Social robot – Jack of all trades?

1.1 Roles of social robots

At the present moment, robots are already used in public places, such as libraries, hotels, and shopping centers, as well as in hospitals [1,2,3,4] mostly as advisers and assistants. Unlike technically assistive robots, a social robot is defined as a robot with a physical body that interacts with humans by following the behavioral norms expected by the humans [5]. It has a human-like body shape, including a head, two arms, and two legs [6].

To date, there are attempts to adapt social robots to be human companions in purely social activities, such as education and special needs [7,8,9], elderly and patient care [6,10,11,12,13], as well as personal and even intimate relationships [14,15]. There are even several examples of human relationships with robots and other artificial agents as intimate companions: The Gatebox (gatebox.ai) is a virtual anime girl appearing in a small glass-box who manages the smart home and is “in love” with her “master” [16]. Recently, there have even been a few marriages with robots, self-explained by an inability to find a partner and by extreme loneliness. Some researchers go so far as to claim that a marriage with a robot will become routine by the year 2050 [17]. These examples, however, remain an exception. Also, other studies found mixed responses to robots in intimate (not only sexual) relationships [18,19]. Nonetheless, although this field is only emerging, the bets are placed on social robots as quasi a Jack of all trades.

Especially, the prevailing situation of the COVID-19 pandemic provides opportunities to investigate roles of social robots. We believe that in such a situation of social isolation, social robots could provide both a social bond and assistance to humans. The aims of the study are twofold. First, we examine the contribution of human-related (age, gender, personality traits, loneliness and need to belong, attitudes toward robots) and contextual factors (lockdown) to acceptance of social robots. Second, we explore possible roles and qualities of social robots in private and public sectors of human lives. In Section 1.2, we elaborate our research hypotheses by providing a short literature review.

1.2 Factors influencing the acceptance of social robots

Despite vast research in this area, factors that impact robots’ acceptance and avoidance are not well understood. Although research in general has made great progress in understanding separate factors influencing robots’ acceptance, their interplay in terms of possible moderation or mediation relationships is not well studied yet.

In 2008, Nomura et al. [20] developed a seminal scale for measuring negative attitudes toward robots (NARS). Since then, this scale has been used in various contexts and has proven to be an excellent measure of participants’ attitudes toward interaction with and influence of social robots. Closely related to NARS is Robot anxiety [20] as a feeling of dread and an avoidance impulse in the presence of robots. Both measurements were found to predict avoidance behavior toward robots [20,21]. We postulate that NARS and Robot anxiety are negative predictors of a Wish to see a robot both at home and in public places (H1a and H1b).

As for further factors influencing the acceptance of social robots, the current evidence is contentious and might be context dependent [22,23]. A recent review identified controversial evidence concerning factors related to both potential users and robots, such as age or exposure [24].

Several studies have evaluated the acceptance of social robots among the elderly [25,26,27], as teaching assistants [28,29,30,31], and in medical care [32,33]. Although in some studies age and female gender were negatively related to acceptance and willingness to use a robot [23,34,35,50,53], other investigations have found no associations of acceptance of social robots with these variables [21,36,37,38], but rather with psychological factors such as loneliness and life satisfaction [26,39]. However, an interaction study indicated that women responded more positively toward the robots and their communication [38]. As for special applications of social robots as a partner or a sex partner, there is evidence that females have a less positive view on them than males [40]. Reasons for these discrepancies are currently unclear. In line with the existing literature, we postulate that age and female gender are negatively linked to Wish to have a robot at home, especially as a partner and a sex partner, and Wish to see robots in public places (H2a and H2b). Moreover, we expect that gender moderates the relationship between the NARS and a Wish to have a robot as a partner (H3).

Further studies demonstrated that education level could also play a role in forming attitudes toward robots. Also here, previous research results are mixed: Although one study showed a negative link of education and perception of robots as a social entity [41], other research did not find such a relation [41]. Crucially, higher levels of personal experience with robots and exposure to them were shown to be linked to higher robots’ acceptance [36]. We posit that Level of education (H4a), Interest in robots (H4b), and Exposure to social robots (H4c) are positively associated with a Wish to have a robot at home and in public places.

