Unveiling the two-sided effects of robotic anthropomorphism on consumers’ experience

Artificial intelligence (AI) has achieved great advancement, and many companies have integrated AI into their applications and systems in order to cut their costs (Aksin et al., 2007). AI has made a revolution in the way companies deal with their customers (McLean and Osei-Frimpong, 2019). It is designed to perform what humans can do but with higher levels of accuracy (Dwivedi et al., 2019). It is embraced by companies with the support of data analytics to achieve better relationships with customers and higher profit margins (Evans, 2019). Moreover, it is being progressed over time and is predicted by experts to become more intelligent than humans before 2060 (Song and Xue, 2019). Our lives became more connected with AI (Bhattacharya and Sinha, 2022), and it has been commonly used in our daily life whether in business, commercial transactions or at home (Huang et al., 2019). According to McKinsey’s Global AI Survey report (2020), the rate of adoption of online and mobile banking channels across countries has increased by an estimated 20 to 50% in the first few months of the COVID-19 pandemic. Furthermore, it is expected that between 15 and 45% of customers are expected to stop visiting the branch because they got used to the standards and the level of customisation they get from AI services (Euart and Ferreira, 2020).

History shows the revolutionary changes that industries have gone through over the years, where AI was brought into the workplace, and this is when mixing between human intelligence and AI started (Mardanghom and Sandal, 2019). It is estimated that by year 2036, the “feeling economy” will become more important than the “thinking economy”, where AI will not only perform physical tasks but will also have the ability to show empathy, emotions and interpersonal relationships (Huang et al., 2019).

AI has become advanced to the level of using human-like characteristics to non-human machines, which leads to the concept of anthropomorphism (Epley et al., 2007). Anthropomorphism is known as “the attribution of human traits, emotions, and intentions to non-human entities or agents” (Guthrie, 1993; Duffy, 2003, p. 180). The term “anthropomorphism” has been linked with “artificial intelligence” in earlier literature; however, it was also noticed that sometimes the word “personification” was used when denoting the concept of “anthropomorphism” (Uysal et al., 2023). It has been already announced by Hanson Robotics in 2021 that anthropomorphised robots such as Sophia and Hanson will be produced on a large scale so they can be used in interaction with customers (Vinas, 2021). Anthropomorphised robots facilitate the social communication and human–AI interaction while delivering a service (Wang and Lin, 2017). According to the European banking federation (2019), robotic advisors offer a great customer’s experience since they are able to suggest different investment solutions that are customised according to the client’s needs and expectations. They can be perceived in various ways by consumers depending on their coping abilities (Belanche et al., 2020). Chatbots have been anthropomorphised to the extent that consumers find it challenging to differentiate between humans and chatbots (Elsharnouby et al., 2025; Pizzi et al., 2023). In the current digital era, service robots have been deployed to deliver customers service replacing human interactions (Mehmood et al., 2024).

