they may also reproduce or exacerbate existing inequalities when not adapted to local socio-economic and institutional conditions [57]. Research examining caregivers’ perceptions of social robots in Latin America remains limited. One notable contribution in the region is the study by Churata et al. [60], which investigated factors shaping the acceptance of social robots among Brazilian caregivers involved in older-adult care. The findings indicate that caregivers generally perceived social robots as valuable tools for

2.5.1 Technology Adoption in the United States Healthcare System In recent decades, healthcare systems in the United States have been characterized by a strong orientation toward technological innovation [51]. Empirical evidence indicates sustained growth in the deployment of surgical robots and artificial intelligence–based systems across hospitals, supported by regulatory approvals and institutional adoption patterns that facilitate their incorporation [52]. By 2022, nearly one-fifth of U.S. hospitals had adopted some form of artificial intelligence [53, 54]. As a result, contemporary hospital environments increasingly involve the coexistence of caregivers, patients, and automated or semi-automated technological systems. This growing technological density not only reshapes clinical practices and organizational processes but also contributes to evolving expectations regarding efficiency, precision, and the role of technology in medical decision-making. However, despite the rapid implementation of advanced technologies in healthcare settings in the U.S., the adoption of AI remains uneven across institutions [53]. The implementation of robotic systems in healthcare and care settings has also reshaped how caregivers conceptualize work organization, care relationships, and patient safety [55]. While these technologies are often promoted as tools for improving efficiency, reducing workload, and optimizing care processes, their integration has also generated significant ethical and professional concerns. In particular, scholars have questioned whether an increasing reliance on technological solutions 6

supporting instrumental care tasks, particularly those related to safety monitoring and medication management. At the same time, participants expressed concerns regarding technical reliability, maintenance requirements, and data security. Interestingly, acceptance was more strongly associated with caregivers’ perceived workload than with prior caregiving experience or familiarity with social robots.

3.1 Participants A total of 298 participants from the United States (N = 152), Mexico, and Chile (N = 146 ), took part in this study. Participants were recruited through the Prolific platform using a purposive sampling strategy. Specific inclusion filters were applied to target individuals residing in the USA, Mexico, or Chile, who were fluent in English (in the case of the United States) or Spanish (in the case of Mexico and Chile), and who reported professional experience or affiliation within the healthcare sector. This sampling approach enabled the deliberate selection of participants relevant to the study’s objectives, ensuring that the sample reflected the sociocultural and professional characteristics necessary to examine variations in technology acceptance and ethical reasoning across sociocultural contexts. The study lasted approximately 27 minutes, and participants were compensated with £10.23.

2.5.3 Research Gaps and Questions The related literature provides the theoretical and empirical foundations for addressing the research questions that guide this study. In particular, it reveals two interrelated gaps. First, technology acceptance research has largely emphasized instrumental and functional determinants of adoption, while paying comparatively limited attention to how stakeholders interpret and evaluate the ethical implications of robotic design and implementation. Second, despite the growing body of research on care robotics, comparative studies examining caregivers’ perceptions of care robots across diverse sociocultural contexts remain limited. This is especially evident in comparisons involving stakeholders from highly technologized healthcare environments, such as the United States, and stakeholders from Latin American contexts, such as Mexico and Chile, which differ in terms of technological development.

3.2 Experimental Setup To address the research questions, we employed a mixed-methods design that integrated both quantitative and qualitative analytical approaches within a mixed factorial structure, consisting of one within-subjects factor and one betweensubjects factor. The within-subjects factor was robot type (Robots Delivering Supplies, Robots Helping Patients into Bed, Robots Monitoring Vital Signs, and Robots Assisting with Mobility), as all participants saw videos of each robot type and then evaluated each through questionnaires. The between-subjects factor was sociocultural context, operationalized through participants’ country of residence and language, and subsequently grouped into two broader groups: English group (EN) and Mexico/Chile group (SP). At the beginning of the study, all participants received a general introduction to the concept of care robots. They then received a standardized written description of each of the four categories of the health robots included in the experiment (see Table 1, Figure 2), followed by a short video clip illustrating the operational functionality of the respective robots. The selected robot types were chosen for two main reasons: (1) they represent distinct and sufficiently differentiated robot

• RQ1: To what extent do acceptance-related perceptions of four categories of healthcare robots, measured using UTAUT, CAN constructs, and an overall rating, vary across three countries with distinct socioeconomic and technological contexts (the United States, Mexico, and Chile)? • RQ2: What similarities and differences emerge in stakeholders’ understandings of the ethical design and deployment of care robots when analyzed through the proposed literature-based ethical framework, and how can these variations be interpreted in relation to ethical reasoning and values-based considerations?

3 Methods This section describes the study design, data collection procedures, and analytical approach.

7

(a) Robot Delivering Supplies

(b) Robot Helping Patients into (c) Robot Monitoring Vital (d) Robot Assisting with MobilBed Signs ity

Fig. 2: Representative examples of the four healthcare robot categories included in the study: (a) autonomous delivery robot (TUG) [61], (b) patient-assistive robot (Robear) designed to support patient transfer and lifting tasks [62], (c) monitoring robot (Florence) used for patient assessment and vital-sign monitoring [63], and (d) rehabilitation exoskeleton designed to support mobility and physical recovery [64].

functions, allowing for the examination of how variations in robot tasks and decision-making influence human acceptance; and (2) they involve different levels of potential risk and forms of human–robot interaction, thereby enabling an analysis of acceptance across a continuum of ethical and practical concerns. After viewing each video, participants completed a questionnaire consisting of quantitative measures based on the UTAUT and CAN scales, as well as an overall rating. In addition, after each video, participants responded to a set of openended questions designed to capture their ethical reflections. Figure 3 presents a schematic overview of the experimental procedure. Each participant evaluated all videos presented in randomized order, ensuring balanced exposure across conditions and enabling within-subject comparisons. Two parallel versions of the experimental materials were developed: one in English and one in Spanish. Content consistency across both versions was verified by two researchers, one a native Spanish speaker and the other a native English speaker. Each version was accessed through a separate study link.

Table 1 Description of the four robot categories used as experimental stimuli. Robot Type Description Delivering Mobile robots that transport medicaSupplies (DS) tions, sterile equipment, and laboratory samples, reducing nursing workload and improving logistical efficiency, especially under high-pressure conditions or isolation protocols. Helping Robots that assist caregivers in lifting Patients into and transferring patients with limited Bed (HPB) mobility, helping to prevent injuries and physical strain while improving safety and efficiency in patient-handling tasks. Monitoring Robots equipped with sensors to Vital Signs autonomously measure heart rate, blood pressure, respiratory rate, and oxygen saturation, improving accuracy and reducing human error in routine monitoring. Assisting with Exoskeleton-based or mobility-assistive Mobility robots that support rehabilitation by helping patients walk and stand, monitoring posture, balance, and gait, and reducing caregiver workload and risk of falls.

3.4 Survey Instrument and Procedure The questionnaire was composed of two sections designed to capture participants’ perceptions, evaluations, and reflections. The structure and rationale of each component are described below.

3.3 Measures For the analysis, two primary outcome variables were analyzed: 1) Acceptance of care robots, measured and compared across countries using constructs derived from UTAUT, CAN, and an overall rating; and 2) Ethical and normative considerations examined through a qualitative-theoretical analysis to identify dimensions of ethical reasoning and values-based judgments influencing acceptance.

3.4.1 Quantitative Measures: Selection and Adaptation of Scales The questionnaire combined constructs from the UTAUT model [18] and the CAN model [65]. UTAUT captures key determinants of technology acceptance, including Performance Expectancy, Attitudes Toward Using Technology, Self-Efficacy, and Behavioral Intention, whereas the CAN model complements these constructs by incorporating 8

healthcare robots and complemented the quantitative findings by providing richer insights into acceptance and ethical considerations. The following were the open-ended questions:

Welcome Study

Informed Consent Initial Scenario Test

• What are your main reasons for accepting or rejecting the use of this robot in caregiving tasks? You may refer to specific situations, feelings, or expectations you have. • Do you see any risks or ethical concerns in using this robot in healthcare? Please explain. • Based on your professional experience, what ethical issues do you think should be considered when developing or implementing this robot?

