Users’ ambivalent sense of security with humanoid robots in healthcare

ABSTRACT

Humanoid robots have already been shown to be useful in healthcare. To ensure successful interactions with humanoid robots, is it essential that the factors that influence users’ sense of security be understood. Ensuring patients’ sense of security is considered a key principle of good caring. The aim of this study was to illuminate users’ sense of security with humanoid robots in healthcare. Twelve semi-structured interviews were conducted. The participants consisted of five women and seven men aged 24-77. Before being interviewed, the participants were shown a video vignette with Pepper, a fully developed humanoid robot, used in the daily care of patients. The data material was analyzed using qualitative content analysis. Most participants perceived the use of humanoid robots in healthcare to be both positive and negative. The overarching theme was: Ambivalent sense of security with humanoid robots in healthcare. The four categories revealed were: Humanoid robots are both reliable and unreliable, Humanoid robots are both safe and unsafe, Humanoid robots are both likable and scary, and Humanoid robots are both caring and uncaring. Here we increased knowledge of whether patients perceive a sense of security with humanoid robots in healthcare, including which fears users have. This is vital information that should be taken into consideration when further developing and introducing humanoid robots into the healthcare setting.

KEYWORDS:

• Humanoid robot
• sense of security
• ethics
• qualitative approach
• healthcare

Introduction

The provision of good and adequate care is a challenge today, because an aging population can be associated with an increased need for care.¹ The integration of robots into healthcare can help provide a solution to the need for more resources in healthcare,^2,3 especially humanoid robots.⁴ Humanoid robots have already been shown to be useful in healthcare. Designed to look like a human, with legs, arms and a head, most also have “eyes” and a “mouth”, i.e., sensors or design features used to imitate the actions of a human’s eyes and mouth. Most humanoid robots are designed with vision systems, such as sensors or cameras that allow them to focus attention on specific objects or screens.⁵ Care-O-Bot,⁶ Hobbit,⁷ BEAR,⁸ Nao,⁹ Romeo¹⁰ and Pepper¹¹ are all humanoid robots suitable for performing daily chores usually performed by humans.¹² Humanoid robots are able to assist with physical tasks (e.g., walking, retrieving, bathing), cognitive training (e.g., reminding, playing memory games), health management (e.g., monitoring blood pressure, detecting falls) or interaction (e.g., companionship, entertainment).¹³ As the number of humanoid robots increase and hardware and operating costs decrease, it is inevitable that humans will be interacting with larger numbers of robots that perform increasingly complex tasks in healthcare.

To ensure that humanoid robots are useful and of benefit in healthcare, a human rather than a technical point of view should be taken into consideration during their technical development. Various factors such as safety and user preferences should be taken into account.¹¹ Researchers have investigated users’ attitudes toward humanoid robots in several studies. In some studies different scales for measuring user’s attitudes have been used (among others,^12,13,14) while in other studies research models have been developed and used to analyze users’ acceptance of humanoid robots.^15,16 Even though users’ attitudes toward and acceptance of humanoid robots are important, users’ sense of security with humanoid robots with regard to sustained interaction is perhaps of greater interest, because those receiving care can be considered vulnerable (cf. 17). We found few qualitative studies on users’ experiences of interacting with humanoid robots. The aim of this study was to illuminate users’ sense of security with humanoid robots in healthcare.

Background

Ensuring patients’ sense of security is considered a key principle of good caring.¹⁷ As a concept, sense of security can be related to other concepts such as feeling safe, safety, confidence and assurance. In Maslow’s (1972) theory of human needs, the need for safety is one of the most important human needs, and feeling safe and secure are related to being free from the threat of physical or emotional harm. Segesten¹⁸ divides security into two dimensions, basic security and situational security. According to Segesten, basic security is related to what one can call personal circumstances: when a person has self-confidence that is created at an early age through a belief in and knowledge of oneself, which then lays the foundation for an inner security. Situational security, in contrast, is related to external circumstances, such as material security, environmental security or relationship security. Furthermore, Segesten maintains that while preconditions for security can exist and be created, security is ultimately an individual experience.¹⁸