There is no doubt that AI anthropomorphism is a multifaceted system. According to Musk (2014), AI technology can also be more dangerous than nukes. In addition, several researchers fear that these machines may spread threats and risks to humans (Yaninen, 2017). The idea of using anthropomorphised robots in the service sector has become a controversial topic due to the sophistication of applying human-like characteristics to non-human entities. Although anthropomorphism has been receiving a great attention in research, there is still a lot of misrepresentation and miscommunication about it (Barrow, 2024). Moreover, despite its growing interest in marketing, significant gaps remain in understanding its antecedents and behavioural outcomes. Existing literature explains that there are mixed findings about the impact of AI anthropomorphism on user behaviour. Therefore, in service literature, there is no agreement whether human-like robots facilitate or reduce the intention to use them (Blut et al., 2023). A technology that adopts a design with human-like features can promote the acceptance and usage as its more engaging (Kuo, 2011). Some studies demonstrate that anthropomorphic features enhance social presence and ease of use, therefore offering a positive customer’s experience. For instance, a meta-analysis demonstrated that anthropomorphism evokes a sense of social presence, enriching social interaction (Blut et al., 2023). Similarly, the results by Troshani et al. (2021) revealed that anthropomorphic design cues in chatbots enhance brand attitudes through increased social presence. Furthermore, prior studies indicate that in general anthropomorphic design features have a positive effect on attitudes and perceptions towards robots (Roesler, 2021). Conversely, other researches indicate that anthropomorphic features generate feelings of discomfort or creepiness among users, resulting in deterioration in customer’s engagement. For example, it has been noted that uncanniness moderates the effect of perceived anthropomorphism on social presence, impacting brand attitude (Troshani et al., 2021). Moreover, the available literature is still scant when it comes to customer’s experience due to the rapid changes in consumers’ perceptions. Accordingly, the advancement in this technology requires further understanding for how consumers perceive it, in addition to deeper understanding for their overall experience. Besides, the cultural and individual differences that shape the interactions with anthropomorphised AI require an interdisciplinary study in order to grasp the nuanced consequences of integrating AI into human life everyday decisions.

There are various theoretical and conceptual frameworks that have been adopted in the area of AI anthropomorphism. The technology acceptance model (TAM) (Davis, 1989) suggests that the usefulness of technology and its ease of use promote the intention to adopt the technology. In addition, the social response theory (Reeves and Nass, 1996) suggests that user satisfaction and engagement are higher when machines have human-like characteristics, specifically in financial services. Trust theory emphasises that users have more compassion towards agents with human-like robots (Lankton et al., 2014). Casa paradigm suggests that anthropomorphised robots are not perceived as programmed computers and users naturally relate to them as social actors replacing employees (Kim et al., 2019). On the other hand, one of the critical issues in the area of anthropomorphism is the uncanny valley theory (Mori, 1970), which explains that users feel discomfort when they interact with robots that extremely look like humans, especially if they fail to perform up to the human standards (Mori, 1970). Similarly, in a context like banking, users have a negative experience when they deal with human-like agents (Luger and Sellen, 2016). Studies indicated that when service robots exhibit human characteristics, customers unconsciously deal with them as if they are humans (Song and Kim, 2020). This encourages users to have high expectations from humanised robots which can lead to disappointments (Calo, 2015). Last, the parasocial interaction theory clarifies how the relationships between people and technologies have developed (Whang and Im, 2021).

Moreover, AI has the ability to understand feelings, motives, intentions, emotions and expectations and they can interact socially (Hassani et al., 2020). A new wave of AI, which is expected to exceed the human intelligence levels, is expected to start by the 2030s (Fouse et al., 2020). Prior literature suggests that when it comes to providing a service, consumers look for human interference if technology failed to fulfil their request (De et al., 2015). Recent studies showed that consumers perceive anthropomorphised robots as competent when they tend to have high human-like language (Nguyen et al., 2023). Nowadays, with the huge improvement in AI, robots do not only look like humans but can also acquire a nationality and a legal citizenship like Sophia, the robot (Pagallo, 2018). Previous studies propose that both the functional abilities of service robots along with their social capabilities determine how consumers react to the interactions with them (Huang et al., 2019; Tung et al., 2018; Wirtz et al., 2018). Studies suggest that anthropomorphism is influenced by social cognition, familiarity with AI and the design of the features (Nass and Moon, 2000). Thus, the research gap below has been developed.

