Keywords

1 Introduction

Modern technology is female. At least when it comes to the “outward appearance” of the systems or the software we are interacting with. When we drive our car, the navigation device guiding us will talk with a female voice. When we address our iPhone, Siri will answer and when we activate our smart home assistant, Alexa will take over. Both systems come with a female name and a female voice. Are they consequently recognized as female entities? Or do they possibly trigger gender stereotypes? You might be vehemently opposed to this idea, pointing out that electronic devices do not have a gender, regardless of whether the hardware or the software is considered. Consequently, they should not elicit gender ascriptions or stereotypes. In sum, these are objections that seem logically consistent. Everybody knows that Siri is not a girl, of course!

However, research on human-computer interaction has bidden defiance to common sense by challenging these rationally plausible assumptions. Back in the 1990ies, researchers analyzed users interacting with desktop computers by focusing on principles originally known from human-human interaction. They examined if, for example, social norms or stereotypes of human-human interactions will be adopted if the counterpart is not another human but a computer. Experiments revealed the users’ tendency to automatically react in a way which is comparable to the reactions known from human counterparts. Consequently, Clifford Nass and his colleagues, who were early researchers in this field, conceptualized “computers as social actors” (e.g. Nass and Moon 2000; Nass et al. 1994). Their CASA paradigm basically assumes that computers are able to elicit social responses from their human users. Among these social responses is the utilization of gender stereotypes, resulting in, for example, a computer speaking with a male voice being evaluated as more knowledgeable when it comes to typical male topics such as technology, while a computer using a female voice was perceived to be more knowledgeable when it came to love and relationships (Nass et al. 1997). Male computers were also perceived to be more dominant and had an even greater influence on participants’ decisions than female computers (Lee et al. 2000).

Research considering computers as social actors peaked in the (late) 1990ies resulting in experiments primarily focusing on desktop PCs at that time. Twenty years later, the technological status quo has changed fundamentally. Devices have become portable and their usage has become mobile. Particularly smartphones have become popular devices, which have overtaken stationary computers regarding distribution, frequency and variety of usage (Pew Research Center, 2018). Compared to the computers of the 1990ies, smartphones accompany us throughout the day, they provide support for a variety of issues and offer a number of different ways of both input and output which results in interactions integrating a variety of cues. Thus, referring to the idea of CASA with computers playing the part of the human counterpart, today’s devices might meet the requirements of a “social actor” to an even greater extent than desktop PCs ever did. Coming from both (a) the questions asked at the beginning of ascribing gender attributes and corresponding stereotypes to “talking devices” and (b) the outlined pioneer research on desktop PCs, our study focuses on smartphones as potential social actors triggering gender stereotypes. Consequently we ask the following question: Will users ascribe gender attributes to smartphones if only minimal gender cues are available? And, furthermore, do these gender cues affect the evaluation of the phones?

2 Theoretical Framework

2.1 CASA Paradigm: Computers Are Social Actors

By conceptualizing “computers as social actors” a psychological perspective is introduced complementing the primarily technological approach in the field of human-computer interaction. Hence, following the psychological questions users are analyzed regarding their cognitions, emotions, motivations, and behavior while interacting with devices (Johnson and Gardner 2007). As outlined above, the guiding idea of this paper is not new but was established back in the 1990ies. Clifford Nass and his team revealed that computers can elicit responses in their human users which were originally regarded as being exclusive for human-human interactions. In a sequence of experiments they showed that sociopsychological concepts regarding social norms of human-human interaction are also adopted when interacting with a computer. Furthermore, they could show that these reactions are not limited to certain users who are e.g. not experienced in handling computers. It is rather an inevitable reaction which Reeves and Nass (1996, p. 8) identify as “fundamentally human” and rather unconscious. Hence, if asked directly users would deny that this behavior pattern applies (Nass and Moon 2000; Nass et al. 1994).