Robot 1 Description + Video UTAUT/CAN/Overall Rating Open-ended Questions Robot 2 Description + Video UTAUT/CAN/Overall Rating Open-ended Questions Robot 3 Description + Video UTAUT/CAN/Overall Rating Open-ended Questions

3.5 Analysis Strategy This subsection describes in detail the data analysis procedures employed in the study.

Robot 4 Description + Video UTAUT/CAN/Overall Rating Open-ended Questions

3.5.1 Quantitative Analysis Prior to conducting the statistical analyses, the internal consistency of all multi-item scales was assessed to ensure the reliability of the measurement instruments. Following established psychometric practice, composite scale scores were computed only after acceptable levels of internal consistency had been confirmed. Specifically, participants’ responses were aggregated across the items within each scale to obtain scale-level scores. Descriptive statistics (means and standard deviations) were then computed for each construct and robot type to provide an initial overview of participants’ evaluations and to inspect the distributional properties of the data. These descriptive results served as the basis for subsequent inferential analyses addressing the study’s research questions. To address RQ1 and examine differences in participants’ evaluations across robot types, repeated-measures analyses of variance (RM-ANOVA) were conducted, with robot type specified as a within-subject factor and language as a between-subject factor. Separate models were estimated for each dependent variable, including performance expectancy, attitude toward using the robot, behavioral intention to use, ethical judgment, and overall acceptance. Interaction effects between robot type and language were tested to assess whether evaluations of different robot types varied across cultural contexts. When omnibus effects were significant, post

End of Study Fig. 3: Schematic representation of the experimental procedure, including participant onboarding, scenario assessment, and repeated evaluations of the four care robots.

normative dimensions. Given the relational and ethically sensitive nature of care robots, integrating both models enabled a more comprehensive assessment of acceptance. We also included an overall rating. Because the original UTAUT items were developed for broader technology contexts, minor wording adaptations were introduced to improve ecological validity in healthcare settings. For example, references to ”technology” were replaced with ”the robot.” This adaptation approach is consistent with previous acceptance research [66, 67] and preserves the theoretical meaning of the original scales. Table 2 summarizes the rationale and items included for each construct.

3.4.2 Open-Ended Questions on Ethical Considerations The online questionnaire included a set of optional open-ended questions for each robot type presented. These questions explored stakeholders’ perspectives on the ethical implementation of 9

Table 2 Overview of the scales used in the user study. For each item, participants indicated their agreement on a 7-point Likert scale. Dimension

Rationale

Items used in the study

Performance Expectancy [68]

Assesses the perceived usefulness of care robots for supporting clinical work and improving job performance.

1. I would find the robot useful in my job.

2. Using the robot would enable me to accomplish tasks more quickly. 3. Using the robot would increase my productivity. 4. If I used the robot, I would have more chances of getting a raise. Attitude Toward Using Technology [68]

Assesses participants’ affective evaluations of interacting with care robots, including perceived enjoyment, interest, and satisfaction.

1. My interaction with the robot would be clear and understandable.

2. The system would make work more interesting. 3. Working with the system would be fun. 4. I would like to work with the robot. Self-Efficacy [68]

Assesses caregivers’ confidence in their ability to use care robots effectively under different support conditions.

I could complete a job or a task using the robot...

1. If there was no one around to tell me what to do as I go. 2. If I could call someone for help if I got stuck. 3. If I had a lot of time to complete the job for which the robot was provided. 4. If I only had the built-in help facility for assistance. Behavioral Intention to Use the System [68]

Assesses caregivers’ willingness to use care robots in their future clinical practice.

1. I would intend to use the robot in the next 12 months. 2. I predict I would use the robot in the next 24 months. 3. I would plan to use the robot in the next 24 months.

Ethical Judgments (CAN model) [19]

Assesses ethical evaluations of care robots, including fairness, morality, and cultural acceptability.

How would you rate the robot’s behavior in the video shown? 1. Unethical/Ethical; 2. Unfair/Fair; 3. Not morally right/Morally right; 4. Not acceptable to my family/Acceptable to my family; 5. Culturally unacceptable/Culturally acceptable; 6. Not personally satisfying/Personally satisfying; 7. Violates an unwritten contract/Does not violate an unwritten contract.

Overall Rating (own)

Helps evaluate the acceptance of human–robot collaboration.

1. Overall, I would rate the robot positively. 2. In general, I would support the integration of this robot in healthcare settings.

hoc pairwise comparisons were conducted using Bonferroni-adjusted tests to control for multiple comparisons. Effect sizes (η2 or partial η2) were reported alongside p-values.

of themes emerging from the data. To capture the diversity of ethical concerns expressed by participants, the framework integrated three complementary ethical perspectives: care ethics, the capability approach, and consequentialism (see ethical framework in Section 2.2.1). This combination enabled the analysis to remain grounded in established ethical theory while remaining sensitive to themes that emerged from the empirical data (see categories in Figure 4). The coding process was conducted collaboratively by the first author and a research assistant

3.5.2 Directed Qualitative Content Analysis The analysis followed a directed qualitative content analysis approach [69]. A literature-informed ethical framework, drawing on the work of [27], was developed to establish deductive analytical categories while allowing for the inductive refinement 10

1. Research question, theoretical background 2. Definition of the category system from theory (Ethics of Care, Capability Approach, Consequentialism) 3. Definition of the coding guideline, containing, for all categories: definitions, anchor examples and coding rules 4. Material run-through, preliminary codings. Complementary anchor examples, coding rules 5. Revision of the categories and coding guideline after 20% of the material, comparison coding between researchers

Fig. 4: Ethical framework for the evaluation of care robots. The framework combines concepts derived from Consequentialism, Care Ethics, and the Capabilities Approach with empirically derived concerns emerging from participants’ responses to openended questions. The circular structure reflects the interrelated and non-hierarchical nature of these perspectives in informing ethical evaluations of care robots.

6. Final coding and review of the material 7. Analysis, category frequencies and contingencies interpretation

Fig. 5: Steps involved in the qualitative content analysis.

using QCAmap software. In the first stage, both researchers independently coded 20 percent of the data and subsequently compared and discussed their coding decisions to establish a shared understanding of the coding scheme and refine it where necessary. In the second stage, the first author coded the remaining data and conducted the content analysis. All coding decisions, category assignments, and final interpretations remained the responsibility of the researchers. Through an iterative process of discussion and interpretation, the first and last authors reviewed, refined, named, and defined the themes. Any questions, ambiguities, or disagreements were discussed and resolved through joint meetings until consensus was reached. Open-ended responses were originally provided in English and Spanish. To standardize the dataset for qualitative analysis, all Spanishlanguage responses were translated into English using a single machine-translation system via an API-based scripted workflow in RStudio. Originallanguage responses were preserved in a separate column for auditability. To enhance translation trustworthiness, translated entries were reviewed by a bilingual researcher. Figure 5 illustrates the qualitative analysis process that was carried out.

JASP (version 0.19.3). Qualitative data were analyzed using the web-based application QCAmap, in combination with RStudio.

4.1 Descriptive Statistics Table 3 presents the descriptive statistics (means and standard deviations) for all study constructs by robot type and language group. All scales demonstrated acceptable to very good internal consistency across robot types and language versions (Cronbach’s α ranging from .80 to .96). Across constructs and robot types, mean scores were generally above the midpoint of the scale, meaning that, in general, robots were positively rated. Mean values ranged from 3.55 to 6.59, with the lowest scores observed for behavioral intention and the highest scores observed for ethical judgment and overall ratings. Descriptive statistics are reported for participants with complete data and correspond to the sample included in the repeated-measures analyses.

4.2 Quantitative Results To examine differences in participants’ evaluations across robot types and to assess whether these evaluations varied as a function of language group, a series of RM-ANOVA were conducted. Robot type (Delivery of Supplies, Helping Patients into Bed, Monitoring Vital Signs, and Assisting with Mobility) was specified as a within-subject factor, and country group (English vs. Spanish) as a betweensubject factor. Separate models were estimated for each dependent variable. Specifically, this analysis

4 Results In this section, we present the key findings derived from both the quantitative and qualitative components of the study. Quantitative analyses were conducted using RStudio (version 2026.04.0) and

11

Table 3 Descriptive statistics (Means and Standard Deviations) by robot type and language. Construct

Robot Type

EN M

EN SD

EN α

SP M

SP SD

SP α

Performance Expectancy

DS BED MVS AM

5.038 4.678 5.003 4.831

1.558 1.661 1.553 1.655

.87 .90 .87 .89

5.389 5.152 5.445 5.277

1.192 1.481 1.212 1.368

.87 .89 .87 .89

Attitudes Toward Using Robot

DS BED MVS AM

5.411 5.115 5.360 5.457

1.339 1.559 1.506 1.361

.89 .91 .90 .89

5.877 5.803 5.993 6.045

1.107 1.218 1.084 1.025

.87 .87 .86 .87

Self-Efficacy

DS BED MVS AM

5.337 5.133 5.288 5.023

1.249 1.364 1.356 1.320

.84 .85 .84 .84

5.267 5.459 5.491 5.262

1.221 1.222 1.188 1.311

.80 .85 .84 .84

Behavioral Intention

DS BED MVS AM

4.851 4.349 4.524 4.336

1.912 2.010 2.058 2.007

.96 .96 .95 .95

3.667 3.744 3.779 3.546

2.015 2.040 1.969 2.016

.95 .95 .92 .93

Ethical Judgment (CAN)

DS BED MVS AM

6.109 5.816 5.847 6.160

1.103 1.346 1.272 1.067

.95 .95 .95 .96

6.210 6.147 6.070 6.450

1.002 0.969 1.200 0.814

.91 .86 .91 .88

Overall Rating

DS BED MVS AM

5.842 5.461 5.734 6.069

1.464 1.698 1.467 1.270

.88 .88 .88 .88

6.209 6.103 6.092 6.586

1.210 1.390 1.354 0.862

.89 .89 .89 .89

Note. Values are scale means; EN = English, SP = Spanish; α = Cronbach’s alpha.