Furthermore, intelligent robots are employed to provide different support services instead of human workers (Belanche et al., 2020; Huang et al., 2019). For instance, the World Health Organization introduced an AI influencer named Knox Frost during the COVID-19 pandemic, who used to make health announcements and promote hygiene practices. It is expected that AI machines will have the cognitive abilities to perform as humans and even more in 40 years (Song and Xue, 2019). As stated in the mission statement of OpenAI, a “non-profit research company” in 2015, “It’s hard to fathom how much human-level AI could benefit society, and it’s equally hard to imagine how much it could damage society if built or used incorrectly”. This raises the question: If human-like robots are used in some situations instead of real humans, will this enhance the service? And will it make it easier for consumers to accept it? This research gap leads to the research question below:

Privacy plays an important role when it comes to the use of digital platforms and devices (Horppu, 2023). Previous studies show that consumers have concerns about the privacy of their interactions which raise some trust issues towards the security of data, and therefore the resistance to provide their personal details (McLean and Osei-Frimpong, 2019). Moreover, prior literature acknowledged that gender could have a great impact on how consumers perceive AI anthropomorphism (Liu and Tao, 2022). Given that female-gendered AI assistants are predominant in the creation of humanised AI assistants (Hermann, 2022) such as: Alexa, Siri, Sophia, etc. It has been demonstrated that users can form emotional attachment towards anthropomorphised robots (Turkle, 2011). This research also aims to explore the antecedents of using anthropomorphised robots and how it shapes consumers’ behavioural outcomes. Thus, the research question below has been raised:

This study adopts exploratory qualitative research in the form of interviews since the research area of AI anthropomorphism is still not fully discovered. Before starting the interviews, the participants were asked to watch a 5-min video of Sophia, the humanoid robot, being interviewed by Ross Sorkin at the future investment institute panel in Saudi Arabia (https://youtu.be/S5t6K9iwcdw), so they can relate to the topic. The researcher has chosen semi-structured interviews with predetermined questions in order to provide an optimum coverage for the subject. All interviewees were asked identical questions, ensuring better comparability of the results. The interview included open-ended questions in order to allow participants to freely express their views, enabling researchers to gain deeper insights into their experience while staying focused on the research topic. Each interview lasted for an average of 45 min. The interview questions are divided into five parts. Demographic questions (name, age, gender, monthly income and educational level) have been captured at the beginning so the researcher can gain information about the background of the participants. The second set of questions aims to explore the term “AI anthropomorphism” which has achieved new eminence lately. The third set of questions aims at exploring whether the antecedents are using it. The fourth set of questions involve looking into the behavioural outcomes of such interactions, events and translating them into actual outcomes. The final part intends to address future recommendations for a better experience or additional services (see Table 2 for the interview questions).

The interviews were done with 18 interviewees. No new information related to the experience of the respondents emerged during the final interviews which indicated data saturation (Saunders et al., 2018). The sample type chosen for the interviews is judgmental or purposive sampling method because the researchers were looking for respondents with specific criteria to participate in the interviews. Respondents were selected based on two characteristics: first, they should be above 18 years old so they could have bank accounts. Second, they should have previous experience related to the topic being researched (Groenewald, 2004). They should have performed services through AI (tech-savvy customers) without interference with employees so we can understand their overall experience. Thus, anyone who is only aware of AI but has not tried it was excluded. Thus, the researchers used non-probability sampling.

Table 1 presents the demographic profile for each participant being interviewed so we have information about the background of the respondents.

Data collected from the qualitative research were analysed using a thematic analytic approach due to its usefulness, adaptability and flexibility in the identification of repetitive themes which can provide detailed and rich data with broader meanings (Braun and Clarke, 2006). Thematic analysis is suitable when the researcher aims to understand the context and depth of human experiences across an entire interview (Vaismoradi et al., 2013). Furthermore, thematic analysis has been chosen over the content analysis because the researcher is seeking a purely qualitative data with no need to quantify data or count frequencies (Gbrich, 2007). In this study, the researchers followed the approach proposed by Creswell (2009) in the thematic data analysis. The researcher conducted the transcription manually by writing notes and defining potential patterns and codes (Braun and Clarke, 2006). It was conducted without the use of transcription tools or outsourcing this step as it provides an opportunity for a better depth and breadth (Braun and Clarke, 2006). Moreover, the researcher made notes during the interviews to capture important words or comments, and some verbatim comments of the interviewees have been documented and added to support relevant text (see Table 2) (Ereaut, 2002). The researcher read the transcripts and notes captured in order to make preliminary observations about the content. Then, the raw data was prepared, sorted and categorised based on the responses of the interviewees to generate initial codes. Data were then retrieved and organised into codes based on their related characteristics (Coffey and Atkinson, 1996; Saldaña, 2015) which enabled to come up with major themes, and some have been categorised into sub-themes.

