Stanford Seminar - The Rise of the Robot Waiter

Introduction and Background

• The work on the rise of the robot waiter is led by co-authors, including Samier Aran, and expands on the job characteristic model, focusing on automation rather than age (17s).
• The researcher's background is in Information Systems from a business school, but they later transitioned to the School of Information, where they integrated themselves into different communities, including the Academy of Management and INFORMS (27s).
• The researcher's involvement in various communities, including the Academy of Management, INFORMS, and the Association of Information Systems, has led to a diverse range of research interests, including autonomous vehicles and robots (50s).
• The researcher is currently involved in the ATR community and has appointments in both the School of Information and Robotics, where they play different roles, including conducting experiments and studying people (2m2s).

Defining Robots and Service Robots

• The researcher defines a robot as a physical enactment of AI, with a physical body, and distinguishes it from AI (2m34s).
• Service robots are designed to help humans perform service tasks, and the market for these robots is growing rapidly, with benefits including communication, data analysis, and productivity (2m51s).
• However, there are also challenges associated with service robots, including fear of job loss, frustration, loss of autonomy, depersonalization, and surveillance (3m32s).

Human-Robot Collaboration and the Future of Work

• The concept of right surveillance is the other side of the data analysis coin, and competition can make people feel like they're competing with robots (3m48s).
• Humans are considered useful, and going forward, there will be a human-robot hybrid workforce, with the goal of building synergy between humans and robots through collaboration (4m8s).
• The big picture of this research is to provide quality work for employees collaborating with robots, ensuring their work remains engaging, meaningful, and satisfying while promoting human-robot synergy in the workplace (4m32s).

The Restaurant Industry as a Case Study

• Restaurants are an interesting place to study human-robot collaboration, as labor is a huge component of their costs, and they are one of the largest private employers in the country with a huge turnover rate (4m50s).
• Restaurants have a surprisingly high rate of technology adoption, and they provide a unique ecosystem to understand how humans and robots work together (5m20s).
• Examples of robots used in restaurants include Flippy, Hyper Robots, M Pizza, Bella Bots, and bar robots that make drinks (6m3s).

Historical Context of Job Changes and the Job Characteristics Model

• The question of how to create an environment where people can work with robots together is not new and dates back to the transition from the pre-industrial age to the Industrial Age (7m7s).
• In the pre-industrial age, people would learn a trade from family members and be responsible for the entire process, whereas in the Industrial Age, tasks became more specialized (7m17s).
• Historically, jobs have shifted from being a source of identity and autonomy to being routine, boring, and tedious, leading to a loss of meaning and enjoyment in work (7m34s).

The Job Characteristics Model and its Components

• The Job Characteristics Model was developed in the 1970s by Hackman and Oldham to ensure employees remained engaged, motivated, and satisfied with their work (8m8s).
• The model provides a structured approach to job design, identifying five key job characteristics that promote psychological states leading to motivation, performance, satisfaction, and reduced turnover (8m36s).
• The five job characteristics are: skill variety, task identity, task significance, autonomy, and feedback (8m52s).
• Skill variety refers to the degree to which a job requires a range of skills, with higher variety leading to greater enjoyment and motivation (8m55s).
• Task identity refers to the degree to which an employee can complete a whole task, rather than just a part of it, with higher task identity leading to greater enjoyment (9m5s).
• Task significance refers to the degree to which an employee believes their job is important to the organization, with higher significance leading to greater motivation (9m22s).
• Autonomy refers to the degree to which an employee has the ability to choose how they do their job, with higher autonomy leading to greater motivation and satisfaction (9m26s).
• Feedback refers to the degree to which an employee is informed about their performance, with higher feedback leading to greater knowledge of results and motivation (9m34s).
• The five job characteristics impact three psychological states: job meaningfulness, responsibility, and knowledge of results, which in turn lead to motivation, performance, satisfaction, and reduced turnover (9m45s).