Table 4 Repeated-measures ANOVA results across dependent variables. F

dfn , dfd

p

η2

Robot Type Language Robot Type × Language

5.714 10.297 0.192

3, 888 1, 296 3, 888

< .001 .001 .902

.007 .021 2.445 × 10−4

Attitudes Toward Use

Robot Type Language Robot Type × Language

5.159 23.123 0.671

3 1 3

.002 < .001 .570

.006 .049 7.339 × 10−4

Behavioral Intention

Robot Type Language Robot Type × Language

4.009 17.306 3.306∗

3 1 3

.008 < .001 .020∗

.003 .041 .003∗

Self-Efficacy

Robot Type Language Robot Type × Language

4.339 2.065 3.005∗

3, 888 1, 296 3, 888

.005 .152 .030∗

.005 .005 .003∗

Ethical Judgment (CAN)

Robot Type Language Robot Type × Language

15.367 4.910 1.465

3 1 3

< .001 .027 .222

.016 .011 .002

Variable

Effect

Performance Expectancy

Note. RM-ANOVA = repeated-measures analysis of variance. Effect sizes are reported as η 2 . Greenhouse–Geisser corrections were applied where the assumption of sphericity was violated (Performance Expectancy, Self-Efficacy, and Ethical Judgment). ∗ Indicates statistically significant interaction effects between Robot Type and Language.

aims to assess acceptance-related perceptions of four categories of care robots. The RM-ANOVA results (see Table 4) indicate that robot type had a consistent and statistically

significant effect across all dependent variables, suggesting that participants’ evaluations systematically varied according to the type of care robot [36, 37]. Language also showed significant main

12

effects for most outcomes, with the exception of self-efficacy, indicating overall differences between language groups in the perceived acceptance and evaluation of care robots. Overall, both language groups tended to evaluate the different robot types in relatively similar ways, as reflected by the largely non-significant interaction effects between Robot Type and Language. These findings point to a general crosscultural consistency in perceptions of healthcare robots, suggesting that participants from different cultural contexts shared comparable evaluations of the various robotic applications. Moreover, although several effects reached statistical significance, their small effect sizes indicate that the practical magnitude of these differences was limited. An exception to this pattern emerged for behavioral intention and self-efficacy, where significant interaction effects between robot type and language were observed. These findings suggest that, while attitudes toward care robots were generally positive across groups, participants differed in their willingness to use these technologies and in their perceived ability to do so. Specifically, participants in the English-language group reported higher behavioral intentions to use care robots than participants in the Spanish-language group. Given the composition of the sample, these differences may reflect broader distinctions between the U.S. context and the Latin American contexts included in the study. Taken together, the results suggest that sociocultural context may play a more prominent role in shaping anticipated adoption and perceived competence in using care robots than in influencing broader attitudinal evaluations. Bonferroni-adjusted post hoc comparisons (see Table 5) revealed that, in the performance expectancy dimension, robots helping patients into bed (BED) received less favorable evaluations than robots delivering supplies (DS) and monitoring vital signs (MVS). In the ethical judgment dimension (CAN model), assisting with mobility (AM) robots received more favorable evaluations than BED and MVS robots, while DS robots were also evaluated more positively than BED and MVS robots. Although descriptive trends (see Table 3) suggested lower evaluations for BED robots in attitudes toward using robots, Bonferroni-adjusted pairwise comparisons did not reveal consistent significant differences across all robot categories.

Table 5 Significant Bonferroni-adjusted post hoc comparisons. Outcome

Comparison

MD

pbonf

Performance Expectancy

DS > BED

0.298

.004

BED < MVS

−0.309

.002

Ethical Judgment (CAN)

DS > BED

0.178

.029

DS > MVS DS < AM BED < AM MVS < AM

0.201 −0.145 −0.323 −0.346

.004 .041 < .001 < .001

Note. Only statistically significant Bonferroni-adjusted pairwise comparisons are presented. Mean differences retain the original comparison direction reported in the post hoc analyses. Symbols indicate the direction of comparatively more favorable evaluations between robot categories. DS = Delivering Supplies; BED = Helping Patients into Bed; MVS = Monitoring Vital Signs; AM = Assisting with Mobility.

Overall, these comparisons suggest that healthcare workers differentiated between forms of physical assistance, evaluating robots involved in direct patient handling less favorably than robots performing logistical or monitoring functions. Likewise, this distinction suggests that not all physically assistive robots are perceived equally, and that the acceptability of robotic care may depend on the specific nature of the caregiving task being performed. These findings support prior research suggesting that acceptance of care robots is taskdependent and shaped by the specific caregiving functions performed by the technology [36–39].

4.3 Qualitative Results In the following section, we present the most salient qualitative findings of the study. The analysis is organized into three parts. First, we examine participants’ ethical perceptions of care robots at a general level. Second, we explore how ethical evaluations varied according to the specific characteristics, functions, and tasks performed by the different robot categories. Finally, we analyze crosscultural similarities and differences in participants’ ethical assessments, highlighting how sociocultural contexts shaped the interpretation, prioritization, and negotiation of ethical concerns surrounding care robot implementation.

4.3.1 Ethical Themes in Participants’ Responses Table 6 provides an overview of the ethical considerations identified in participants’ responses 13

regarding the acceptance and potential rejection of care robot technologies. To analyze these responses, we draw on the categories derived from the theoretical framework described in section 3.5.2. Table 6 therefore identifies both the theoretical origins of each category and those that emerged inductively during the coding process. Importantly, the interpretation of these categories does not rely on a simple distinction between positive and negative evaluations. Participants frequently invoked the same ethical considerations to both support and criticize the use of care robots, highlighting the complex and often ambivalent nature of ethical reasoning surrounding these technologies. Overall, the theoretically derived categories captured a substantial proportion of participants’ responses, suggesting that the proposed framework provides a useful lens for understanding stakeholders’ ethical evaluations of care robots. In the following sections, we elaborate on how these ethical considerations emerged across the different analytical categories.

regulations designed to protect the privacy, security, and integrity of Protected Health Information. Although participants from Mexico and Chile also raised concerns regarding data privacy and confidentiality, they generally did not connect these concerns to specific regulatory frameworks or legal standards. Additional contextual differences were observed in the way participants framed ethical concerns surrounding robotic systems. Responses from the US more frequently emphasized issues related to harm reduction, safety, accountability, and responsibility for foreseeable consequences, suggesting a stronger focus on efficiency, governance, and regulation. In contrast, participants from Mexico and Chile more commonly referred to contextual adaptation, capability enhancement or deprivation, and the relational dimensions of care, emphasizing the importance of understanding the specific social and emotional realities in which robots are implemented. One particularly salient finding concerned the use of robots in psychiatric and pediatric care settings. Across both cultural contexts, participants frequently expressed skepticism about implementing robots in these environments. Many emphasized that patients in such settings may have difficulty understanding what robots are or how they operate, potentially resulting in fear, emotional distress, mistrust, or even accidental harm.1 Overall, the findings suggest that ethical perceptions regarding care robots are shaped not only by universal concerns surrounding healthcare technologies, but also by the social, economic, and regulatory contexts in which these technologies are interpreted and evaluated.

4.3.2 Cross-Cultural Comparison of Ethical Evaluations Based on the qualitative findings, some sociocultural differences emerged in participants’ ethical perceptions regarding the use of care robots. However, these differences were relatively limited, suggesting that participants across cultural contexts generally identified similar ethical benefits and risks associated with robotic care technologies. Nevertheless, some interpretative differences became evident across specific thematic categories. For instance, among participants from Mexico and Chile, the concept of impartial evaluation was predominantly associated with the economic capacity to acquire or access robotic systems, reflecting concerns related to inequality and accessibility in healthcare technologies. In contrast, participants from the US tended to associate impartial evaluation with concerns regarding algorithmic bias, discriminatory outcomes, and biases embedded within datasets used to train robotic systems. Similarly, when discussing data privacy, participants from the US consistently referred to compliance with the Health Insurance Portability and Accountability Act of 1996 (HIPAA), emphasizing the importance of adhering to federal

4.3.3 Ethical Evaluations by Robot Type Regarding the comparison between different robotic systems, certain ethical categories were more frequently associated with specific types of robots, suggesting that participants did not evaluate robotic technologies as homogeneous systems, but rather interpreted their ethical implications 1 This interpretation is further supported by an exploratory part of the study in which participants were asked to identify the healthcare settings in which care robots would be most useful. Across both cultural contexts, psychiatric settings were consistently evaluated less favorably than other medical environments (see Figure 6 in the supplementary material).