2.2 Media Equation: Media Equals Real Life

Reeves and Nass (1996) resolve this contradiction of conscious belief and unconscious behavior by arguing from an evolutionary psychological perspective. They introduce their concept of media equation which briefly says that if a medium “communicates” with us we unconsciously react as if it was a human resulting in their oft-cited core thesis: “media equals real life” (Reeves and Nass 1996, p. 5). Thus, users react to the output of an electronic device as if this device sends these cues intentionally. As a result, these reactions are similar to reactions to humans resulting in social rules and social norms established for human counterparts. From an evolutionary point of view this is the consequence of evolutionary adaptation. The human brain and the human body are products of evolution, which were designed by natural selection to serve survival and reproduction (Hagen 2002). However, this adaptation refers to our ancestors’ world and not to today’s world. Back then, interacting with other human beings was fundamental for survival resulting in a psychological mechanism evolved to automatically and therefore resource-efficiently detect them and react to them. To do so, the following heuristic equation was sufficient: everything that appeared as a person sure enough was a person. Today, defining these cues as a sufficient condition for humanity is not applicable anymore because modern media and technology send similar cues. Nevertheless, the old mechanisms still shape our behavior. Former exclusively human cues still automatically elicit (social) reactions regardless of the entity sending them (Buss 2015; Reeves and Nass 1996).

As these processes occur automatically and independent of conscious intention or control, users would deny equating media and humans if asked. Anyhow, they follow media equation principles on a behavioral level almost unavoidably. Consequently, research referring to media equation does not implement explicit surveys but experimental approaches. Social dynamics of human-human interaction are transferred to human-computer interaction with most studies following a similar approach: Replacing the human counterpart of human-human interaction by a media device to see if the same social rules apply (Johnson et al. 2004). The requirements of this replacement are low-threshold because research revealed that even minimal social cues are sufficient to trigger social responses (Nass and Moon 2000). Presenting a text on a screen or giving the computer a name resulted in users attributing personality traits to the computer. Moreover, a computer appearing similar to the participant’s personality characteristics was rated significantly better by this specific user than a computer appearing dissimilar (Nass et al. 1995). Nass and his colleagues could also show that simply telling participants that a computer belongs to the same team as they do (signified by a color) resulted in significantly better evaluations of both performance and friendliness compared to a computer of the other team (Nass et al. 1996).

In sum, the paradigm of “computers as social actors” refers to research conceptualizing the computer as a “social” counterpart to whom social traits are ascribed and who elicits social responses in the human user (Nass and Brave 2005). The number of studies applying this paradigm peaked in the 1990ies. However, devices as well as user behavior have changed fundamentally since then, leaving the general question unanswered whether the paradigm and the resulting assumptions apply to modern technologies.

2.3 Social Categorization: Social Identity Theory and Stereotypes

Navigating through a social environment means navigating through complex social information. The social cognition approach refers to the processing of this information constituting the cognitive component of social psychological phenomena (Aronson et al. 2008). Organizing the complex social stimuli is central for a successful navigation because human information processing capacities are limited. Thus, we categorize objects to simplify information processing. Similarly, we categorize humans into groups with ourselves belonging to some of these groups. According the social identity theory, the individual’s affiliation to certain groups is one aspect of the person’s self-concept. Social identity is defined as a result of the individual’s group “membership” and the value ascribed to them. The individual develops a variety of social identities, e.g. “we, the women” or “we, the men” or “we, the scientists” (Tajfel and Turner 1979). Hence, we show group behavior, e.g. the tendency to favor our own group (the in-group) and evaluate groups we do not belong to less favorably (the out-group). Consequently, a positive social identity is maintained resulting in self enhancement (Abrams and Hogg 1988).

Stereotypes refer to social categories which encompass beliefs about a certain social group in terms of associations between the social object and certain attributes (Fishbein and Ajzen 1975). These beliefs can be held by an individual or can be shared on a societal level reflecting a more general consensus (Eagly and Mladinic 1989). Meeting one representative of the category (e.g. scientist) triggers associations and guides both judgements and subsequent behavior without checking the accuracy of these conclusions. This tendency will be quite useful because the category provides us with more information (e.g. clever) than the single representative actually gives and with more than we are able or motivated to find out. However, regarding the scientist as a representative of a group and not as an individual might result in wrong attributions if the stereotype is wrong or the representative is not typical of the group. Furthermore, neglecting individuality raises ethical questions of prejudices and discrimination. Cunningham and Macrae (2011) even identify a “pernicious social problem” concluding that cultural stereotypes influencing “thoughts and behaviour are completely unacceptable” (p. 598).