Researchers have investigated patients’ sense of security from many different perspectives and in many different contexts. Children and young people in residential care experienced an increase in their sense of security when cared for by competent and trustworthy staff they perceived to be caring, proactive and tenacious in building relationships and when they perceived that staff listened to them and ensured that they had a voice.¹⁹ Elderly patients experienced a sense of security when nurses provided oversight, were predictable, provided personalized care and were willing to advocate for them.²⁰ Patients in intensive care experienced a sense of security when their psychosocial needs were met, when they regained control, when healthcare professionals were perceived as being trustworthy and when their family were present.²¹ Hospitalized patients experienced an increase in their sense of security when they perceived that they were cared for and received adequate knowledge and when nurses and family were present.²² Wijk²³ maintains that the healthcare environment influences patients’ sense of security, and notes that an environment that can be easily navigated leads to an increased sense of independence, which in turn gives individuals an increased sense of security. Other researchers find that a sense of security is an important factor that contributes to patients’ recovery and health.²⁴ For example, Fagerström et al²⁵ find that older people who are safe perceive that life feels meaningful and crises are manageable.

Methods

Participants and data collection

Twelve semi-structured interviews were conducted in Ostrobothnia, Finland in the spring of 2018. The participants consisted of five women and seven men aged 24–77. The participants were recruited through convenience selection within personal networks of the researchers and represented different professional groups: students, teachers, engineers, consultants and pensioners, among others. Prior to inclusion in this study, the participants were aware of the existence of humanoid robots but had not seen them in use. Before being interviewed, the participants were shown a video vignette. The video vignette was made by the commercially available entity (the vendor for Pepper), and it depicted the robot assisting with care of patients in a hospital. The use of a video vignette can stimulate participant responses by awakening emotions and attitudes (cf.²⁶) In the video vignette used here, Pepper, a fully developed humanoid robot, was shown being used in the daily care of patients. Pepper stands 140 centimeters tall and can capture information, identify human voices and detect human emotions through cameras and sensors linked to a cloud-based Artificial Intelligence (AI) service.²⁷ Pepper is designed to learn from interaction with humans, and its learning is shared with other robots. The intention underlying the use of the video vignette was to give the participants a realistic picture of Pepper’s functional abilities.

Following the video vignette, individual semi-structured interviews were conducted, during which the participants were asked to reflect on humanoid robots in healthcare. The interviews took place in a location chosen by each participant. The interviews were conducted in the form of a dialogue, and an interview guide (cf. ²⁸) was used. The interview guide included the following questions, among others: How would it feel to be greeted by or taken care of by a humanoid robot? What are the advantages and disadvantages of humanoid robots? Would you feel safe and secure when meeting a humanoid robot (why/why not)? The questions included in the interview guide were supplemented with follow-up questions to deepen the participants’ responses. The interviews lasted between 25–60 minutes and were digitally recorded. The recordings were thereafter transcribed into text, resulting in 52 pages of text in total.

Data analysis

Data were analyzed using qualitative content analysis, inspired by Graneheim and Lundman.²⁹ The interview material, 52 pages of text from all twelve interviews, was first read through several times so that we could gain an initial understanding of the content of the entire material. Meaning units (words, sentences or paragraphs) relevant to the study purpose were then identified in the material. The meaning units were condensed, coded, compared (to find similarities and differences), and sorted into categories and subcategories. The coding was conducted within the research group; discrepancies were discussed, which resulted in an agreement between the researchers on the categorizations.The final analysis resulted in four categories. As a final step, latent analysis was used to elicit an overarching theme, derived from the categories’ underlying meaning.

Ethical considerations

The guidelines for good scientific praxis as described by the Finnish Advisory Board on Research Integrity³⁰ have been followed during the course of this research. The participants were given information about the study, and the purpose of the study was explained. Participants were guaranteed confidentiality and informed that participation was voluntary. The collected data material was securely stored, with access given only to those researchers directly involved with the study.