Personality factors, in particular, conscientiousness, extroversion, and neuroticism, were found to be predictors to positive attitudes toward robots [42]. However, a recent meta-analysis indicated that agreeableness, extroversion, and openness are positively correlated with robots’ acceptance, whereas conscientiousness and neuroticism did not play any role [43]. In the current study, we aim to explore the role of personality factors (Big Five) on the Wish to have social robots in different roles (friend, companion, partner, and sex partner) at home and also in public areas (EH1).

Importantly, the roles of social robots in a human society differ in different countries and cultures [44]. So, Haring et al. [45] found that, besides some similarities in robots perception, Japanese people prefer human-like robots and even trust them to do more personal tasks such as a massage, whereas Europeans see them rather as utilitarian machines. Japanese people have also shown a heightened sensitivity to anthropomorphism [46]. Koreans were even reported to perceive robots as part of society [47]. Earlier studies pointed out that Europeans are both fascinated and afraid of robots because they define humans similar to machines and perceive technology as challenging for their own nature [48]. Interestingly, there is evidence that European people perceive social robots more positively than Easterners after interacting with them [47]. The present study does not aim to test intercultural differences in robot perception, although one needs to keep in mind that the findings could differ in participants with different cultural backgrounds.

1.3 The COVID-19 pandemic and robots as a solution

Loneliness was identified as one of the possible factors influencing acceptance of social robots [49]. Now that because of the COVID-19 pandemic we are facing a drastic change in social reality all over the world, loneliness is a ubiquitous topic. Humanity faces challenges in both the private domain (at home), in terms of loneliness and isolation, and the public domain in terms of contamination risk.

Closely related to Loneliness is the Need to belong as a wish to be connected to others. A considerable body of literature has considered aspects of belonging and loneliness together. Both factors were found to be related to social and psychological functioning [50] and health outcomes [51]. Traditionally, the Need to belong has been conceptualized as a predictor of Loneliness [52]. Although the two terms are often used interchangeably [47], some studies found that they must not necessarily be interrelated [53]. The novel Dual Continuum Model of Belonging and Loneliness proposed by Lim et al. [54] strives to combine the two constructs in one model on a continuum, as independent constructs, with each contributing to the complexity of human social needs.

Social isolation might drive people to search at least some contact – even if the social agent is a machine rather than a human being. Employing social robots as social partners and companions to fight loneliness, especially in a situation of forced social isolation, could be one of the possible solutions. Moreover, these days, robots are especially advantageous because they are immune to infections and less prone to spreading infections than humans are, thus being safer than other humans as interaction partners. In its turn, a feeling of loneliness might boost acceptance of robots in different roles at home. Research with elderly people shows that a robot might help reduce isolation, improve the moods when a person feels lonely [11], and ameliorate the psychological condition of the participants [55,56]. Also, in the context of the COVID-19 pandemic, there is evidence that social robots can help overcome loneliness [57] and satisfy our Need to belong. We predict that Loneliness and the Need to belong are positively associated with participants’ Wish to have a social robot at home, especially as a companion and as a friend (H5a and H5b). We also predict that the Length of the lockdown will be positively related to the Wish to have a robot at home, especially as a friend, a companion, and a sex partner (H6a). This relationship will be mediated by Loneliness (H6b).

Importantly, overcoming loneliness presupposes a certain amount of physical closeness and contact. This is of particular interest when researching social robots as a possible social agent that helps overcome a feeling of loneliness. A preferred distance can depend on the interaction and the qualities, such as emotional impact, of the robot [58]: Although little is known about a comfortable distance to a robot in an intimate relationship or at home, participants reported that in public corridors, robots entering the intimate zone of about 45 cm is considered uncomfortable [59]. And when cooperating with social robots, a comfortable distance of about 80 cm was found [60]. Therefore, we aim at exploring a comfortable Distance to a robot (EH2) and Time participant wish to spend with a robot (EH3).

Finally, earlier surveys indicated that robots were generally perceived as helpful devices and that they ought to be reliable and assistive [61,70,71]. To obtain a more global view on requirements for robots, we also explore the Roles and Tasks robots could have at home and in public places (EH4) and what Qualities robots should have (EH5).