14

15

Ethical arguments emphasizing that care robots are perceived as desirable when they enhance efficiency, improve well-being, and reduce workload, and as undesirable when they fail to contribute meaningfully to these aspects.

Welfare maximization (Consequentialism)

Analytical interpretation

Positive/Neutral quote

Critical/Neutral quote

This category was among the most fre“It would free up a person to work “Although the robot helps me to quently invoked across both cultural on something else”. (Robots Deliver- make the movement and to load the contexts. Participants commonly framed ing Supplies) patient, it still has to be watched at robots as tools for improving efficiency and “I consider that this type of robot is the time of performing the action, so reducing workload by streamlining routine extremely necessary and beneficial I can not be so productive because I tasks. However, they also questioned whether in the health area. It could help a have to be watching that the movethese anticipated benefits would be realized lot of patients. In addition, it would ment is done correctly and that it in practice or whether robot adoption might also benefit caregivers who usually does it a little slow, so I could lose instead introduce new forms of workplace dis- rely on their own physical strength time if it is done in a place with ruption. for these tasks.” Robots Assisting many patients.” (Robot Helping with Mobility). Patients into Bed) “I feel that taking vital signs myself is faster than using the robot. It would also only be usable with patients who can actively participate in the process.” (Robots Taking Vital Signs).

Impartial evaluation (Consequentialism)

Ethical arguments emphasizing fairness, equal consideration, and balanced evaluation across stakeholders.

Note. Categories identified inductively from the data are indicated with ∗ .

This category captured responses in which “I don’t see how the robot can pose “Equity and accessibility should be participants evaluated care robots in terms any ethical problems as long as it addressed. The technology should not of their consequences for multiple stakehold- treats each person equally.” (Robots only be available to well-funded hosers. Participants expressed concerns about Delivering Supplies) pitals but also accessible to smaller distributive justice, questioning whether “I think as long as the robot facilities and diverse patient popularobotic technologies would reinforce or remains neutral and it doesn’t tions.” (Robots Helping Patients into reduce existing inequalities. In particular, assess the patient based on anything Bed). they highlighted unequal access to robot that would be considered as being “Fairness. Developers must ensure technologies. Participants also raised conprejudiced or biased, there would be that the robot does not reinforce cerns about algorithmic bias, emphasizing no problems with ethics” (Robots biases or discriminate against any that AI systems trained on biased data may Helping Patients into Bed). group of patients.” (Robots Deliverproduce discriminatory outcomes. ing Supplies).

Harm reduction Ethical considerations Safety, alongside welfare maximization, “I believe it would facilitate the “I would only be afraid of it hurting, and safety (Con- emphasizing that care emerged as one of the most salient moral transfer of overweight patients and dropping someone, or malfunctioning.” sequentialism) robots are morally jus- concerns. Participants viewed care robots prevent accidents/injuries to nurses, (Robots Helping Patients into Bed) tified insofar as they positively when they were perceived as reduc- orderlies and physiotherapists.” “Robots need thorough testing to reduce risk, prevent ing errors, physical strain, burnout, or harm (Robots Assisting with Mobility). prevent malfunctions that might harm, or minimize to patients. Conversely, robots were also seen “I think it is a wonderful advancedelay or misdeliver important items negative outcomes. as potential sources of accidents, malfuncment in technology. It is incredibly like medications or lab samples .” Conversely, they are tions, or loss of control, particularly during hard to support the weight of a per- (Robots Delivering Supplies) unjustified if they physically intimate care tasks. Participants son and assist in mobility and it compromise individfurther stressed that safety depends not only is also riskier for injuries to both uals’ well-being by on technical performance but also on careclient and caregiver without a causing accidents or givers’ responsible use of these systems. device like this. (Robots Assisting harm. with Mobility).

Definition

Category

Table 6 Overview of ethical categories, analytical interpretations, and ambivalent evaluations across participants’ responses.

16

Ethical arguments emphasizing human relationships, emotional connection, interpersonal presence, and appropriate human oversight.

Relational care and human oversight (Care Ethics)

Capability enhancement and deprivation (Capability Approach)

Ethical arguments addressing the impact of robots on autonomy, agency, independence, dependence, and opportunities for meaningful participation.

Robots literacy Ethical considerations emphasizing education, training, and understanding of robotic systems.

Ethical arguments emphasizing anticipation of future risks, unintended effects, and responsibility for foreseeable outcomes.

Responsibility for foreseeable consequences (Care Ethics)

∗

Definition

Category

Analytical interpretation

Positive/Neutral quote

Critical/Neutral quote

“It seems to me that any tool that “None, they will only make trainees is capable of providing a patient and nurses ”lazier”, since in the with a greater degree of autonfuture they may stop teaching how omy is a tool that should not be to take vital signs.” (Robots Taking overlooked and should be considVital Signs) ered when evaluating the quality “Over-reliance on automation: of medical care and quality of life Employees may grow unduly reliant of the patient.” (Robots Assisting on the robot and fail to notice critical with Mobility) “I am in favor of its clinical cues that call for human judguse to support mobility therapies, ment.” (Robots Taking Vital Signs) help people to walk again and give them a more dignified life.” (Robots Assisting with Mobility)

Note. Categories identified inductively from the data are indicated with ∗ .

This category captured the ambivalence surrounding robots as both capability-enhancing and capability-restricting technologies. Some participants viewed robots as tools that could expand patient autonomy, mobility, independence, and participation in healthcare. Others, however, expressed concern that excessive reliance on robotic systems could reduce healthcare workers’ active involvement, weaken professional expertise, or limit human decision-making.

Participants emphasized that the successful “As long as a person is trained, “I can see where some patients may implementation of care robots depends on there should not be risks”. Robots be fearful of the robot-assisted med providing patients, caregivers, and healthcare Assisting with Mobility “Just mak- cart moving. Educate the patients stakeholders with clear guidance on how the ing sure the patient has a good that this is what is going to be haprobot operates and what it is designed to do. understanding of how the robot will pening.” Robots Delivering Supplies Such explanations were considered essential operate in assisting staff with their “Risk of lack of staff training.” Robots for reducing uncertainty, facilitating approcare.” Robots Helping Patients into Assisting with Mobility priate interaction, and establishing realistic Bed expectations. Conversely, insufficient understanding of robotic systems was associated with confusion, misuse, and unrealistic expectations about the robot’s capabilities.

Participants frequently emphasized that “Human oversight should always “Over-reliance on automated monrobots should support rather than replace be maintained, ensuring staff can itoring may reduce direct human human caregivers. Responses reflected a ten- intervene if needed” Robots Helping interaction, which is important for sion between the efficiency of robotic systems Patients into Bed. patient comfort, trust, and emotional and the ethical importance of empathy, emo- “The robot should be designed to support.” Robots Monitoring Vital tional warmth, and human presence. While support patient dignity, motivation, Signs some viewed robots as valuable complements and emotional well-being without “Patients may feel a loss of personal to care, they stressed that meaningful human replacing essential human interacdignity or autonomy if they are oversight must remain central and that tion.” Robots Helping with Mobility moved by a machine instead of a responsibility for decisions affecting patients’ human caregiver.” (Robots Helping health should always rest with caregivers. Patients into Bed)

Participants used this category to express “Hospitals need systems in place “Accountability matters if a robot a future-oriented ethical stance centered on to track deliveries accurately and malfunctions during a transfer who is preventing foreseeable risks associated with assign responsibility if errors occur, responsible? The nurse?The hospital care robots. They emphasized that hospisuch as delivering the wrong medica- questionmark? the manufacturer tals, designers, and healthcare institutions tion.” (Robots Delivering Supplies) questionmark? ethical use demands should establish clear protocols and response “Accountability: Is the manufaccrystal clear accountability and failprocedures before implementation to mititurer, the hospital, the managing safes because one error could mean gate potential harms. Participants also raised therapist, or the robot’s programa serious injury” (Robots Helping concerns about accountability in complex mer legally and morally liable if a Patients in Bed) socio-technical systems, particularly when system malfunctions and a child is “What if the robot supplies the wrong responsibility for errors or adverse outcomes hurt? Prior to implementation, it is drugs to the wrong room. Who takes may become diffuse or unclear. necessary to establish clear lines of the blame?” (Robots Delivering accountability.” (Robots Assisting Supplies) with Mobility)

17

Definition

Ethical considerations related to sensitive information, data protection, cybersecurity, and consent to interact with robots.

∗

Analytical interpretation

Positive/Neutral quote

Critical/Neutral quote

Note. Categories identified inductively from the data are indicated with ∗ .

“As a matter of ethics, people losing their jobs as they are replaced by robots or AIs.” (Robots Delivering Supplies) “This model can do many things, and I would be a bit concerned about job losses.” (Robots Monitoring Vital Signs)

This category was among the least frequently None mentioned, and all responses expressed negative views. Participants raised concerns about job displacement and the impact of robotic technologies on the healthcare workforce. Consistent with the category of relational care and human oversight, many emphasized that robots should support rather than replace caregivers, warning that economic and efficiency-driven incentives could ultimately favor robotic over human care.