Findings

Most participants perceived the use of humanoid robots in healthcare to be both positive and negative. The overarching theme was: Ambivalent sense of security with humanoid robots in healthcare. The four categories revealed were: Humanoid robots are both reliable and unreliable, Humanoid robots are both safe and unsafe, Humanoid robots are both likable and scary, and Humanoid robots are both caring and uncaring.

Humanoid robots are both reliable and unreliable

We saw that the participants held ambivalent views on humanoid robots with regard to reliability. The participants maintained that they would feel safe being cared for by a humanoid robot. They also noted that they perceived robots to be reliable, because robots work with precision, do what they are programmed to do, make no mistakes and only act in accordance with how they are programmed. They stated that robots act predictably and consistently, are quick, and never miss anything important. According to one participant, “Robots never have a bad day and never need to rest”. The participants maintained that good intentions underlie how robots are programmed, with the intention being to help. Some participants even maintained that they would feel safer being cared for by robots than humans, because robots are more reliable. Especially with regard to surgery, the participants maintained that robots were more dependable, “because the robot [has a] steadier hand”. Some participants also emphasized that robots can access much more information than humans.

Yet some participants noted their reservations about being cared for by robots, even though they generally perceived robots to be safe. The participants stated that whether they felt safe with a robot depended on whether the robot was supervised by healthcare personnel or capable of independent decision-making. As one participant stated, “I probably think there is another surgeon standing next to the robot and following everything unerringly minutely. It is the robot that to a great extent does the procedure, but in some ways the doctor controls it a little, the robot itself”. Other participants maintained that because robots are unreliable they would not feel safe being cared for by a robot, further revealing that they perceived that robots could be of use in industry, but not healthcare. “Robots can move things here and there; they can clean and make beds. But robots are not meant to be used with people”.

Humanoid robots are both safe and unsafe

We also found that the participants held ambivalent views on humanoid robots with regard to patient safety. The participants maintained that a person’s sense of security when receiving care depends on the situation and one’s illness. Some participants noted that being cared for by a robot was risk-free and safe as long as one was not seriously ill, i.e., the treatment could be considered a minor procedure or the first step in treatment. These participants expressed the belief that the use of humanoid robots was acceptable for “simpler” treatment for non-threatening illnesses, because the situation was “under control”. They continued by stating that it could be advantageous for robots to perform what they themselves called “simple examinations and tests”. Other participants perceived that robots facilitated the registration of patient data, noting that they found it beneficial to be able to enter their own data directly into information systems.

Nevertheless, some participants stated that they would not feel safe receiving care from a robot, referring to the risks associated with using technology, including that they were afraid that the technology would suddenly stop working or even break down. We saw that the participants were not technical experts and were somewhat unaware or uncertain of humanoid robots’ technological capacities. Several participants highlighted that they worried about robots’ software being manipulated by malicious hackers. Some maintained that there was a

risk that the information collected by a robot could be sent to unauthorized persons or used for improper purposes. “After all, it has cameras and everything in it so it can pass [the information] on to someone who wants to know if it is someone who has valuable items with them. I mean, a hacker can merely take control of it like that, so that’s probably something to consider”.

The participants’ ambivalent views on humanoid robots with regard to sense of safety even extended into an association between safety and patient age. Some argued that the use of robots was not appropriate for older people, while others argued that the use of robots was especially good for older people. Several participants stated that children and younger people would not be disinclined toward the use of robots as companions to inhibit a sense of involuntary loneliness, because children and young people are used to new technology and consider it to be harmless. Still, one participant maintained that robots were not safe for children.

Humanoid robots are both likeable and scary

We discovered that the participants held ambivalent views on how humanoid robots are perceived. Some participants found robots to be kind and likeable, which, according to these participants, increases their sense of security. Others stated that they did not like robots that were too lifelike or had too many human traits. “[The robot] gives a trusting impression and it has a pleasant voice and a nice look. It is positive they have not tried to make it too human.”