2.1 Participants and procedure

One hundred forty participants (11 native English speakers, 109 native German speakers, 19 participants with another native language, one did not answer this question) filled out an online survey (28 males, 111 females, 1 nonbinary; mean age = 26.06 years, SD = 8.33; 40.7% with university qualification [bachelor and above]; the majority [about 97%] of the participants came from Germany or another European country). The survey was conducted in both German (for German native speakers) and English (for others). All non-native English-speaking participants were fluent in English by self-report. The participants were recruited in social networks such as Facebook^® and Linkedin^® and via the subject pool system SONA at the University of Potsdam during a 6-week period between April 2020 and May 2020.

Before commencing the study, the experimental procedures were reviewed and approved by the Ethics Committee at the University of Potsdam (approval number 25/2020). All participants submitted their informed consent at the beginning of the survey by clicking the relevant online link and were reimbursed with course credits for their participation if required. Participants could also enter their email address to take part in a raffle of one of two vouchers.

The questionnaire was created and hosted with SoSci Survey [62]. Participants were told that the study was dedicated to social robots and their acceptance in the society. They were instructed to answer the survey questions honestly and spontaneously. At the end of the survey, participants were debriefed and given a link to leave their internal subject pool ID for receiving a credit or entering an email address to take part in the raffle.

2.2 Measures

The following variables were measured: (1) Human-related factors: Demographic factors (age, gender, native language, education, country of residence at the moment of the survey conduction), Personality, Loneliness, Need to belong; (2) Attitude factors: NARS, Robot anxiety (to a robot at home and on public places), Wish to have a robot in different spheres of life (at home as a friend, companion, partner, sex partner; in public places), Distance to a robot, Time spent with a robot at home and in public places, Interest in robots, Exposure to social robots; (3) Robot-related factors: Roles and Tasks robots could fulfill at home and in public places, Qualities robots should have at home and in public places; (4) Context-related factors: Whether there was a Lockdown in the country/federal state they lived in, and if yes, for how long (Length of Lockdown).

The variables are summarized in Table 1.

2.3 Data preparation

3.1 H1, H2, H4, H6 correlation analyses

Spearman’s rank correlation was calculated separately between (1) NARS and RAS, (2) Personality variables, and (3) Loneliness and Need to belong and Wish to see a robot at home and in public places in different roles. The results are summarized in Tables 3–5.

As seen in Table 3, NARS have a small positive association with Wish to have no robot at home or in public places and small to moderate negative associations to having a robot at home, as a companion, a friend, a partner, or a sex partner (H1a). NARS also have a small to moderate negative association with Wish to see a robot at the counter, at school, in a hospital, and in other public places (H1a). Especially, the sub-factor of Negative attitudes toward emotional interactions with robots (NARS S3) plays a role.

Robot anxiety toward a robot both as a companion and in public places is positively associated with a refusal to have a robot at all (H1b). The former, especially the Anxiety about the behavioral characteristics of a robot (RAS Companion S2), is also negatively linked to having a robot as a sex partner. The majority of correlations can be classified as small or moderate.

Interestingly, the Anxiety to communication or discourse with robots (RAS S1 and RAS S3) both as a companion and in public places is not or almost not linked to the Wish to have a robot in any of these roles. The majority of negative, although weak, associations with different roles for robots exist related to the Behavioral aspect of robot anxiety (RAS S2), namely, how the robot would move or act.

Further, as shown in Table 4, age was negatively associated with an Acceptance of a robot in a hospital (H2a). Contrary to our expectations, no other correlations were found between age (H2a) and gender (H2b) of participants and a Wish to have a robot at home or in public places.

As expected, education was positively linked to Wish to have a robot at home (H4a): Participants with university education were rather in favor of getting a robot.

In line with our expectations, the more previous Exposure participants had with social robots and the more they were interested in them, the more they were willing to have a robot at home and in public places in most of the roles (H4b and H4c). Again, the majority of correlations can be categorized as small or moderate.