Table 2 provides examples of the respondents’ responses on each question and reflect them on the research questions being addressed along with a grouping of the thematic codes.

Researchers grouped consumers’ perceptions about AI anthropomorphism into four main themes: Perceived anthropomorphism, individuals’ emotional responses to anthropomorphism, emotional intelligence and gender.

AI has made huge improvements, and it can now act similar to humans in so many ways. Respondents indicated that humanised robots possess many features derived from human characteristics such as their external features, facial expressions and body language yet their eye contact is still fake. They have a lot of human-like technical qualities such as conscientiousness, knowledge and competence which allow them to respond objectively based on the data embedded in them. Nevertheless, they lack some human characteristics such as creativity, empathy, compassion and neuroticism.

Even though anthropomorphic robots have a lot of human characteristics, all interviewees believe that interacting with them is not comparable to engaging with real humans. Besides, they believe that they still need improvements in their intonations, body language, eye contact and emotions. They are monotonic, and they do not reply spontaneously.

When the interviewees were asked about their feelings towards robots with anthropometric direction, only a few participants showed negative emotions describing them as fake or silicon-like. In addition, very few participants had pure positive emotions towards anthropomorphised robots such as excitement that encouraged users to try them. On the other hand, the majority had mixed feelings such as impressive yet scary, interesting yet scary, exciting but creepy, etc. They felt impressed with the technology at the beginning but then felt creepiness.

This indicates that most individuals experienced negative emotions towards anthropomorphic robots such as creepiness and discomfort, even if their initial reactions were positive. These feelings were stronger when robots performed tasks requiring human-like intelligence. This supports the “uncanny valley theory” (Mori, 1970) and aligns with Gursoy et al. (2019), suggesting that human-like robots can provoke negative emotions that may reduce users’ willingness to engage with them. Hence, AI and robotics can be a repulsive idea for some people, even to those who were initially enthusiastic about them.

Although humanised robots are designed with communication and emotional features, most respondents do not expect them to show any emotions. They believe that emotional intelligence is still too complex for robots. They are perceived as logical decision-makers rather than emotional beings.

On the other hand, only a few participants expected humanised robots to react and show emotions and feelings because they perceive them as highly advanced with greater potentiality.

Anthropomorphised robots are currently designed to mimic human conversations; however, most respondents believe that they lack genuine emotional interactions. Even the very few participants who expect them to show emotions, it is not because they are actually emotional, but because they have illusions that humanised robots are an expression of a great technological advancement so they should be able to act humanly. Therefore, AI can replicate human actions, but it cannot express emotions. Respondents believe that emotional intelligence in robots may be possible in the future, but for now, it remains unlikely.

It has been recommended by previous studies to consider gender bias in future research and develop perspectives about how it might affect consumers’ interactions and perceptions of AI (Diederich et al., 2022). Studies recommend that firms can adopt robots that have a female look, name and voice rather than male robots (Blut et al., 2023). However, when participants were asked if they care about the gender of the AI-robot, only one male was gender biased, since he felt disturbed by the female voice. He preferred to have a male voice in AI-powered services.

Furthermore, three female participants said that they never thought about it but generally, they expect a female voice by default as the female voice is commonly used in the interactive voice responses (IVRs). However, most participants were neutral about the gender of AI voices, prioritizing competence and service quality instead. In general, according to the vast majority, making AI sound more interactive, warm and human-like is more important than its gender.