2.4 Gender Stereotypes: Warm Women and Competent Men

Gender stereotypes can be defined as cognitive structures which encompass shared social knowledge about characteristics of men and women, e.g. physical features, personality traits, interests or job preferences (Ashmore and Del Boca 1979). In contrast to stereotypes regarding nationalities or age, gender stereotypes encompass descriptive as well as prescriptive parts. Both refer to traditional assumptions about male and female characteristics, on the one hand how men and women are, on the other hand how they should be. According to this, women are emotional and sympathetic while men are dominant and determined. Violations of descriptive aspects result in surprise but violations of the prescriptive parts result in rejection or punishment (Eckes 1994). According to Prentice and Carranza (2003), stereotypes are highly resistant to change and violations only rarely result in an adjustment. Only if the invested amount of mental effort is sufficiently high and the individual is aware of the stereotype as well as motivated to avoid stereotypical thinking, will he or she be able to avoid it (Bodenhausen and Macrae 1998).

Gender stereotypes are often associated with gender roles. Referring to social role theory, Eagly et al. (2000) assume gender stereotypes to be highly elaborated resulting in a “set of associations concerning men and women” as well as “a range of overall differences between these groups” (p. 124). Hence, social role theory argues that these beliefs result from observations of different behaviors shown by men and women as a consequence of their different social roles in society, which Eagly et al. (2000) summarize as “breadwinner and homemaker roles” (p. 125). Thus, shared expectations for appropriate behavior promote different behaviors for men and women. As sex roles are associated with different activities, they require different skills. Women and men who conform to their roles regarding e.g. house work vs. paid work, consequently acquire different abilities and competencies.

Research on the content of gender stereotypes reveals two fundamental dimensions: communion (interpersonal orientation: friendly, caring), which is associated with women and agency (action orientation: competitive, individualistic), which is associated with men (e.g. Diekman and Eagly 2000; Fiske et al. 2002). Furthermore, both dimensions are linked with social roles and different levels of social status. Fiske et al. (2002) established a model of stereotype content confirming that a higher degree of warmth is ascribed to women. However, they are rated to be less competent. On the contrary, men are regarded as more competent but less warm. Hence, contributions made by men are usually met with more attention and have a bigger impact on another person’s decisions (Altemeyer and Jones 1974; Jacklin and Maccoby 1978; Propp 1995). Regarding social influence in terms of affecting another person’s emotions, cognitions and behaviors (e.g. Deutsch and Gerard 1955), men exert more influence on others than women (e.g. Lockheed, 1985) resulting in effects of conformity. Basically, conformity is elicited in two ways: the individual is convinced by a valid source of information (informational social influence) or the individual complies with another person’s expectations (normative social influence), which can be real or perceived (Deutsch and Gerard 1955; Kelman 1958; Kiesler and Kiesler 1969). Social influence is facilitated by the perceived trustworthiness of an information source thereby reflecting the reliability and expertise of the source (McCroskey et al. 1974; Fogg and Tseng 1999). Men’s higher social influence goes along with higher credibility (Pearson 1982) resulting in a generally more significant impact on another person’s decisions (Altemeyer and Jones 1974; Jacklin and Maccoby 1978; Propp 1995).

2.5 Colors and Gender Stereotypes: Blue Is for Boys, Pink Is for Girls

Gender stereotypes are deeply rooted within society resulting in gender specific sports, food, clothes or cars (Gal and Wilkie 2010). Gender-related color is another notable variation which Cunningham and Macrae (2011) trace back to the middle of the 20th century. So-called gender marketing causes a flood of products establishing the color scheme from childhood on. Children’s rooms, clothes and toys seem to be colored following the color scheme. Pomerleau et al. (1990) examined the environment of 40 children and showed that girls have more pink and colorful clothes while boys wear mostly blue, red and white. Moreover, there were more blue objects in the boys’ environments. According to Pomerleau et al. (1990), children are equipped conforming to gender before they can even express own preferences. Moreover, before they know about biological sex differences they learn that color is a sex-specific feature (Picariello et al. 1990). Thus, the association of gender and certain colors is regarded as a product of socialization that starts with birth and lasts an entire lifespan (Goddard and Meân 2009). Among other findings, Cunningham and Macrae (2011) demonstrated that associations between color and gender automatically trigger stereotypes: participants attributed more feminine features to men and women wearing pink clothes and more masculine features to men and women wearing blue clothes. Cunningham and Macrae (2011) concluded that associations between colors and gender are strong enough to create an associative pathway triggering stereotypical assumptions.