Hypotheses and Applications of the Job Characteristics Model

• The Job Characteristics Model has three overarching hypotheses, including that skill variety, task identity, and task significance promote job meaningfulness, and that job autonomy and feedback lead to increases in responsibility and knowledge of results, respectively (10m18s).
• The model is widely used to guide job design and can be used to determine the outcome of a new job on employees, as well as to identify areas for improvement in existing jobs (11m0s).
• The goal is to increase productivity while maintaining a balance to prevent making people miserable, and to apply this concept to people working with robots and designing better work technology (11m19s).

Research Questions and Studies on Human-Robot Collaboration

• There have been calls in academia to view intelligent technologies in different ways and to develop or extend theories to these new contexts, particularly in the study of providing services and how to keep people engaged when working with AI and robotics (11m39s).
• The research question is focused on what insights the job characteristic model can provide into enhancing job quality for employees collaborating with robots, ensuring that work remains engaging, meaningful, and satisfying (12m17s).
• Two studies were conducted: the first was an exploratory study of restaurant employees not working with robots, and the second was a confirmatory study of 424 restaurant employees working with robots, both using mixed-method approaches (12m30s).
• The first study surveyed employees to see what degree they think the job is going to change and what problems they see, while the second study is a random sample of a different organization with no overlap with the first study (12m41s).
• The studies used a mixed-method approach, combining quantitative and qualitative analysis, with random sampling and established scales to survey employees and gather data (13m22s).

Study 1: Exploratory Study of Restaurant Employees

• The survey asked employees questions, such as whether they think working with a robot will increase or decrease their opportunity to complete the work they start, and provided an open text box for them to explain their answers (13m42s).
• The sample of 220 restaurant employees in the first study had a gender breakdown and worked over eight hours a day, and the data met the requirements for convergence and discriminant validity (14m27s).
• The quantitative analysis found significant results for hypothesis one, except for skill variety and task identity, which had no evidence of a relationship with job meaningfulness (14m49s).
• A study was conducted to analyze the impact of introducing robot waiters on job meaningfulness, autonomy, responsibility, motivation, and performance satisfaction, as well as turnover intention, using a mixed-methods approach combining quantitative and qualitative data (15m3s).
• The quantitative analysis found that anticipated changes in job autonomy and responsibility led to increased motivation, but not performance satisfaction or turnover intention (15m15s).
• Feedback and knowledge results were found to be significant, except for turnover intention, suggesting that the model could predict behavior from a quantitative standpoint (15m29s).
• A qualitative analysis was conducted using thematic coding, bridging, and bracketing to find examples that supported or contradicted the quantitative findings (15m44s).
• The qualitative analysis found that collaboration and cooperation comments suggested that the introduction of robots would impact job meaningfulness, with some employees feeling they would have a lesser role (16m27s).
• Employees anticipated that having robots would help them with other tasks, improving their attitude and motivation, but also expressed concerns about being evaluated by the robots (16m49s).
• The qualitative analysis also found evidence that contradicted the quantitative findings, including concerns about job meaningfulness and responsibility (17m22s).
• Despite not finding quantitative evidence for the link between changes in skill variety and task identity, job meaningfulness, the qualitative analysis found evidence that employees were concerned about the impact of robots on their work (17m30s).

Emerging Themes and Perceived Usefulness of Robots

• The study found support for the overall job characteristics model, but also identified emergent themes that suggested the theory may not apply in the same way in this context (18m13s).
• Two emerging themes were identified, but not specified in the provided text (18m36s).
• A study was conducted to understand how employees perceive the usefulness of robots in their jobs, with two emerging constructs identified: perceived usefulness with the employee and perceived usefulness without the employee (18m38s).
• Perceived usefulness with the employee refers to the degree to which the robot is seen as helpful in making the employee's job better, with comments suggesting that the robot would help employees work more efficiently and effectively (18m41s).
• Perceived usefulness without the employee refers to the degree to which the robot is seen as capable of performing tasks without the employee's help, with comments suggesting that the robot would take over jobs and make employees feel depressed or outsourced (19m20s).