The following section addresses RQ2: “How can the anthropomorphic features in AI bridge the gap between human and machines? And how can this undermine the richness of human interactions?”.

When participants were asked about the services they perform in an automated branch (bank), they reported using AI for simple tasks such as issuing a replacement card or opening an account. The majority preferred AI for routine tasks but looked for human interactions when tasks are complex or require engagement. Preferences for using AI largely depended on two factors: Users’ personality traits (tech freak, technophobe, introvert, etc.) and the complexity of the service performed. This goes in line with the results of Poushneh, (2021) that shows that users’ personality traits have an influence on their attitudes and behaviours and also influence their readiness to interact with the brand (Shah et al., 2024).

When respondents were asked whether they prefer to visit an automated branch or a physical branch in case they have a problem, all respondents even those who are introverts or tech-savvy preferred visiting a physical branch seeking human interaction.

Respondents are not only seeking efficiency but also they are looking for emotional understanding during issues. Furthermore, consumers are expecting to gain emotional and hedonic benefits throughout the overall experience (Collier and Barnes, 2015), which AI still lacks. It is concluded that customers prefer human interaction in complex situations.

Lee and Chen (2022) demonstrated that intelligent or smart chatbots are effective in routine tasks. Similarly, most respondents preferred chatbots over human agents for loan information, viewing it as a simple and straightforward task.

Although the participants considered gathering information as an easy task, they mentioned that the final decision to take a loan is a risky and a complicated decision. Thus, they prefer human interactions in this situation.

Participants’ responses to the second research question indicate that while anthropomorphised AI has replaced many routine tasks, it still lacks the emotional intelligence and communication skills needed for complex situations. This supports Immonen et al. (2018), showing that users still prefer human interactions in challenging situations. Consequently, the results show that anthropomorphised robots have not closed the gap between AI and human interaction. However, they have allowed employees to focus on more complex and emotion-driven tasks which should enhance the overall customer experience.

The analysis of the qualitative data revealed the key antecedents of AI anthropomorphism. They include personalisation and waiting time.

Data-driven personalisation has always been a paradox. It can either be an effective or an ineffective marketing strategy. Although there is evidence that greater personalisation improves response rates, it can make consumers uncomfortable which could lower response rates (Aguirre, E. et al., 2023). The interviewees were asked to share their thoughts about receiving messages of offers based on their past behaviour or previous preferences, and to tell whether this enhances their experience or not. The majority of the respondents find that personalisation is useful; however, there was a concern about data privacy, whereas fewer participants did not like personalisation and believe it is annoying or a repetition of the same service besides feeling exposed.

Regardless of consumers’ views on personalisation, the researchers concluded from the responses that it often triggers privacy concerns. This aligns with the Orwellian’s perspective that proposed the concept of the “Big Brother effect” where individuals feel constantly monitored and controlled by the technical power of entities, leading to resistance towards using smart technologies (Mani, 2021).

When the interviewees were asked whether waiting in a queue is a factor that affects their decision to go to the bank or not, all interviewees agreed that waiting time significantly influences their decision to visit the branch. They would prefer automated branches to avoid long queues.

Based on the qualitative data analysis, a number of behavioural outcomes related to AI has emerged. They include intention to adopt AI anthropomorphism, customer retention, participation intention and switching behaviour.

When respondents were asked whether humanised robots promote their acceptance, the majority said that they do promote their acceptance because they grab attention, make people excited and sound more real.

Most participants felt that anthropomorphised robots accelerated technology acceptance, while less participants found them unsettling as they evoke fear and discomfort.

When respondents were asked if they believe that their bank has an edge over other banks by having fully automated branches, the majority of the respondents believe that fully automated branches do not give banks a competitive edge. Consumers prioritise service quality, data security, accessibility and customer service. The findings offer suggestions to industries, companies or financial services to improve their online services as it is more important for the customer experience than having automated branches.