2.6 CASA Research Focusing on Gender Stereotypes

Two experiments following the CASA paradigm focused on the potential effects of gender ascribed to desktop computers. Nass et al. (1997) operationalized gender by implementing computers “talking” with either a female or a male voice. Participants interacted with a computer talking either with a female or a male voice and evaluated the performance of that particular computer afterwards. In accordance with gender stereotypes, results revealed that the male-voiced computer was perceived as more dominant and less likable than the female-voiced computer. Lee et al. (2000) received similar results. Their participants cooperated with a computer (again: male or female voice) to manage a social-dilemma situation. In line with gender stereotypes, the male-voiced computer had a greater impact on the participants’ decision than the female-voiced computer. Furthermore, the male-voiced computer was perceived to be more socially attractive and trustworthy. Regarding social identification processes, female participants were found to rather conform to the female-voiced computer and male participants rather conform to the male-voiced computer.

Contemporary CASA research focusing on today’s devices is rare with literature research revealing no studies focusing on both modern devices and potential gender stereotypes. However, e.g. Kim (2014) analyzed smartphones which acted as a specialist in an advertisement context and which were found to be rated more trustworthy. Carolus et al. (2018) showed that the evaluation of a smartphone depends on the device which is used for this evaluation afterwards. Evaluations were best when participants were asked directly by the phone which the evaluation refers to, evaluations were worse when the evaluation took place on another phone and worst if participants used their own phone to evaluate.

In summary, research regarding computers as social female or male actors is rather rare. Studies so far have focused on the effects of voices leaving open the question of effects of gender-stereotypic colors. In line with CASA research in general, participants of these first gender studies interacted with desktop PCs. Thus, the transfer to modern devices such as smartphones remains open.

2.7 Hypotheses

Based on the results of previous CASA research, computers sending gender cues should elicit gender stereotypes in their human users. These gender stereotypes were shown to affect both cognitions and behavior. Men and women are perceived differently, with men being e.g. more competent and trustworthy. Furthermore, colors seem to be gender-specific with tones of blue being associated with men and pink with women. Research also revealed these colors to trigger gender stereotypes in terms of gender-stereotypic ascriptions of characteristics. Our study brings together the experimental design implemented by Lee et al. (2000) and the idea of gender-specific colors. In contrast to the original study, we focused on smartphones instead of desktop computers. Furthermore, we replaced the male or female voice by either a pink or a blue sleeve as a minimal gender cue. In line with gender stereotypic colors, we postulate that these sleeves result in an ascription of gender to the phone. Hence, our first hypothesis is:

  • H1a: A phone presented in a blue case is rated to be significantly more masculine than a phone in a pink sleeve.

  • H1b: A phone presented in a pink case is rated to be significantly more feminine than a phone in a blue sleeve.

Lee et al. (2000) showed that “male” devices are perceived as more competent than “female” devices, which is in line with the stereotype regarding competence of men and women. Accordingly, we postulate differences regarding the evaluation of the smartphones’ competence and trustworthiness as a consequence of their gender. Furthermore, and in line with both the results by Lee et al. (2000) and social identity theory (Tajfel and Turner 1979), we postulate the evaluation of the phones to be affected by the participants’ gender.

  • H2: A male (blue) smartphone is evaluated to be significantly more competent than a female smartphone.

  • H2a: A male (blue) smartphone is evaluated to be significantly more competent by male participants.

  • H2b: A female (pink) smartphone is evaluated to be significantly more competent by female participants.

  • H3: A male (blue) smartphone is evaluated to be significantly more trustworthy than a female smartphone.

  • H3a: A male (blue) smartphone is evaluated to be significantly more trustworthy by male participants.

  • H3b: A female (pink) smartphone is evaluated to be significantly more trustworthy by female participants.

Research on gender differences revealed men to have greater social influence, while women’s attempts to influence someone are more often ignored having a smaller impact on others (Lockheed 1985). In accordance, Lee et al. (2000) showed that participants more often complied with suggestions made by a computer speaking with a male voice. Therefore, our third hypothesis is as follows:

  • H4: Participants show significantly more conformity with a male (blue) smartphone compared to a female (pink) smartphone.

3 Method

3.1 Participants

A total of 108 volunteers (54 males; mean age 24.63 years; sd = 6.74 years) voluntarily participated in the study, with ages ranging from 18 to 52 years. 83 volunteers were mainly students (76.9%) or employees (13.9%). Volunteers were recruited through social media platforms and via flyers to participate in an experimental study conducted at the University of Wuerzburg, Germany. After completion of the experiment, participants were offered the opportunity to enter a prize draw to win restaurant vouchers.