Ethical concerns related to replacement of human labor and socio-economic consequences of automation.

∗

Job loss

Participants rejected one-size-fits-all “It is only a robot that assists in “Care should be adapted to each approaches to robotic implementation, mobilization and rehabilitation, so person, which may be difficult for emphasizing that the appropriateness of there are not really many ethical standardized technology.” (Robots robots depends on patients, tasks, and care issues to comment on, only that Delivering Supplies) contexts. They expressed particular concern the robot is well adapted to each “I believe that in specific cases, such about psychiatric and pediatric settings, person’s mobility needs.” (Robots as pediatric, geriatric or psychiatric where patients may not fully understand Assisting with Mobility) patients, it could become a problem, or feel comfortable interacting with robotic “When implementing this in hosespecially for this public to undersystems, increasing the risk of distress. Par- pitals, they should test different stand the way in which they are used.” ticipants also stressed that robots should populations within a hospital. That (Robots Monitoring Vital Signs) respect patients’ wishes and preferences and way they could see what type of be adapted to the architectural and spatial population receives this type of conditions of hospitals. robot more appropriately and, on the other hand, be able to adapt different types of robots for different populations.” (Robots Monitoring Vital Signs)

Participants associated robotic implemen“It can help to have a more accu“Data protection and confidentialtation with risks related to data collection, rate record of patient data without ity are major concerns. These robots storage, surveillance, and institutional trust. having to annotate, and reduces the often rely on sensors, cameras, or AI While some viewed the storage of personal risk of human error.” (Robots Moni- systems that record sensitive health data as enhancing safety and security, others toring Vital Signs) information, so developers must perceived it as increasing surveillance and “None at all. I fully recommend ensure strict data encryption and reducing privacy. Participants emphasized these robots. Of course, we would compliance with healthcare privacy that acceptable implementation depends on need to obtain the patient’s consent regulations.” Robots Helping Patients informed consent and individuals’ freedom to or the family member of the patient into Bed decide whether to use robotic systems. Par- if the patient has cognitive impair- “Data privacy and security is also ticular concerns were raised about obtaining ments such as dementia/Alzheimer.” important, as these robots often informed consent from psychiatric patients (Robots Helping Patients in Bed) operate through centralized systems and individuals with mental illnesses. that track deliveries and item details.” (Robots Delivering Supplies)

Contextualization Ethical arguments (Capability emphasizing adapApproach) tation to individual needs, abilities, circumstances, and care contexts.

Data privacy, security, and informed consent (Capability Approach)

Category

according to the specific forms of care, task, interaction, and vulnerability involved in each context. Overall, participants appeared to differentiate between robotic systems that support bodily assistance and patient autonomy, and those that mediate clinical judgment, monitoring, or interpersonal interaction. For example, the category ”capability enhancement and deprivation” was particularly prominent in discussions concerning robots assisting with mobility and robots monitoring vital signs; however, the connotations associated with this category differed substantially between the two systems. In the case of robots assisting with mobility, participants generally used this category positively, frequently describing the robot as a tool capable of supporting patient autonomy, improving mobility, and facilitating rehabilitation, particularly in situations involving physical dependence or reduced movement capacity. In contrast, discussions surrounding robots monitoring vital signs reflected a predominantly negative interpretation of the same category. Participants frequently associated these systems with the potential loss or deterioration of caregivers’ clinical skills, particularly regarding the manual assessment and monitoring of vital signs. Rather than viewing automation as enhancing, participants often perceived these systems as potentially replacing human competencies and reducing direct clinical engagement. These findings suggest that participants distinguished between forms of automation perceived as augmenting human care and those perceived as substituting essential human skills and interactions. Some illustrative quotes related to these findings can be found in Table 6. Notably, robots assisting with mobility and those helping patients into bed were more frequently associated with promoting dignified care. Participants often emphasized that assistance with mobility and bodily support could enhance patients’ dignity by reducing situations in which they feel exposed, dependent, or physically vulnerable in the presence of others. Interestingly, these findings suggest that participants did not necessarily associate robotic intervention with dehumanization; rather, in certain physically sensitive situations, robotic assistance was perceived as potentially less invasive and more dignitypreserving than human intervention alone.

At the same time, robots assisting patients into bed were also strongly associated with the category ”harm reduction and safety”. Participants frequently expressed concerns regarding the possibility of technical malfunction, mechanical errors, or physical accidents that could directly harm vulnerable patients during bodily transfers. These findings reveal an important ethical tension in participants’ perceptions: while robotic systems were recognized as capable of preserving dignity and reducing physical dependency, they were simultaneously perceived as introducing new forms of physical risk in highly vulnerable care situations. Similarly, robots monitoring vital signs and delivering supplies were more frequently associated with the category of relational care and human oversight. Participants emphasized that activities such as monitoring vital signs are not merely technical procedures but also important opportunities for interpersonal connection, emotional support, and human presence. Although these robots were generally viewed positively, participants consistently stressed that their use should remain subject to human and clinical oversight, expressing greater concern about fully autonomous care systems lacking human validation and accountability. Overall, the qualitative findings regarding the robot types indicate that participants’ ethical evaluations of care robots were strongly shaped by the type of human relationship affected by the technology. Robotic systems were generally perceived more positively when they supported patient autonomy, reduced physical vulnerability, or preserved dignity, whereas greater skepticism emerged when robots were perceived as replacing human interaction, reducing relational care, or eroding clinical competencies.

5 Discussion The goal of this study was to investigate caregivers’ acceptance of four categories of care robots across the United States, Mexico, and Chile; and second, to evaluate a literature-informed ethical framework for analyzing stakeholders’ ethical perceptions of these technologies. Three key findings emerged from our analyses. First, participants generally expressed more favorable attitudes toward robots performing logistical and physically demanding tasks than toward robots operating in contexts requiring close interpersonal interaction. Second, 18

although participants from Mexico and Chile reported positive evaluations across most acceptance dimensions, they expressed lower behavioral intentions to use these technologies, suggesting greater uncertainty regarding their practical implementation in local healthcare systems. Finally, the qualitative findings revealed both shared and context-specific ethical concerns related to the design, deployment, and governance of care robots, highlighting how sociocultural values and robot type shape their acceptance. For the discussion, we focus on integrating the quantitative and qualitative findings to provide a comprehensive understanding of participants’ perceptions, evaluations, and ethical reflections regarding care robots across different sociocultural contexts, as well as to discuss some of the most latent ethical.

tasks [35]. This interpretation was reinforced by the qualitative findings, in which mobility-assistance robots were frequently associated with positive ethical outcomes, particularly in relation with increasing autonomy and enhancing quality of life (capability enhancement and deprivation category), whereas robots monitoring vital signs were more often linked to negative ethical evaluations. Together, these findings suggest that ethical evaluations of care robots depend not only on the presence of physical interaction but also on how agency, control, vulnerability, and assistance are embodied within the caregiving relationship. Technologies perceived as supporting patients’ autonomy, rather than replacing caregivers, appear to be viewed as more ethically acceptable [38]. Similarly, the quantitative analyses revealed generally limited interaction effects between robot type and language, although behavioral intention varied across language groups for some robot categories [47–49]. Participants from Mexico and Chile consistently reported lower behavioral intention scores than those from the United States. One possible explanation for this pattern emerges from the qualitative findings, as participants from Mexico and Chile frequently referred (coded under the category of contextualization) to the high costs of care robots and unequal access to technological resources. These concerns may reflect the challenges faced by countries that rely primarily on imported healthcare technologies, where acquisition, maintenance, and implementation costs constitute substantial barriers to adoption, particularly for healthcare institutions with limited financial and infrastructural resources [56–59, 71].

5.1 Task-Dependent Acceptance: Relational and Contextual Dimensions Some of the qualitative data provided complementary insights that helped explain participants’ quantitative evaluations of care robots. Consistent with previous research [11, 35–37, 40], acceptance varied across the four robot categories, although the overall evaluation patterns remained broadly consistent across cultural contexts. One of the clearest findings was that robots designed to help patients into bed received less favorable evaluations than the other robot types. Qualitative responses suggested that, although participants frequently associated these robots with greater patient autonomy, enhanced dignity, and reduced dependence on caregivers for mobility-related or potentially embarrassing situations, they also perceived them as particularly high-risk technologies. Concerns about accidents, physical harm, loss of control, and technological malfunction appeared to outweigh sometimes their perceived benefits [70]. Another prominent quantitative finding was that, within the ethical judgment dimension (CAN model), robots assisting with mobility received more favorable evaluations than robots helping patients into bed or monitoring vital signs. Although all three involve direct patient interaction, mobility-assistance robots are typically perceived as wearable technologies that support, rather than independently perform, caregiving

5.2 Beyond Instrumental Acceptance: Moral Tensions in Care Robotics The qualitative analysis, guided by the theoretical framework proposed in this study, showed that users’ ethical perceptions of care robots are shaped by both the sociocultural context and the robots’ characteristics and functions. Rather than being stable or purely instrumental, ethical evaluations emerged as relational and context-dependent negotiations between competing moral values [29]. Participants frequently evaluated the same robotic systems from multiple, and sometimes conflicting, ethical perspectives, highlighting the moral 19

complexity of technologies that mediate practices of care [38]. What some participants perceived as ethically beneficial, others interpreted as ethically problematic or potentially harmful. Next, we discuss some of the most important ethical tensions.

and operational risks, they were also seen as introducing technological vulnerabilities that require continuous human oversight, clear operational protocols, and well-defined accountability structures. Accordingly, concerns about harm reduction and safety were closely intertwined with the categories of responsibility for foreseeable consequences and relational care and human oversight [27].