Several participants maintained that how individuals perceive robots will change over time as people become more accustomed to them. These participants revealed that the first time an individual encounters a robot in healthcare, the robot could be perceived as being scary, but that after the individual has become accustomed, robots “blend into the background”. These same participants stated that they believed robots would be considered a natural element in care in the future. Other participants expressed a completely divergent opinion, stating that robots were scary when they did not look human enough. Some participants found it difficult to understand what robots were saying and perceived robots’ speech as being unclear. These participants noted that they perceived robots’ voices to be unpleasant and stated that the robot they saw in the vignette was emotionless. One participant related that he found the physical presence of robots to be more frightening than when other “invisible” health technology services were used: “Perhaps I myself feel [the robot] to be more threatening than if it were somewhere in a cloud service assisting the nurse in his/her tasks, and not like it is, physically present”.

Humanoid robots are both caring and uncaring

Participants held ambivalent views on whether humanoid robots can be considered caring. Some participants argued that because humanoid robots have positive human traits and can be perceived as caring, users’ sense of security is increased. One participant stated that robots could help eliminate misunderstandings by acting as interpreters between individuals who speak different languages, and that being understood increases both the quality of care and users’ sense of security. Various participants remarked that robots can help patients navigate in a building by giving directions or can monitor patients, while other participants stated that robots can entertain lonely people and provide companionship to make patients feel more secure. Some participants stated that robots can give the impression of being empathetic and caring. Others highlighted that robots can give patients extra attention and improve patients’ functionality. One participant remarked that robots should have warm and soft hands, because this would give older people the feeling that “an actual being” is touching them and give them the sense that robots actually care about their patients. Another participant stated that robots could provide support and encouragement in various situations, i.e., when patients cannot or do not want to eat. Some participants maintained that robots could be more caring than humans and noted that “robot care” would sometimes be preferential. The reasons given for such were that robots treat all patients equally, are present in a completely different way than human caregivers and can help ensure that human caregivers have more time for tasks that robots cannot perform. “I would like it and the knowledge that other staff will be relieved is positive”.

Still, other participants maintained that robots could not in any way be considered caring. They stated that they would not feel safe in encounters with robots, because robots have no emotions. Other participants noted that robots were impersonal and are unable to do what human caregivers can do, i.e., respond to certain questions. “They cannot feel or anything, if you have broken an arm, how it feels and so on.” All of the participants maintained that robots cannot replace human caregivers. Some participants also expressed concern that human caregivers would no longer have employment if robots were used to perform caregivers’ jobs. These participants furthermore expressed the concern that humanity and human contact would disappear from healthcare, maintaining that robots can never replace living, sensitive, human caregivers. “Real people will always be needed in care. It is difficult to beat a human, but it also of course depends on what kind of person it is. If that person can respond to [others] in a good or bad way”.

Discussion

Meeting patients’ needs is a fundamental aspect of caring, and this also includes patients’ sense of security. In our opinion, ensuring that patients feel secure when encountering robots in the caring context should be one of the cornerstones of healthcare robotics in the future. Here we saw that the participants expressed an ambivalent sense of security with humanoid robots in healthcare.

As expressed in the study themes, the participants perceived humanoid robots to be reliable and unreliable, safe and unsafe, likable and scary, caring and uncaring – all at the same time. We perceived that the participants’ ambivalence was strongest with regard to whether humanoid robots can be considered replacements for human caregivers. We discerned that the participants were fearful that the introduction of humanoid robots into the care setting would result in the loss of human contact and, thereby, would lead to dehumanized healthcare. This is in line with earlier research, in which researchers found that technology could not entirely replace genuine contact with another human being.³¹ The ambivalence that the participants demonstrated can possibly be attributed to the manner in which robots are portrayed in movies, computer games, and so on (cf.³²) Some participants demonstrated limited knowledge of humanoid robots and how they function, which we perceive affected the participants’ perceptions. Lack of knowledge can make it challenging for individuals to consider phenomena in an informed and balanced way.

In terms of perceptions of reliability, we found that if patients are to feel safe, robots should act in a reliable and predictable manner. This is in line with earlier research on older people in relation to caregivers, where researchers found that patients’ sense of security increases when nurses were predictable.²⁰ Another important factor that emerged was that, for some participants, their sense of safety was contingent on human caregivers being present during robot-patient interactions. This can be compared to Moore et al.,¹⁹ who found that available staff increased patients’ sense of security.