3.2 Procedure

To analyze the idea of gender in human-smartphone interaction we adopted the CASA paradigm. Hence, our laboratory study followed the basic principles established by Lee et al. (2000) presented above. Participants who had interacted either with a female-voiced or a male-voiced computer to solve a social-dilemma situation were analyzed regarding their conformity with the computer’s recommendation and their evaluations of competence and trustworthiness. In contrast to Lee et al. (2000), we used smartphones instead of desktop computers. Furthermore, we did not use voice output to manipulate the gender, but simply manipulated the color of the case the phone was presented with. The smartphones we used were Samsung Galaxy S5 models (operating system: Android) with either a blue (male) or a pink (female) sleeve.

In a 2 × 2 mixed factorial design participants were randomly assigned to either a “male” phone (blue case) or a “female” phone (pink case). A social dilemma was presented to them with the smartphone arguing for one of the two possible choices (e.g. “In my opinion Mr. D should think economically and take the risk to invest in the unknown shares. If his investment doubles, he will finally have enough money to buy the things he wants to. If the shares lose their value, he can still sell them and invest the rest in more secure blue-chip shares.”). Then, participants had to make their decision. This process was repeated five times. Afterwards, participants were asked to evaluate the smartphones regarding competence and trustworthiness as well as masculinity and femininity. This evaluation concluded with demographic information and was conducted on a separate desktop PC. Finally, participants were debriefed. By using blue and pink smartphone-sleeves we asked whether these minimal gender cues are enough to trigger stereotypical responses from participants concerning their evaluation of the smartphones as well as their behavior when interacting with them.

3.3 Measures

Conformity to the smartphone was measured using five choice dilemma items. After the smartphone had argued for one of the two options, participants reported their decision on an eight-point scale (1 = absolutely option A, 8 = absolutely option B). Competence and trustworthiness were assessed using a 10-point Likert scale derived from Lee (2008). To evaluate competence participants rated the phone regarding three items (intelligent, competent and knowledgeable), to evaluate trustworthiness participants rated the phone regarding another three items (trustworthy, reliable and honest). Masculinity and femininity were measured using two four-items scales (masculine, dominant, competitive, ambitious and feminine, sensitive, understanding, sympathetic). Again, adjectives were rated on a 10-point Likert scale.

4 Results

Regarding hypothesis 1, we checked if the color of the sleeve affected ascription of gender to smartphones. Hence, we analyzed the participants’ ratings of masculinity and femininity of the phone they interacted with. Results revealed that phones presented in a blue case were rated to be significantly more masculine than phones in pink cases, t(106) = −1,94, p = .03, d = .37, while phones with pink cases were rated to be significantly more feminine than the blue ones, t(106) = 3.14, p < .001, d = .61. Thus, hypothesis 1 can be confirmed.

Hypotheses 2 and 3 postulated effects of the colored sleeves on users’ evaluation. We conducted analyses of variance (ANOVA) to check for the main effects of “phone gender” and the interaction between “participant gender” on the evaluation of the phone (H2: competence, H3: trustworthiness) and the tendency to follow its recommendation (H4: conformity). Regarding competence, the main effect of phone gender was significant, indicating that “male” phones were rated to be more competent than “female” phones (F(1, 104) = 4.73, p = .03, η2 = .043). Analyzing female and male users separately showed that men (but not women) rated “male” phones to be significantly more competent than “female” phones, t(52) = −2.40, p = .01, d = .65. Table 1 shows the descriptive values of all hypotheses tested.

Table 1. Means and standard deviations of smartphone evaluations: competence, trustworthiness and conformity
Full size table

Contradicting hypothesis 3, ANOVA did not reveal significant results for trustworthiness, F (1, 104) = 1.16, p = .28. However, focusing on men and women separately, again male (not female) participants favored “male” phones and rated “male” phones to be more trustworthy than “female” phones (t(52) = −1,93, p = .03, d = .53).

Finally hypothesis 4 is corroborated with conformity ratings of “male” phones being significantly higher. Both female and male participants followed the recommendations of a “male” phone (M = 4.50, SD = 1.36) significantly more often than those of a “female” phone (M = 4.09, SD = 1.13), t(106) = −1.71, p = .04, d = .33.