5.2.1 Welfare Maximization Within the category of welfare maximization, participants acknowledged that robots designed to help patients into bed or monitor vital signs could reduce the physical burden associated with patient handling and continuous supervision, thereby promoting overall well-being. However, they also emphasized that introducing these technologies into clinical workflows may generate new cognitive, technical, and organizational demands. In particular, caregivers would need to learn not only how to operate the robots but also how to coordinate their use within everyday care practices. In this regard, this category was linked to Robots Literacy. Consequently, the anticipated gains in efficiency were not perceived as automatic, as the additional workload associated with implementation could partially offset the robots’ practical benefits [22, 72]. These findings illustrate how values commonly associated with technological innovation, such as efficiency and welfare maximization, become ethically and practically contingent when embedded within the organizational and relational realities of care work [27, 73].

5.2.3 Capability Enhancement and Deprivation The category Capability enhancement and deprivation revealed important tensions in how participants understood autonomy in relation to the use of care robots. On the one hand, some robotic technologies were frequently associated with increased independence, improved mobility, and enhanced support for both patients and caregivers. On the other hand, participants also expressed concerns that excessive reliance on robotic systems could foster new forms of technological dependency that might ultimately undermine autonomy itself [39, 74]. Importantly, participants suggested that the increasing delegation of caregiving tasks to robots could contribute not only to the erosion of practical and interpersonal caregiving competencies [75], but also to broader forms of social isolation, loneliness, and the weakening of human interaction within caregiving environments [55, 76, 77]. In this sense, autonomy was not understood as a stable condition, but rather as something relationally constructed and potentially transformed through human–robot interactions. This category was consistently connected to relational care and human oversight, suggesting that participants perceived the ethical integration of care robots as dependent not only on technological functionality, but also on the preservation of meaningful human relationships and on users’ ability to critically understand, supervise, and appropriately engage with these systems.

5.2.2 Harm Reduction and Safety The category of harm reduction and safety revealed important ethical tensions regarding the role of automation in caregiving environments. Participants frequently emphasized that care robots could reduce risks and physical strain for healthcare personnel, for example by minimizing medication delivery errors when robots deliver supplies or by preventing back injuries and physical exhaustion when robots help patients into bed. Nevertheless, participants repeatedly expressed concerns regarding the possibility of mechanical failures, malfunctions, or loss of control, which could ultimately generate even greater forms of harm [28]. These findings suggest that participants viewed robotic technologies as shifting, rather than eliminating, risks within the caregiving process. While robots were perceived as reducing certain physical

5.3 Implications for Care Robot Design and Implementation Based on the findings of the present study, several practical implications can be derived for the design and implementation of care robots.

20

First, although participants across the examined cultural contexts generally expressed positive attitudes toward care robots, subtle differences in perceptions and concerns may influence their acceptance in diverse sociocultural settings. These differences were particularly evident for robots expected to operate in situations involving close interpersonal interaction and physical proximity, such as those monitoring patients’ vital signs or assisting patients into bed. The findings therefore highlight the importance of moving beyond purely technical considerations and adopting a user-centered, culturally sensitive design approach that carefully considers how a robot’s appearance, behavior, communication style, and decision-making processes shape users’ trust, comfort, and sense of control. Second, the findings highlight the importance of considering the socioeconomic contexts in which care robots are expected to be deployed. While research on care robotics often emphasizes technical performance and usability, the present study suggests that economic accessibility is also a key determinant of public acceptance. Developers should therefore design care robots with affordability and accessibility in mind, adapting both the technology and its implementation to the realities of different healthcare systems. Importantly, affordability should be regarded not only as a market consideration but also as an ethical design principle, as technologies developed primarily for high-income healthcare systems may exacerbate existing inequalities in access to care. Third, the findings raise important concerns regarding the use of care robots in mental health and psychiatric settings. Although robotic technologies are increasingly being developed to support individuals with mental health conditions [78– 80],participants feared that robots could increase stress, anxiety, confusion, or emotional distress among vulnerable patients. These findings suggest that deploying care robots in mental health settings should be approached with caution, supported by rigorous risk–benefit assessments and continuous evaluation to ensure patient well-being, particularly for populations that may be disproportionately affected by technological failures or inappropriate interactions [81]. Finally, the findings of this study suggest that the ethical and practical factors influencing the acceptance of care robots should not be understood as inherently positive or negative. Rather,

participants frequently perceived the same robot system as both beneficial and potentially problematic, revealing a fundamental ambivalence in how these technologies are evaluated. This highlights the importance of recognizing and addressing such ambivalence throughout the design process. Instead of focusing exclusively on maximizing functionality or usability, developers should adopt a more holistic and human-centered approach that considers the complex and sometimes conflicting expectations users hold toward care robots.

6 Limitations Several limitations of this study should be acknowledged. First, although statistically significant differences were identified, most quantitative effects were small in magnitude, suggesting that the findings should be interpreted as general tendencies rather than strong predictive relationships. Second, the study focused on participants from the United States, Mexico, and Chile; therefore, the findings cannot be generalized to all sociocultural or healthcare contexts. Additionally, participants evaluated hypothetical care robot scenarios rather than interacting directly with robotic systems in real clinical settings, which may differ from actual experiences of use and implementation. We also only evaluated four types of care robots, however, the interpretation might change with those types. Finally, while the qualitative analysis provided important interpretive insights, the proposed ethical categories remain theoretically situated and interpretive in nature, meaning that alternative analytical perspectives and frameworks could lead to different interpretations of the data.

7 Conclusion Overall, the present analyses demonstrate that caregivers’ acceptance of care robots is shaped by a wide range of interconnected factors, including the specific type of robot, the sociocultural context in which it is introduced, and the characteristics, expectations, and concerns of the different stakeholders involved. The findings further suggest that care robots are not evaluated solely according to their technical functionality or efficiency, but also according to the particular forms of caregiving relationships they mediate. In this sense, participants’ evaluations appeared to be strongly 21

References

influenced by how different robotic systems configured autonomy, bodily vulnerability, control, intimacy, assistance, and human oversight within caregiving interactions. Moreover, the empirical findings revealed the ethical complexity involved in evaluating care robots, as participants frequently articulated contradictory moral evaluations, competing ethical priorities, and tensions between different values. Rather than relying on isolated ethical principles, participants often evaluated these technologies through interconnected ethical values that attempted to account for caregiving practices in a more holistic, relational, and context-sensitive manner. Ultimately, the findings highlight the importance of developing care robotic technologies that are not only functional and efficient, but also socially meaningful, culturally responsive, and aligned with the values, needs, and lived realities of the communities in which they are intended to operate. More broadly, the study demonstrates that understanding care robot acceptance requires moving beyond instrumental evaluations of technology to a more complex ethical discussion in order to examine how robotic systems transform the social, relational, and ethical dimensions through which care itself is understood and practiced.

[1] Silvera-Tawil, D.: Robotics in healthcare: a survey. SN Computer Science 5(1), 189 (2024) [2] Sequeira, J.S.: Can social robots make societies more human? Information 9(12), 295 (2018) [3] Mohan, R., Arce, J., Mokhtar, S., Cattaneo, D., Valada, A.: Syn-mediverse: a multimodal synthetic dataset for intelligent scene understanding of healthcare facilities. IEEE Robotics and Automation Letters 9(8), 7094– 7101 (2024) [4] Hindel, J., Mekic, E., Enamundram, N.K., Mohan, R., Cattaneo, D., Kalweit, M., Valada, A.: Dynamic robot-assisted surgery with hierarchical class-incremental semantic segmentation. In: International Workshop on Applications of Medical AI, pp. 246–257 (2025). Springer [5] De Graaf, M.M., Allouch, S.B.: Exploring influencing variables for the acceptance of social robots. Robotics and autonomous systems 61(12), 1476–1486 (2013) [6] Johansson-Pajala, R.-M., Gustafsson, C.: Significant challenges when introducing care robots in swedish elder care. Disability and Rehabilitation: Assistive Technology 17(2), 166–176 (2022)

Funding. This work was funded by the Carl Zeiss Foundation with the ReScaLe project and the German Research Foundation (DFG) Emmy Noether Program grant number 468878300. Data Availability. The complete dataset and coding materials have been deposited in a public repository and will be made available upon reasonable request.