As postulated, our results show that the lonelier a person feels, the more disposed he or she is to have a robot as a companion, as a friend, a partner, and even as a sex partner (H5a). That being said, the separate Loneliness subscales were related to this wish in a different way. Although the Intimate others and the Social others aspects of Loneliness were linked to all the four potential roles of the robot, the Affiliative environment aspect was not associated with any of them. The Intimate others aspect focuses on the lack of companionship and isolation, the Social others aspect deals with being close to other people and having someone to turn to, and the Affiliative environment aspect is related to being a part of a social group and not feeling alone.

However, Need to belong was negatively linked only to the Wish to have a robot as a partner (H5b). This shows that the more a person wants to belong and to be accepted by other people, the less they can imagine having a robot as a partner. Apparently, such a person does not consider an artificial agent as possible substitute as human bonds.

Almost all other correlations can be classified as weak or moderate (0.20 to 0.39). Willingness to see a robot in various public places (counter, school, hospital, other public places) was not linked to any of the variables in question. Finally, the longer a Lockdown in the country had been, the more favorable the participants were to having a robot as their sex partner (H6a). This relationship was not mediated by Loneliness (H6b, Table 6).

In summary, various parameters such as Loneliness, Attitudes toward robots, Robot anxiety, Exposure to and Interest in robots, education and native language, but not personality qualities (except for Conscientiousness), or gender were found to be linked to Acceptance of a robot at home in different qualities. Age was only negatively associated with the Wish to see a robot in a hospital.

3.2 Exploratory analyses EH1–EH3

Conscientiousness was negatively associated with a robot at school (EH1). No other correlations were found between Big Five personality measures (EH1) of participants and a Wish to have a robot at home or in public places.

EH2 and EH3 were tested using a Wilcoxon signed-rank test. Distance to a robot (EH2) and Time spent with a robot companion at home and a robot in public places (EH3) did not differ significantly. However, there was a slight descriptive tendency showing that in public places, participants prefer to stay further to robots (M _home = 2.71, M _public = 2.49, Z = −1.68, p = 0.09) and wish to spend more time with them (M _home = 2.21, M _public = 2.39, Z = −1.28, p = 0.20).

3.3 Predictors of robots’ acceptance

The results of the regression analyses are summarized in Tables 7 and 8.

Linearity assumption check showed that the relationship between Exposure and Wish to have a robot at home is better explained by a square term of the Exposure variable (Figure 1), and the relationship between Loneliness (Intimate) and Wish to have a robot as a companion is better explained by the cubic term of the Loneliness variable (Figure 2).

Figure 1

Wish to have a robot at home explained by Exposure to social robots.

Figure 2

Wish to have a robot as a companion explained by Loneliness (Intimate).

Regression analyses support partially H1a, H1b, H4b, H4c, H5a, and H6 and contradict H2a and H2b. The relationship between Exposure to social robots and a Wish to have a robot at home is better explained by a square model (Figure 1). The relationship between Loneliness and a Wish to have a robot as a companion is explained by a cubic model (Figure 2). Moreover, we found that conscientiousness and agreeableness predict a Wish to have a robot at school and at other public places (EH1).

To test the hypothesis that participant gender moderates the relationship between NARS and a Wish to have a robot as a partner, a hierarchical multiple regression analysis using the PROCESS Macros [74] (Model 1) was conducted. Indeed, the relationship between NARS and Wish to have a robot as a partner was different for men and women (H3). Specifically, for men, as NARS increased, the Wish to have a robot as a partner decreased (B = −0.67, t(130) = −2.54, p < 0.05), whereas for women, the relationship between NARS and Wish to have a robot as a partner was not significant (B = −0.07, t(130) = −0.71, p = 0.478). In general, women were not inclined to see a social robot as their partner. The results of the moderation analysis are summarized in Table 9. Figure 3 illustrates the model.

Figure 3

Gender as moderator of the relationship between Negative attitudes toward robots and a Wish to have a robot as a partner. Gender is coded as dummy variable with values 0 = male, 1 = female. Note. The y-axis represents the average ratings on the item “Wish to have a robot as s partner.”

3.4 Robot’s roles and qualities EH4–EH5

Table 10 provides an overview about the Roles and tasks of a robot at home and in public places chosen by the participants (EH4). Most participants preferred to see either no robot at all or a robot as a helper and assistant. In these Roles, robots were suggested to be used for cleaning and cooking; only a few participants could imagine a robot as a friend or in an intimate relationship. Also, in public places, the most preferred choice of a task for the robot was cleaning followed by transportation. Only 19.3% of all participants could imagine a robot in childcare.