5 Discussion

Smartphones are ubiquitous in modern society. Coming from the CASA paradigm, which postulates computers to be social actors, smartphones might to be some kind of digital companion fulfilling a variety of needs. Compared to the desktop PCs of the 1990ies, which CASA studies focused on and which were shown to play the part of a “social actor”, it stands to argue that today’s phones might meet the requirements of a “social actor” even more. However, research is rare. This study aims to reveal first insights by transferring the CASA paradigm to smartphones. Hence we adopted an experimental setting Lee et al. (2000) established for desktop PCs. In contrast to the original study, we focused on smartphones. Furthermore, we referred to gender-stereotypes. However, we did not manipulate the gender of the voice the device was talking with but focused on the color of the sleeve. Following gender-stereotypic ascriptions to colors, participants of our study interacted either with a phone in a blue or a pink sleeve and we subsequently analyzed if users would ascribe gender attributes to the phone and, furthermore, if the evaluation of the phone as well as the participants’ conformity is affected by these minimal gender cues.

Confirming our hypotheses, blue phones were rated to be significantly more male than pink phones and pink phones significantly more female, although both types of phones did not differ in terms of the content they presented. Thus, ascriptions of gender were triggered only by the pink or the blue phone case revealing the strong link between these two colors and gender. Furthermore, we could show that the color and the ascriptions they caused affected the evaluation of the phone.

In line with our second hypothesis, blue phones were rated to be significantly more competent than pink phones, thus confirming that not only colors trigger the ascription of gender but also gender stereotypes do, like men being more competent, even though the interaction was exactly the same for both smartphones. A more detailed analysis revealed specifically men to rather favor “male” smartphones, as they rated blue phones to be significantly more competent as well as trustworthy while women did not. Men seemed to be more prone to reject pink phones while the evaluation of blue and pink phones barely differed with women.

Data analysis also supports hypothesis 4. Both male and female participants followed the recommendations of a blue phone significantly more often thus showing more conformity to it. This is a clear indication for “male” smartphones having a greater social influence compared to “female” smartphones which is in line with findings from social psychology regarding gender stereotypes (Tajfel and Turner 1979).

In sum, our study could show that gender does matter, even for smartphones and even if “gender” was nothing more than the color of a phone case. Recognizing a gender seems to trigger stereotypes and deeply rooted beliefs about how men and women “are”.

While this study provides first insights into the idea of conceptualizing smartphones as social actors, there are some limitations as well as questions remaining unanswered regarding both methodological as well as theoretical considerations. Starting with methodology, the part of the experiment in which participant interacted with the phone needs to be reconsidered. The smartphone participants interacted with was not their own but rather a foreign one, which is highly uncommon as we usually only interact with our own smartphone. However, previous research could show that interacting with one’s own smartphone vs. a foreign smartphone did not make any difference regarding its evaluation (Carolus, Schmidt, Muench, & Schneider, submitted).

For further research a broader perspective regarding underlying social psychological phenomena should be considered: for this study we transferred the ascription of gender stereotypes to human-smartphone interaction, however there are many more aspects of human-human interaction that have already been transferred to human-computer interaction that warrant further research. More specifically, widening the perspective regarding other devices has to be considered as for example the concept of smart homes increases, opening a new research area.

6 Conclusion

The focus of this study was to offer first insights into the transfer of the CASA paradigm to smartphones. Conceptualizing “smartphones as social agents” indicates a heuristically fruitful approach for future analyses of human-smartphone interaction. Smartphones seem to elicit social norms in users interacting with them. The findings are considered as important for both theory and practice. The study offers multiple starting points for further research in the field of human-computer interaction thereby pointing out the importance to integrate a psychological perspective. The interaction with a device might be affected by psychological processes that appear rather illogical at first glance. Taking into account the concepts of media equation and its perspective on humans, who are a product of evolution and therefore not adapted to the use of media (devices), results in a theoretical framework which is highly relevant for the analysis of user experience. At the latest when the evaluation of the device itself is regarded, the relevance for a commercial perspective becomes obvious. Intelligent assistants (Siri) as well as smart home assistants (Alexa) are evidently designed to appear female. With regard to our findings the question arises if this should be reconsidered at least for certain areas where performance in terms of competence or trustworthiness and persuasive power are more important than the gender-stereotypic “female” interpersonal orientation associated with triggering warmth and communion.