Wirtz et al. (2018) proposed a service robot acceptance model that adds social-emotional elements, such as perceived humanness and social presence, and relational elements, such as trust and rapport, to the functional core. Gursoy et al. (2019) set out a staged model in which social influence, hedonic motivation and anthropomorphism feed expectations of performance and effort, which in turn generate emotion and then willingness or objection. Lu et al. (2019) developed and validated a scale of willingness to integrate service robots into hospitality transactions. Belanche et al. (2020) organised the growing list of predictors under robot design, customer features and the characteristics of the service encounter. The shared message is that acceptance is an emotional and social judgement as much as a calculation of usefulness.

Whether a hotel robot should look human is the most debated design question. The idea of automated social presence holds that a machine can make customers feel they are in the company of another social being, with consequences for how the service is experienced (van Doorn et al., 2017). The uncanny valley hypothesis warns that near-human likeness produces unease (Mori et al., 2012). Experimental work in service settings supports the warning: Mende et al. (2019) found that humanoid service robots elicited discomfort and compensatory consumer behaviour. In the hotel context, Tussyadiah and Park (2018) compared consumer evaluations of two robots of different form and showed that intention to adopt related to different attributes for each, and an analysis of online comments on humanlike hotel robots recorded fascination mixed with unease (Yu, 2020). Murphy et al. (2019) concluded that anthropomorphism should be treated as a set of separate cues to be matched to the task.

Acceptance is a precondition for satisfaction and is not the same thing. Service quality in hotels has traditionally been assessed on dimensions such as reliability, responsiveness, assurance and empathy (Parasuraman et al., 1988), and robots perform unevenly across them. A delivery robot can be highly reliable and has no empathy to offer. Choi et al. (2020) compared perceptions of human staff and service robots and reported that human staff were rated more favourably on the quality of interaction. Tung and Au (2018) examined guest reviews of hotels that use robots and showed that guests describe their encounters in terms of embodiment, emotion and shared experience, in addition to function. Qiu et al. (2020) found that the contribution of robot attributes to the hospitality experience ran through rapport, between guest and robot and between guest and staff.

Two qualifications recur. Much of the positive response to robots reflects novelty, which fades with repeated exposure and cannot be the basis of a lasting service concept. And robots fail in ways that guests find hard to forgive when no member of staff is at hand to recover the situation. The pandemic altered the balance for a time: experimental studies conducted during it found that the salience of infection risk raised preference for robot-staffed hotels (Kim et al., 2021). Whether that preference has outlasted the emergency has not been established.

Forecasts of large-scale job loss to computerisation (Frey and Osborne, 2017) set the tone of public discussion. Service researchers have generally argued that replacement happens at the level of the task before the job. Huang and Rust (2018) distinguish mechanical, analytical, intuitive and empathetic intelligence and propose that artificial intelligence takes over tasks roughly in that order, leaving human workers to concentrate on the last two for as long as machines cannot perform them. Wirtz et al. (2018) reach a compatible conclusion: robots are likely to dominate cognitive and analytical tasks of low emotional complexity, people to retain tasks of high emotional and social complexity, and mixed teams to handle the remainder.

Hotel work fits this account unusually well, because it combines repetitive physical tasks with what Hochschild (1983) called emotional labour, the managed display of feeling that guests read as hospitality. Figure 1 arranges common hotel tasks by their emotional content and by the judgement they require. It is a reading of the literature and has not been tested as a model.

The best-known empirical study of hotel employees found that awareness of artificial intelligence and robotics was positively associated with intention to leave, and that perceived organisational support weakened the association (Li et al., 2019). The result suggests that the first effect of robots on staff may be felt before any job is removed, through insecurity, and that it can be moderated by how the employer behaves. Qualitative accounts add that staff who work beside robots acquire new duties: loading and maintaining the machines, rescuing them when they stall, explaining them to guests and stepping in when they fail. These tasks are rarely written into job descriptions or training.

A related theme is the meaning that staff attach to their work. Hospitality employees commonly describe contact with guests as the rewarding part of the job and the physical routine as its cost. A robot that takes over night-time deliveries or corridor cleaning removes some of that cost, and on this reading is a welcome colleague. A robot stationed at the front desk takes the contact and leaves the routine, and is more likely to be resented. Which tasks are automated, and who is consulted about the choice, may therefore matter more to staff than the number of machines installed (Ivanov et al., 2019).

The redistribution has an uneven social profile. The tasks most open to automation, in housekeeping, portering and room service, are held disproportionately by workers with the least bargaining power, while the tasks that remain reward language skills and confidence in dealing with guests. A hotel that automates the back of house and concentrates its remaining staff in guest-facing roles may raise the skill demands of the average job while closing the entry-level positions through which many people join the industry.