Extending the Job Characteristics Model and Study Limitations

• The study aimed to extend the job characteristics model to account for these two constructs in the context of humans working with robots (19m51s).
• The first study had limitations, including that it was based on anticipated changes rather than actual experiences, and that the associations may not be reliable (20m9s).

Study 2: Confirmatory Study of Restaurant Employees Working with Robots

• A second study was conducted to confirm the findings of the first study and to quantify the two emerging constructs, using a new set of employees and a different methodology (20m34s).
• The second study used an online panel and similar items to the first study, but with changes to reflect the actual experience of working with a robot (21m54s).
• The study aimed to test the hypotheses that perceived usefulness with the employer and perceived usefulness without the employer would be related to the job characteristics model (21m12s).
• The study used a confirmatory analysis to test the hypotheses and to extend the findings of the first study (20m52s).
• The study analyzed the relationship between humans and robots in the workplace, specifically in the restaurant industry, using a survey with partial least squares and covariance-based structural equation modeling, which yielded the same results (22m49s).
• The average participant in the survey had been working with a robot for at least two years, with some working up to 5 hours or more with the robot (23m13s).
• The organizations involved in the study had been working with robots for at least three to four years, indicating a high level of technology maturity (23m31s).
• The survey participants had a high school or some college education, which is consistent with the expected demographic for the restaurant industry (23m56s).
• The robots in the study performed various tasks, including cleaning, lawn work, front desk duties, and food services, while the human employees worked as front desk staff, counter servers, groundskeepers, cooks, and cleaning personnel (24m17s).
• The study defined a robot as an autonomous system and provided pictures and examples for the participants to identify the type of robot they worked with (25m4s).
• The analysis did not take into account the specific type of robot the participants worked with, but this is a consideration for future studies (25m28s).

Results and Findings of Study 2

• A study was conducted to pair the type of robot with the employee work to see if it had an impact, and the results showed a lot more support than anticipated for changes, except for turnover intention, which was not significant (25m46s).
• The study found that increases in job meaningfulness had an impact on motivation, performance, and satisfaction, but not on turnover intention (26m34s).
• The study tested three hypotheses together using a structural equation model, but the results are presented one at a time to make it easier to understand (26m52s).
• The second hypothesis found motivation and turnover intention, but no relationship between performance and satisfaction, and almost full support for turnover intention (27m11s).
• The study used self-reported data to measure the impact of robots on employees' jobs, including skill variety, job meaningfulness, and motivation (27m36s).
• The data showed that employees who worked with robots reported an increase in skill variety, job meaningfulness, and motivation, but the study did not find evidence for a negative impact on these factors (28m1s).
• The study had a limitation due to sample bias, as only employees who self-selected to work with robots participated in the survey, which may not be representative of all employees (28m47s).
• The study's results may be influenced by a humanistic adaptation, where employees may feel that working with robots has a neutral impact over time (28m40s).

Impact of Robots on Job Responsibility and Turnover Intention

• According to the characteristic model, increases in job autonomy lead to increases in job responsibility, which should decrease the chances of quitting a job, but in this case, it actually increases the chance of quitting a job, possibly due to the added responsibility of taking care of robots (29m31s).
• Working with robots increases job responsibility, which isn't necessarily a good thing, as it raises the degree of responsibility due to performance (30m1s).
• Job responsibility didn't increase performance and satisfaction, possibly due to the association of the robot's work with the human's work, with comments suggesting that the robot's integration into the workflow and help with smaller tasks are key factors (30m24s).
• The robot's work is not valued by the human, as seen in comments stating that the robot cannot affect job performance, regardless of the human's work ethic, and that customer reactions, not the robot, make them satisfied with their jobs (30m50s).
• The study controlled for variables such as attitude toward robots in general but didn't account for personal or psychological traits, such as extroversion or introversion, which may influence how humans interact with robots (31m26s).
• The study found no impact of working with robots on turnover intention, possibly due to the high-stress nature of the restaurant business, where other factors may be more significant in determining job satisfaction and turnover (32m25s).
• The type of restaurant, such as fast food or high-end, may also influence the impact of working with robots on job satisfaction and turnover intention (32m55s).
• Employees in the service industry tend to love their jobs regardless of the presence of robots, and they do not see robots as a threat to their job satisfaction (33m16s).