[7] Frennert, S.: Expectations and sensemaking: Older people and care robots. In: Proceedings of the International Conference on Human Computer Interaction, pp. 191–206 (2020). Springer

Declarations

[8] Johansson-Pajala, R.-M., Thommes, K., Hoppe, J.A., Tuisku, O., Hennala, L., Pekkarinen, S., Melkas, H., Gustafsson, C.: Care robot orientation: What, who and how? potential users’ perceptions. International Journal of Social Robotics 12(5), 1103–1117 (2020)

Ethics Approval and Consent to Participate. Before the experiment, participants received all the information and gave informed consent to participate in the study. Conflict of Interest. The authors declare that they have no financial or non-financial conflicts of interest.

[9] Bruno, B., Recchiuto, C.T., Papadopoulos, I., Saffiotti, A., Koulouglioti, C., Menicatti,

22

R., Mastrogiovanni, F., Zaccaria, R., Sgorbissa, A.: Knowledge representation for culturally competent personal robots: requirements, design principles, implementation, and assessment. International Journal of Social Robotics 11(3), 515–538 (2019)

Handbook of the Law, Ethics and Policy of Artificial Intelligence (7), 131–150 (2022) [18] Venkatesh, V., Morris, M.G., Davis, G.B., Davis, F.D.: User acceptance of information technology: Toward a unified view. MIS quarterly, 425–478 (2003)

[10] Sanoubari, E., Young, J.E.: Explicit, neutral, or implicit: a cross-cultural exploration of communication-style preferences in human robot interaction. In: Companion of the 2018 ACM/IEEE International Conference on Human-Robot Interaction, pp. 237–238 (2018)

[19] Reinares-Lara, E., Olarte-Pascual, C., Pelegrı́n-Borondo, J.: Do you want to be a cyborg? the moderating effect of ethics on neural implant acceptance. Computers in Human Behavior 85, 43–53 (2018) [20] Holden, R.J., Karsh, B.-T.: The technology acceptance model: its past and its future in health care. Journal of biomedical informatics 43(1), 159–172 (2010)

[11] Broadbent, E., Stafford, R., MacDonald, B.: Acceptance of healthcare robots for the older population: Review and future directions. International Journal of Social Robotics 1(4), 319–330 (2009)

[21] Roy, P.: Applying utaut2 in qualitative research: A systematic review. International Journal of Qualitative Methods 24 (2025)

[12] Londoño, L., Röfer, A., Welschehold, T., Valada, A.: Doing right by not doing wrong in human-robot collaboration. arXiv preprint arXiv:2202.02654 (2022)

[22] Torras, C.: Ethics of social robotics: Individual and societal concerns and opportunities. Annual Review of Control, Robotics, and Autonomous Systems 7(1), 1–18 (2024)

[13] Livanec, S., Londoño, L., Gorki, M., Röfer, A., Valada, A., Kiesel, A.: Designing for difference: How human characteristics shape perceptions of collaborative robots. arXiv preprint arXiv:2507.16480 (2025)

[23] Yuan, S., Coghlan, S., Lederman, R., Waycott, J.: Ethical design of social robots in aged care: A literature review using an ethics of care perspective. International Journal of Social Robotics 15(9), 1637–1654 (2023)

[14] Klebbe, R., Friese, C.: Exploring the role of robots in inpatient care: Caregiver’s perspectives on the development and evaluation of collaborative robot applications – a qualitative research approach. International Journal of Social Robotics (2025)

[24] Leineweber, M., Keusgen, C.V., Bubeck, M., Haltaufderheide, J., Ranisch, R., Klingler, C.: Ethical aspects of the use of social robots in elderly care–a systematic qualitative review. arXiv preprint arXiv:2505.09224 (2025)

[15] Bon, A., Dittoh, F., Lô, G., Pini, M., Bwana, R.M., Cheah, W., Kulathuramaiyer, N., Baart, A.: Decolonizing Technology and Society: A Perspective from the Global South. Cham, Switzerland (2022)

[25] Londoño, L., Hurtado, J.V., Hertz, N., Kellmeyer, P., Voeneky, S., Valada, A.: Fairness and bias in robot learning. Proceedings of the IEEE 112(4), 305–330 (2024)

[16] Garcia, E.V.: Technology for whom and for what? a global south view of tech diplomacy. Global Policy (2025). Policy Analysis, Early View

[26] Vandemeulebroucke, T., De Casterlé, B.D., Gastmans, C.: The use of care robots in aged care: A systematic review of argument-based ethics literature. Archives of gerontology and geriatrics 74, 15–25 (2018)

[17] Gordon, R.E.: Development disrupted: Technology and the global south. The Cambridge

[27] Van Wynsberghe, A.: Designing robots for

23

care: Care centered value-sensitive design. Machine ethics and robot ethics, 185–211 (2020)

[37] Niemelä, M., Heikkilä, P., Lammi, H., Oksman, V.: A social robot in a shopping mall: Studies on acceptance and stakeholder expectations. In: Korn, O. (ed.) Social Robots: Technological, Societal and Ethical Aspects of Human-Robot Interaction. Human–Computer Interaction Series, pp. 119–144. Springer, ??? (2019)

[28] Hung, L., Mann, J., Perry, J., Berndt, A., Wong, J.: Technological risks and ethical implications of using robots in long-term care. Journal of Rehabilitation and Assistive Technologies Engineering 9, 1–12 (2022)

[38] Formosa, P.: Robot autonomy vs. human autonomy: Social robots, artificial intelligence (ai), and the nature of autonomy. Minds and Machines 31(4), 595–616 (2021)

[29] Vandemeulebroucke, T., Casterlé, B.D., Gastmans, C.: Ethics of socially assistive robots in aged-care settings: A socio-historical contextualisation. Journal of Medical Ethics 46(2), 128–136 (2020)

[39] Johnston, C.: Ethical design and use of robotic care of the elderly. Journal of Bioethical Inquiry 19(1), 11–14 (2022)

[30] Gilligan, C.: In a Different Voice: Psychological Theory and Women’s Development. Harvard university press, ??? (1993)

[40] Graaf, M.M.A., Allouch, S.B.: Exploring influencing variables for the acceptance of social robots. Robotics and Autonomous Systems 61(12), 1476–1486 (2013)

[31] Tronto, J.C.: Who Cares?: How to Reshape a Democratic Politics. Cornell University Press, Ithaca, NY (2015)

[41] Leterme, C.: New north–south issues in the digital economy. Centre Tricontinental (CETRI) (2020)

[32] Nussbaum, M.C.: Women and Human Development: The Capabilities Approach vol. 3. Cambridge university press, ??? (2000)

[42] Yu, D., Rosenfeld, H., Gupta, A.: The ’AI Divide’ Between the Global North and the Global South. Accessed: 23 January 2026 (2023). https://www.weforum.org/stories/20 23/01/davos23-ai-divide-global-north-globa l-south/

[33] Sinnott-Armstrong, W.: Consequentialism. The stanford encyclopedia of philosophy. Available at: https://plato. stanford. edu/archives/win2015/entries . . . (2015) [34] Floridi, L., Cowls, J.: A unified framework of five principles for ai in society. Machine learning and the city: Applications in architecture and urban design, 535–545 (2022)

[43] Hurtado, J.V., Londoño, L., Valada, A.: From learning to relearning: A framework for diminishing bias in social robot navigation. Frontiers in Robotics and AI 8, 650325 (2021)

[35] Walters, M.L., Syrdal, D.S., Dautenhahn, K., Boekhorst, R., Koay, K.L.: Avoiding the uncanny valley: Robot appearance, personality and consistency of behavior in an attentionseeking home scenario for a robot companion. Autonomous Robots 24(2), 159–178 (2008)

[44] Ahmad, S.M., et al.: Understanding the impact of robots’ embodiment on user acceptance in social robotics. International Journal of Social Robotics (2025) [45] Ferrer, G., Garrell, A., Sanfeliu, A.: Socialaware robot navigation in urban environments. In: 2013 European Conference on Mobile Robots, pp. 331–336 (2013). IEEE

[36] Chatzoglou, P.D., Lazaraki, V., Apostolidis, S.D., Gasteratos, A.C.: Factors affecting acceptance of social robots among prospective users. International Journal of Social Robotics 16(6), 1361–1380 (2024)

[46] Ferrer, G., Zulueta, A.G., Cotarelo, F.H., Sanfeliu, A.: Robot social-aware navigation

24

framework to accompany people walking sideby-side. Autonomous robots 41(4), 775–793 (2017)