In a further question, participants were asked to evaluate different Qualities a robot at home and in a public place should have (EH5). Table 11 summarizes the descriptive statistics of these parameters, compared between a robot at home and in public places. Significant differences were found in only four Qualities. Both robots at home and robots in public places are expected to be extremely reliable, considerate, and agreeable, but moderately emotional, talkative, and companionate (Figure 4). In comparison to robots at home, robots in public places are desired to be more controllable (M _h = 4.16, M _p = 4.49) and predictable (M _h = 3.99, M _p = 4.47), but less breakable (M _h = 2.09, M _p = 1.90) and emotional (M _h = 3.05, M _p = 2.47).

Figure 4

Robots’ qualities desired at home and in public places, measured on a scale from 1 (Not at all) to 5 (To a great extent).

4.1 Human-related factors

4.2 Contextual factor lockdown

As for the specific characteristics of the pandemic situation, we found that the longer a Lockdown in the part of the country was, the more inclined the participants were to have a robot as their sex partner. However, this relationship was not mediated by Loneliness.

4.3 Dependent variables

5.1 Limitations of the present study

Our findings should doubtlessly be scrutinized in future studies to overcome the limitations of the present study. First, the present study was of a correlative nature, which is unsuitable for producing causal conclusions. Further experimental studies are needed to manipulate different robot-related factors, such as emotional interaction or voice [85]. Second, the limited number of participants and the homogeneity of the sample regarding nationality, education, cultural background, and age do not allow generalization of the present results. Most of the participants were of younger age, enjoyed university education, and came from Germany or other European countries. Further, more representative studies are needed to compare the acceptance of social robots in different, not only European societies. Third, there were more female participants than male ones, which makes it difficult to generalize the gender-related results. Fourth, this study did not involve any real-life interaction with a robot and was purely cross-sectional. However, a recent meta-analysis [86] showed that whether participants had interacted with physical robots or merely seen their photos had no influence on a wide range of measures. Fifth, the present study was conducted in an unusual period of social distancing, which prevents us from generalizing results to all situations. Multiple other related factors such as stress or depression might have influenced our results. Last but not least, only some equivocal, general, notion of social robots was referred to, without any actual live interaction with a specific robot, which could deliver a different picture. Experimental, eventually longitudinal field studies with different robots will be useful and shed more light on real-life human–robot interactions.

5.2 Robots as a third social category

Humans tend to assign human traits to other, nonhuman beings, mostly to their pets. The main objective difference between pets and robots is that pets are alive and are not machines: They have physical needs such as hunger, thirst, and even seek the attention of their owner. This difference, however, can disappear in specific cultural contexts, for example, in Japan, where due to Shinto, the native animistic beliefs about life and death, robots are perceived as living beings [87].

In the European culture, to which our participants belong, robots are not believed to have needs. They might imitate them by using natural learning algorithms and, for example, like Paro [88], repeat behaviors leading to being stroked. But, strictly speaking, these are not their real needs, they can go without this. They only serve their humans without asking for attention, food, or drinks. Of course, pet robots exist that pretend to have needs as well, e.g., MarsCat.^[2] Still, human owners know that robots do not emotionally or physically need them as a real other human beings do. One can argue that on these grounds, humans will never develop as much empathy for robots as they have for other humans or for animals because of the knowledge that robots are still machines. Even if the owner is able to ignore the fact that a social robot is a machine, it is questionable whether or not subconscious knowledge would prevent the human from having a feeling of belonging to the robot.

Thus, robots might open a third social category in addition to humans and pets – at least in some cultural contexts. So, there is no wonder that Need to belong cannot be compensated by a social robot – because it is not human – in our results with mostly European participants. Figuratively speaking, despite all the hopes, social robots are hardly masters of all the trades offered. Thus, when creating social robots for remote places or other venues with the aim to mitigate Loneliness, one should bear in mind that there are some genuinely human functions, such as being a part of a group, that cannot be substituted by robots. Future research, preferably using experimental methods, should study aspects of Loneliness as state and trait as predictors of Robots’ acceptance.