Perceptions and Attitudes towards Robots in the Workplace

• Some people perceive robots positively and believe that they must learn to live with the increasing presence of robots and artificial intelligence in the workplace (33m36s).
• The impact of robots on job tasks depends on the type of task being performed, and some people see the value of their job as being in front-facing interactions with customers, which robots are unlikely to replace (33m52s).
• Other factors, such as the need for money, drive turnover in the service industry, and the presence of robots does not significantly affect this (34m20s).

Impact of Perceived Usefulness on Job Characteristics and Psychological States

• The perceived usefulness of robots has a positive impact on job characteristics, and employees who see robots as useful are more likely to have a positive attitude towards them (34m57s).
• The perceived usefulness of robots does not have a significant impact on task identity, possibly because the work is loosely coupled and allows employees to manage their time effectively (35m26s).
• The perceived usefulness of robots has a positive impact on psychological states, including job satisfaction and engagement, but only when employees see the robots as useful to them personally (35m48s).
• The perceived usefulness of robots can reduce turnover intention when employees see the robots as useful to them, but can increase turnover intention when employees see the robots as useful to the employer but not to themselves (36m17s).

Discussion on Perceived Usefulness and Factor Analysis

• The discussion revolves around the concept of robot waiters and their perceived usefulness, with a focus on whether humans are involved or not in the process (36m40s).
• The survey question asked about the usefulness and success of the robot in its task, and the results showed positive loadings, which could be attributed to the robot's ability to accomplish its tasks, regardless of human involvement (37m12s).
• Factor analysis was conducted, and the results showed that the items did not cross-load, indicating that perceived usefulness is a distinct construct (38m4s).

Findings on Perceived Usefulness and Model Modification

• The study found that when people view the robot as a complement, they are more likely to be in favor of it, but when they view it as a substitute, they are more likely to be against it (39m0s).
• However, this relationship did not hold up in the second study, except for turnover intention, and the results showed that increased autonomy led to increased turnover intention and decreased job meaningfulness (39m24s).
• The study also found that job responsibility was not related to performance and satisfaction, and the researchers are still trying to figure out why (39m36s).
• The researchers suggest that the job characteristics model should be modified to include perceived usefulness with and without autonomy when people are working with autonomous systems (39m56s).
• They are currently running analysis to determine where perceived usefulness should be placed in the model, whether it should come before or after job characteristics (40m11s).
• When working with autonomous systems, it's essential to consider the different impacts on human intention and extend the model to account for this interaction (40m19s).

Designing Robots and Work for Effective Collaboration

• Designing robots to handle routine tasks can free humans for more varied responsibilities while retaining control over important decisions (40m57s).
• Ensuring robots complete discrete, identified service portions allows humans to see their contributions clearly and is an essential design consideration (41m8s).
• Designing robots involves not just designing interfaces but also redesigning the work the robot does, considering the context and the fact that work needs to be reorganized with the technology (41m14s).
• Allowing customers to rate their experience with humans and robots separately can address complaints about job performance and tips (41m30s).
• Designing robots to support rather than overshadow humans by emphasizing the human touch and customer service is crucial, and robots should not substitute human interactions, especially greetings (41m57s).
• Designing robot interfaces that offer multiple control options, allowing staff to choose their preferred interaction method, can be beneficial (42m58s).
• Some employees want to be able to adjust the autonomy level of robots to customize their interactions, but this raises concerns about standardizing processes and high employee turnover rates (43m7s).