Participants were asked whether it matters to them to participate in the implementation decisions of AI. The majority of the respondents appreciated being involved in AI implementation decisions. They believe that it enhances their experience, loyalty and engagement. However, fewer participants viewed it as unnecessary focusing more on the quality of service offered.

Finally, participants identified the key challenges related to the use of AI, including system failures, poor Internet connection or privacy concerns such as hacking or spamming.

Accordingly, this paper proposes the following AI taxonomy, as shown in Figure 1, that shows the antecedents and behavioural outcomes for perceived anthropomorphism. These variables have been extracted based on the responses of the interviewees.

The study aims to focus on the conceptualisation of AI anthropomorphism, its antecedents and behavioural outcomes. Regarding the first research question, the interviews showed that anthropomorphised AI shares a lot of human characteristics. It can be competent, intelligent and smart. However, the majority of the respondents perceive anthropomorphised AI as a one way of communication, unlike previous literature that showed that AI is capable of having two-way conversations with customers (Hoffman and Novak, 2018). It lacks emotions and social interaction skills leading to “parasocial interaction”. In other words, respondents agreed that anthropomorphised AI still does not offer a two-way reciprocal communication or behaviour. In addition to this, the responses support the suggestion by the uncanny valley theory, as the majority of the participants elucidated that the personification of a non-human entity in general and human–AI interaction in specific generates mixed feelings starting with excitement and ending with creepiness, discomfort and fear when they are too much human-like (Murphy et al., 2019). There is always a dilemma when designing a humanoid robot whether it should be a male or female. Previous studies discussed the importance of gender in humanoid robots and there was a gender-bias in favour of females, because a female voice gives more warmth, provides better feelings and shows more empathy than males (Rust and Ming-Hui, 2021). Additionally, it was mentioned that AI agents with female characteristics are preferred in the current stage of AI deployment (Alabed et al., 2022). Unlike previous studies, the findings show that users do not consider gender as an important factor in designing humanised robots. Users focus on how natural a robot reacts to their queries in order to decrease the feeling of creepiness and how efficient it is to get the service done smoothly with no complications regardless of whether the robot is a male or female. Moreover, the type of service performed moderates the relationship between perceived anthropomorphism and behavioural outcomes. Respondents mentioned that if the service required is easy, they will prefer to do it through AI. Whereas, if the service is complicated, or in case of escalations, they prefer to deal with humans with no hesitation because humanised robots have not yet reached the level of having emotional intelligence to deal with complaints.

In regard to the second research question, the qualitative findings demonstrated that AI cannot replace human interactions in complex matters when consumers look for human input. From users’ point of views, anthropomorphism has induced customers’ intentions to accept humanised robots, but all of them were sceptical whether they are able to replace humans in cases of escalations. The responses highlighted that human–AI collaboration has enhanced the overall customer’s experience and increased their satisfaction but has not actually replaced humans. They are working complementary to each other not as a replacement to one another.

The main antecedents of using AI can be categorised into two main themes: personalisation and waiting time. Existing literature shows that customers are sceptical about the use of technology in financial services specifically because they are less tolerant to errors or unnatural behaviours related to their finances (Luger and Sellen, 2016). However, the results in this study show that human-like robots did not trigger concerns about privacy or data confidentiality. Personalisation raised a concern about the “big brother effect” referring to consumers’ fears of being watched and vulnerable, which in turn raised concerns about the privacy and confidentiality of users’ information. It can be concluded from the qualitative findings that anthropomorphised AI can make the customer’s journey more fun and would enhance the overall customer’s experience. Moreover, it is interesting to see how humans are able to design robots that can mimic human speech, which can have an impact on the acceptance of robots.

The behavioural outcomes have declared positive outcomes resulting from AI with anthropomorphic features such as enhancing customer’s experience, customer’s retention, increasing its acceptance among users and trust to provide personal information which will in turn increase their adoption intention, while also triggering some negative behavioural outcomes such as emotional creepiness and perceived vulnerability through personalisation.