Regarding the participants’ views on humanoid robots with regard to patient safety, we discerned that more knowledge of what robots actually do can increase sense of safety. Some participants here stated that receiving care from robots for minor ailments was safe, but that such care for serious illness was unsafe. This is in line with Mollon,²² who stressed that patients’ sense of security increases when patients are given sufficient knowledge. Also Moore¹⁹ found that patients’ sense of security increased when competent staff were present. We therefore infer that for patient’s to experience a sense of security when interacting with robots, knowledge caregivers, who possess robotics knowledge and skill, should be present when robots are introduced into a care setting. Furthermore, the participants here emphasized that their sense of security was dependent on their having control of how patient data is disseminated, with some but not all participants finding it acceptable that robots collect data. This is in line with Wassenaar et al.,²¹ who found that patients feel safe when they have control.

We found that some participants’ sense of security was improved in encounters with lifelike robots (cf., for example,³³) Other researchers have also found that individuals prefer interacting with lifelike robots, because such are perceived to be more functional and more capable of engaging in activities than other, less lifelike robots (^34,35) Some participants maintained that less lifelike robots were scary and that it could be difficult to understand what robots were saying because their speech was unclear. We therefore find that if individuals are to feel confident during interactions with robots, they must be able to understand what a robot is saying and be able to interpret the robot’s expression. This is in line with Wijk,²³ who found that an individual’s experience of security is influenced by how clearly and easily they can interpret an environment. That the participants here expressed differing views on how lifelike robots should look can be compared to Wolpin and Stewart,³⁶ who highlighted the importance of a patient-centered approach to technology, including that users are involved at an early stage in the design and evaluation process.

Regarding the participants’ views on how humanoid robots are perceived, we saw that some participants perceived robots as caring while others perceived them as uncaring. Some participants maintained that robots could never replace humans, while others stated that robots have great potential in the care setting, because they can give patients extra attention and improve patients’ functionality. We note that other researchers can be skeptical of robots’ ability to be caring, and that such researchers have raised critical questions about the ethical implications of robots as caregivers (cf.³⁷) Other researchers argue that it is unwise to disregard a future that includes robot-human interaction, while also warning that machines can never replace human caring and interaction.³⁸ The participants here, however, provided an interesting counterpoint when they mentioned the possibility of using robots as an aid or complement to caregivers, in lieu of replacing caregivers. We propose that humanoid robots can be used to manage automatic tasks that consume a lot of caregivers’ time (such as maintaining medical records), thereby ensuring that caregivers have more time to spend with and in close contact with their patients.

Validity and study limitations

The participants had different backgrounds and can therefore be considered a representation of patients in healthcare. Limitations include the small sample size and the participants’ lack of knowledge about humanoid robots. The use of a video vignette can be considered a further limitation. While video vignettes are useful when one seeks to understand individuals’ impressions, real interaction with humanoid robots in a certain context would have given a more comprehensive understanding of sense of security with humanoid robots in healthcare. In the next step of our research we will seek to include a larger sample and investigate sustained humanoid robot-patient interaction. We will incorporate a focus on the differences in users’ perceptions, in relation to age, gender and different care needs.

Conclusions and implications

More knowledge on this topic is needed, and those involved in the research and development of humanoid robots have a great responsibility to truthfully convey what is possible and what is not with regard to robots’ intelligence and capabilities. If robots are to become truly useful in care, patients must feel secure when interacting with them. Therefore, to ensure successful interactions with robots, is it essential that the factors that influence users’ sense of security be understood. Here we have increased knowledge of users’ sense of security with humanoid robots in healthcare, including which fears users have. This is vital information that should be taken into consideration when further developing and introducing humanoid robots into the care setting. Listening to patients is crucial, because they are the ones who will interact with such technology. We conclude by restating that users’ sense of security with humanoid robots in healthcare is still not fully understood and should therefore be the subject of further research.

Disclosure statement

The authors declare no conflict of interest.

Additional information

Funding

No specific grant from any funding agency in the public, commercial or not-for-profit sectors has been received for this research.