Human-Robot Interaction and Employer's Role

• The discussion revolves around the interaction between humans and robots, specifically in a work environment, and how employers can impact employee interactions with robots to encourage collaboration (44m35s).
• A question is raised about how people's perception and interaction with robots change over time, but the current study does not explore this aspect as it only provides a snapshot of the situation (44m56s).
• To encourage humans to collaborate better with robots, one possible approach is to make the robot more social and friendly, which can be achieved through design and work organization (45m43s).
• The work design plays a crucial role in determining the level of collaboration between humans and robots, as it can either facilitate or hinder collaboration (45m59s).

Ethical Concerns of Social Robots and Encouraging Collaboration

• The idea of making robots more social and friendly raises concerns about the ethics of encouraging people to build emotional bonds with artificial entities, with some experts like Ben Sniderman arguing that it is fundamentally unethical (47m9s).
• The concern is that robots, unlike human coworkers, can record and report conversations, potentially creating an environment where employees feel deceived or misled into forming emotional bonds with machines (47m27s).
• The debate highlights the need for careful consideration of the implications of designing robots that encourage humanization and emotional bonding (47m36s).
• To encourage people to enjoy interactions with robots, employers should consider sharing the cost reduction benefits with employees, allowing them to earn more, rather than being concerned about losing their job (47m41s).

Challenges and Considerations in the Restaurant Industry

• One of the challenges in the restaurant industry is the high turnover rate, with people going to leave anyway, which may lead employers to think about accommodating employees or getting more robots (48m49s).
• The first adopters of technologies that take portions of human jobs may be employers who are not very good and are trying to slowly phase out their employees, but this factor is difficult to control for in studies (49m14s).
• The economic justification for the innovation of technology is often to reduce costs, as the return on investment is higher for cost-cutting measures than for increasing sales (50m5s).
• In the restaurant industry, labor costs are 30 to 50%, making cost reduction a significant incentive for automation (50m52s).
• The high turnover rate in the restaurant industry, with 70% of employees leaving within a short period, may lead employers to prioritize protecting themselves from turnover rather than valuing their employees (51m21s).

Further Research and Analysis

• Different types of restaurants and their sizes were analyzed, but no difference was found in the data collected from them (51m57s).
• The attitude of a head chef towards a robot in the kitchen may differ from their attitude towards a server robot, and this data has been collected but not yet analyzed (52m11s).
• A comparison between people's imagination of AI in restaurants and their actual experience of working with robots could provide valuable insights, as people's perceptions may change after working with robots (52m31s).
• In the first study, people initially thought that robots working without human involvement would be a bad idea, but the results showed that it was actually a good thing (53m3s).
• The people who participated in the second study may have self-selected to work with robots, which could mean that the sample is biased towards optimistic individuals (53m20s).
• If the same set of people from the first study had participated in the second study, the results might have been different, but it's possible that people who were unhappy about working with robots in the first study did not participate in the second study (53m35s).

Challenges in Measuring Job Loss and Automation

• Accounting for opinions towards technology in the face of job displacement is a challenge, as people who have lost their jobs due to automation may not be surveyed (53m51s).
• Job loss due to automation can be difficult to measure, as jobs that are not created due to automation are not visible, and it's hard to know how many jobs would have been created if automation had not occurred (54m20s).
• It's challenging to account for the people who would have been negatively affected by job loss due to automation, as they may not be represented in the data (54m55s).

Follow-up Studies and Methodological Considerations

• Running a follow-up study with a fixed organization could provide more insights, and it may be possible to use methodological tricks to account for the effects of job loss due to automation (55m11s).
• The idea of tracking changes in people's attitudes over time was mentioned, with a specific interest in whether negative attitudes towards robot waiters would change over time (55m18s).
• The demographics of the second study were found to be similar to those of the first study, indicating that the participants were not an entirely different set of people (55m27s).
• The similarity in demographics between the two studies was noted, suggesting that the results are comparable (55m31s).