[57] Dutta, N., Dhar, D.: Understanding medical technology innovation in low-and middleincome countries: factors, impact, and a model proposal. She Ji: The Journal of Design, Economics, and Innovation 10(2), 192–222 (2024)

[47] Lawrence, S.: The role of social norms in human–robot interaction. ACM Transactions on Human-Robot Interaction (2025)

[58] Gereffi, G.: Global Value Chains and Development: Redefining the Contours of 21st Century Capitalism. Cambridge University Press, Cambridge (2018)

[48] Marchesi, S., Roselli, C., Wykowska, A.: Cultural values, but not nationality, predict social inclusion of robots. In: Ge, S.S., Cabibihan, J.J., Salichs, M.A., Broadbent, E. (eds.) Social Robotics. Lecture Notes in Computer Science, vol. 13062, pp. 49–58. Springer, ??? (2021)

[59] Dachs, J.N.W., Ferrer, M., Florez, C.E., Barros, A.J., Narvaez, R., Valdivia, M.: Inequalities in health in latin america and the caribbean: descriptive and exploratory results for self-reported health problems and health care in twelve countries. Revista panamericana de salud publica 11(5-6), 335–355 (2002)

[49] Li, H., Milani, S., Krishnamoorthy, V., Lewis, M., Sycara, K.: Perceptions of domestic robots’ normative behavior across cultures. In: Proceedings of the 2019 AAAI/ACM Conference on AI, Ethics, and Society (AIES ’19), pp. 345–351. ACM, ??? (2019)

[60] Churata, D.V.P., Fantinato, M., Peres, S.M., Yassuda, M.S., Melo, R.C., Cachioni, M., Paradeda, R.B.: Exploring factors shaping social robot acceptance in older-adult care: Insights from brazilian caregivers. Expert Systems, 70220 (2026)

[50] Kovner, A.R., Knickman, J.R., Weisfeld, V.D.: Jonas and Kovner’s Health Care Delivery in the United States. Springer, ??? (2011) [51] Matheny, M., Israni, S.T., Ahmed, M., Whicher, D.: Artificial intelligence in health care: The hope, the hype, the promise, the peril. NAM Special Publication (2020)

[61] Wohlsen, M.: The Incredible Hospital Robot That Saves Lives—and Makes Everyone Hate It (2015). https://www.wired.com/2015/02/ incredible-hospital-robot-saving-lives-also-h ate/ Accessed 2025-07-16

[52] Lee, D., et al.: Levels of autonomy in fdacleared surgical robots. npj Digital Medicine 7(1), 45 (2024)

[62] Nursing Times: Experimental Robot Could Help Nurses of the Future Lift Patients (2015). https://www.nursingtimes.net/archive/exp erimental-robot-could-help-nurses-of-the -f uture-lif t-patients-27-02-2015/ Accessed 2025-07-16

[53] Abdullah, A., et al.: How are us hospitals adopting artificial intelligence? early evidence from 2022. Health Services Research (2024). Early online publication [54] Cruz, P., et al.: Robotics applications in the hospital domain: A literature review. Robotics 13(6), 125 (2024)

[63] GovInsider: How Robots Can Help Healthcare Workers’ Bedside Manner (2023). https://go vinsider.asia/intl-en/article/how-robots-can -help-healthcare-workers-bedside-manner-u ipath-ncs Accessed 2025-07-16

[55] Soljacic, F., Law, T., Chita-Tegmark, M., Scheutz, M.: Robots in healthcare as envisioned by care professionals. Intelligent Service Robotics 17(3), 685–701 (2024)

[64] Haseltine, W.A.: Revolutionizing Rehabilitation: How Exoskeletons Are Restoring Mobility for Neuromuscular Patients (2023). https: //www.forbes.com/sites/williamhaseltine/20 23/07/31/revolutionizing-rehabilitation-how -exoskeletons-are-restoring-mobility-for-neu

[56] Schneegans, S., Lewis, J., Straza, T.: The Race Against Time for Smarter Development. UNESCO Paris, ??? (2021)

25

romuscular-patients/ Accessed 2025-07-16

Research 19(11), 367 (2017)

[65] Reinares-Lara, E., Olarte-Pascual, C., Pelegrı́n-Borondo, J.: Do you want to be a cyborg? the moderating effect of ethics on neural implant acceptance. Computers in Human Behavior 85, 43–53 (2018)

[73] Wynsberghe, A.: Healthcare Robots: Ethics, Design and Implementation. Routledge, London (2016) [74] Liu, L., Daum, C., Cruz, A.M., Neubauer, N., Perez, H., Rı́os Rincón, A.: Ageing, technology, and health: Advancing the concepts of autonomy and independence. Canadian Journal of Occupational Therapy 89(5), 402–410 (2022)

[66] Shiferaw, K.B., Mengiste, S.A., Gullslett, M.K., Zeleke, A.A., Tilahun, B., Tebeje, T., Wondimu, R., Desalegn, S., Mehari, E.A.: Healthcare providers’ acceptance of telemedicine and preference of modalities during covid-19 pandemics in a low-resource setting: An extended utaut model. Plos one 16(4), 0250220 (2021)

[75] Apostolova, I., Lanoix, M.: The importance of developing care-worker-centered robotic aides in long-term care. Bioethics 36(2), 170–177 (2022)

[67] Rahimi, B., Nadri, H., Afshar, H.L., Timpka, T.: A systematic review of the technology acceptance model in health informatics. Applied clinical informatics 9(03), 604–634 (2018)

[76] Turkle, S.: Alone Together: Why We Expect More from Technology and Less from Each Other. Basic Books, New York (2011) [77] Vallor, S.: Moral deskilling and upskilling in a new machine age: Reflections on the ambiguous future of character. Philosophy & Technology 28(1), 107–124 (2015)

[68] Venkatesh, V., Morris, M.G., Davis, G.B., Davis, F.D.: User acceptance of information technology: Toward a unified view. MIS Quarterly 27(3), 425–478 (2003)

[78] Scoglio, A.A.J., Reilly, E.D., Gorman, J.A., Drebing, C.E.: Use of social robots in mental health and well-being research: Systematic review. Journal of Medical Internet Research 21(7), 13322 (2019)

[69] Hsieh, H.-F., Shannon, S.E.: Three approaches to qualitative content analysis. Qualitative Health Research 15(9), 1277–1288 (2005)

[79] Kabacińska, K., Prescott, T.J., Robillard, J.M.: Socially assistive robots as mental health interventions for children: A scoping review. International Journal of Social Robotics 13(5), 919–935 (2021)

[70] Broadbent, E., Garrett, J., Jepsen, N., Ogilvie, V.L., Ahn, H.S., Robinson, H., Peri, K., Kerse, N., Rouse, P., Pillai, A., et al.: Using robots at home to support patients with chronic obstructive pulmonary disease: pilot randomized controlled trial. Journal of medical Internet research 20(2), 8640 (2018)

[80] Rabbitt, S.M., Kazdin, A.E., Scassellati, B.: Integrating socially assistive robotics into mental healthcare interventions: Applications and recommendations for expanded use. Clinical Psychology Review 35, 35–46 (2015)

[71] Wall, P., Saxena, A., Brown, I.: Ai in the global south: Potential and risks. AI & Society (2021)

[81] Riek, L.D.: Robotics technology in mental health care. In: Luxton, D.D. (ed.) Artificial Intelligence in Behavioral and Mental Health Care, pp. 185–203. Elsevier, San Diego, CA (2016)

[72] Greenhalgh, T., Wherton, J., Papoutsi, C., Lynch, J., Hughes, G., A’Court, C., Hinder, S., Fahy, R., Procter, J., Shaw, S.: Beyond adoption: A new framework for theorizing and evaluating nonadoption, abandonment, scaleup, spread, and sustainability of health and care technologies. Journal of Medical Internet 26

Fig. 6: Observed proportions of perceived robot suitability across medical contexts by language using Cochran’s Q analysis.

8 Supplementary Material As part of the study, participants were also asked to indicate the healthcare settings in which they believed each robot could be appropriately implemented. This question was included to quantitatively examine whether different types of robots were perceived as more suitable for particular healthcare contexts than others (see Figure 6). The results revealed a remarkably similar pattern across cultures. In both language groups, robots were consistently perceived as less appropriate for psychiatric settings than for any other healthcare context. This finding complements the qualitative results, in which participants repeatedly expressed concerns that robot deployment in psychiatric environments could increase stress, anxiety, confusion, or emotional distress among vulnerable patients.

27

Related documents

• Healthcare reimagined: Carebots from hospital to home#382003
• Digital transformation of home-based care: a narrative review of nursing, technology, and management synergies.#409368
• CARE-HOUSE: developing a framework for conceptualizing social implications of digital health technologies in palliative care.#140689
• Home-Based Digital Healthcare Interventions for Dementia: A Systematic Review of Patient and Family Caregiver Outcomes.#12970
• The impact of inpatient medication administration technologies on nursing autonomy and practice: a meta-ethnographic synthesis of the qualitative literature.#236104