Should Service Robots Look Like Us?

By Markus Blut, March 2021

Technology is vital for expansion of the service economy. Service robots are expected to change the way services are provided and to alter how customers and firms interact. They are defined as autonomous agents whose core purpose is to assist customers by performing physical and non-physical tasks.

The market value for service robots is forecast to reach US$699.18 million by 2023. SoftBank has sold more than 10,000 of its humanoid service robot, Pepper, since launching it in 2014.

Pepper is four feet tall and weighs around 61 pounds, and is employed by service providers in restaurants, airports and cruise liners to greet guests and help them navigate the location. It’s highly likely that robots will become more common and that customers will encounter them more in the future. While Pepper has a humanlike appearance, other robots employed by service firms, such as Lowe’s LoweBot, look more machinelike. Robot anthropomorphism refers to the extent to which customers perceive service robots as humanlike; this perception often results from the attribution of human characteristics or traits to non-human agents.

Perceived threat

Customers anthropomorphise all kinds of marketing objects, including brands, products and services. But whether anthropomorphism in service robots enhances customers’ experiences the ‘uncanny valley’ effect – an eerie discomfort and potential threat to their human identity.

Although scholars have frequently examined the impact of anthropomorphism on customers’ intentions to use a service robot, the results are inconsistent. Firms need to carefully consider how to use artificial intelligence to engage customers in a more systematic and strategic way, but they lack the support of clear management guidelines. Against this background, we have conducted a comprehensive meta-analysis of individuals interacting with different service robots, to provide some clarity on this issue.

The results suggest that anthropomorphism has strong positive effects on several marketing outcomes. It seems that humanlike perceptions are more likely to facilitate human–robot interactions, helping customers to apply the familiar social rules and expectations of human–human interactions.

This finding emphasises that it is important to consider robot anthropomorphism when understanding customer interaction in the context of this technology. Our findings highlight to managers the potential consequences of employing humanlike versus machinelike robots in service firms. Since we found mainly positive effects associated with anthropomorphised service robots, managers should not worry that an uncanny valley effect will lead customers to avoid the use of service robots.

Customer experience

Anthropomorphism is positively related to vital outcomes, including ease of use, perceived safety and usefulness of the technology, and social presence, and it does not negatively impact customer experience. However, anthropomorphism is not positively related to rapport, which indicates that service robots may not (yet) develop a personal connection with customers. Therefore, when personal connection is key to a firm’s business model, managers should employ human employees as well as service robots.

Moreover, this research examined the determinants of robot anthropomorphism. We have assessed which customer characteristics and robot design features impact the likelihood of anthropomorphising a service robot. Our research found that a customer’s general negative attitude towards robots in daily life, their need for interaction, and their age in particular, are related to robot anthropomorphism.

Managers can use these findings to identify which target customers are most receptive to the employment of humanlike robots. They should pay particular attention to these customer traits and predispositions, and we recommend they complement these criteria with characteristics of the customer, such as customer age. Managers can also use these findings to assess whether their own customer base is ready for robot services.

Furthermore, we found that different physical and non-physical robot features, such as gesture, mimicry and voice, trigger perceptions of anthropomorphism. Robots are perceived as humanlike when displaying emotions and imitating human behaviours and gestures, whereas the gaze of the robot has no effect. The results help managers to understand which robots trigger anthropomorphism perceptions.

Which robots?

Finally, our results indicate that the positive effects of anthropomorphism are contingent on the robot and service types. The results provide managers with further guidance on which human-like robots to choose when offering services to customers. Anthropomorphism had stronger effects for female robots than for non-female robots. Therefore, we suggest that service firms employ female-gendered robots (e.g. robots with a female look, name and voice). However, managers can choose any robot size, level of ‘cuteness’ and even body form, since no differences were observed in these areas.

Regarding service type, we found that the extent of a robot’s anthropomorphism can influence how it can be best used in a business. Our findings help managers to establish the service contexts in which it is most important to employ humanlike versus machinelike robots.

Managers should employ humanlike robots for critical services including ticket selling and shopping advice in the retail sector; for uncritical services, robot anthropomorphism is less important.

Similarly, humanlike robots should be employed for information-processing services such as banking and financial services. Managers in these industries have more need to invest in human-like service robots and to monitor customer–robot interaction.

More information on Professor Blut's research interests.