Finally, the results of this paper contradict with the concept of “AI exceptionalism” which demonstrates that AI has epistemic expertise and capabilities to perform tasks successfully therefore it can have more authority and control over humans (Ferrario et al., 2024).

The interview questions were related to the banking sector, so the findings could be different if it is applied on other service industries. Gender showed no importance in this study, even though it showed significant outcomes in other studies that were done in other service sectors. For instance, when it comes to healthcare services, results showed that female AI agents resulted in more positive outcomes (Liu and Tao, 2022). On the contrary, male AI agents were preferred in the sales context (Liew et al., 2017). Thus, future research should consider that gender might have different outcomes depending on the type of service provided. Besides, future research should consider the long-term psychological implications for dealing with AI such as the parasocial relationship which is having a bond and emotional attachment to AI, since it will have an impact on the social well-being across various contexts.

In general, attitudes towards service robots are shaped according to the culture, where some countries may have positive attitudes towards service robots unlike others (Bartneck et al., 2007; Belanche et al., 2019). Most studies about the importance of AI anthropomorphism are conducted in Western contexts (Bartneck et al., 2007; Li et al., 2016; Park and Kim, 2023). However, this research area is very limited in emerging countries (Idrees et al., 2020; Jain and Gandhi, 2021). Users from different cultures may have diverse perceptions to AI with human-like features (Nomura et al., 2006). For instance, studies explained that consumers in Canada have better tolerance to accept anthropomorphised services, while consumers in India are reluctant towards using technologies with anthropomorphic features (Mehmood et al., 2024). Therefore, this paper contributes to the literature by understanding how cross-cultural differences can influence the perspectives about AI anthropomorphism. Thus, understanding how users in Arab countries perceive human-like robots adds to the literature and helps gain insights about the differences between each culture. Attesting to the prominence of AI anthropomorphism, this paper offers a more fragmented insight as it integrates positive and negative possible behavioural outcomes in a single model. Unlike previous studies that demonstrate the influence of AI anthropomorphic features from one direction either positive or negative, the interviews focused on both cognitive (usefulness, ease of use) and affective aspects (feelings, emotions), offering a better comprehensive understanding for the concept of anthropomorphism. Furthermore, this paper offers many variables related to the antecedents and consequences of AI anthropomorphism which can provide various suggestions and recommendations for future research topics. This paper used a qualitative approach to dig deeper into consumers’ perceptions while the majority of previous literature conducted a quantitative approach.

Companies are now investing in AI in general and in service robots in specific, aiming to establish collaboration between humans and robots for the purpose of completing a task effectively and maximising productivity (Chacón et al., 2021). Therefore, this paper has several implications for marketing managers so they can successfully employ anthropomorphised service robots. Practitioners and developers should consider the results of this study while developing robotics in the service context. First, users are still not ready to deal with humanised robots in all aspects and still perceive them as scary and creepy, so practitioners should make robots more friendly, less monotonic, more spontaneous, showing more empathy and they should enhance their intellectual abilities to handle complex situations. Even if humanised robots are able to serve customers, they cannot fully replace employees, and human agents should be present for further assistance. Thus, consumers believe that they can be employed to routine or basic tasks as they prefer to deal with humans in complex matters. Besides, users believe that humanised service robots have been improved in the way they look, yet they do not have the capabilities of a human service provider. Generally, marketing managers should allow the presence of AI in easy tasks because users expressed their satisfaction in how this saves their time, but they will be quite frustrated if they do not find a human service provider to deal with when they have a problem. Gender does not matter in the development of service robots, so a service robot can either be a male or female; however, their reactions and capabilities to solve the matter are the most important. To sum up, the results show that customer’s participation is an important factor in developing new technologies, so it is recommended for companies to engage customers and listen to their feedback for better outcomes. Accordingly, companies should focus on and invest in their online platforms and avail more through AI.