Exploring the Impact of Robot Occupations on the Perception of the Robot-Human Border

By Soham NandiReviewed by Susha Cheriyedath, M.Sc.Nov 9 2023

In an article published in the journal Nature, researchers aimed to investigate the influence of different robot occupations on the perception of the robot-human border among participants while also considering the potential effects of age and gender.

Background

Recent advancements in robotics and artificial intelligence (AI) have led to a growing variety of robots assuming various job roles, with a rising demand for humanoid designs. This shift is driven by the belief that making robots more human-like enhances user trust and favorability. However, the concept of anthropomorphism in robots is a subject of debate.

The "Uncanny Valley" theory posits that as robots become more human-like, they become more relatable, but they are not fully accepted until they are indistinguishable from humans. Additionally, people's expectations of a robot's appearance can influence their comfort level. Different occupations also shape our preferences for anthropomorphism in robots, and age and gender play a role in these perceptions. Researchers aim to pinpoint the robot-human border, influenced by robot occupation and participant characteristics, using a classification task with morphed face images.

Methods

Three robot occupations were selected for analysis: robots, service robots, and security robots. A total of 1159 participants from Japan took part in the study, which was conducted in January 2023 and approved by the Research Ethics Committee of the University of Tsukuba, adhering to the Declaration of Helsinki. Participants aged between 30-39 and 60-69 years were recruited through Yahoo! Crowdsourcing.

The study utilized 25 images, including a robot, two male, and two female faces. The images were edited to remove hair and convert them to grayscale to ensure the focus was on facial features rather than skin color. The researchers created a set of morphed images, gradually transitioning between a robot and a human appearance, ranging from 0% to 100% human features.
Participants completed a questionnaire in which they categorized each image into one of three options: human, service robot, or security robot. The experiment was conducted in a controlled environment with specific guidelines for screen brightness, distance, and visual acuity.

Upon completing 75 categorization tasks, participants provided their personal information, including age, gender, and occupation. The collected data were then analyzed using IBM SPSS Statistics 29.0. This study aimed to shed light on how different robot occupations affect the perception of the robot-human border and how age and gender might play a role in this perception.

Results

In this study, a three-way analysis of variance (ANOVA) was conducted to examine the impact of robot occupation, participant gender, and participant age on Point of Subjective Equality (PSE) results derived from categorizing morphed face images. The data were thoroughly checked for equality of error variance, normal distribution, and sphericity. The results revealed that all main effects were statistically significant, demonstrating that each factor influenced participants' perception of the robot-human border.

The PSE, indicating the human photo proportion at which participants were equally likely to classify an image as human or as a particular type of robot, was influenced by participant age, participant gender, and robot occupation. Notably, a significant interaction effect was observed between participant age and gender, revealing variations in PSE between different age and gender groups. Post-hoc analyses further illuminated specific differences, such as higher PSE values for females in their 60s compared to males in the same age group.

The main effect of robot occupation demonstrated that participants had varying perceptions of anthropomorphism towards different types of robots, with service robots showing a higher PSE compared to robots and security robots. Participant age and gender independently influenced PSE as well, with older participants and females exhibiting higher PSE values.

Discussion

The findings highlighted the role of robot occupation; service robots, involved in direct human interactions, require more human-like features to be seen as humans. Older individuals tend to perceive images as more human, emphasizing the importance of considering generational differences in robot design, especially for older adults who prefer highly anthropomorphic robots.

Gender differences are evident, with females favoring more human-like robot features. The study also uncovers the interplay of age and gender in influencing perceptions, suggesting complex cultural and physiological factors. Despite acknowledging limitations, such as occupational stereotypes and contextual factors, this research provides valuable insights into anthropomorphic expectations, catering to diverse demographics in robot design.

Conclusion

In conclusion, this study revealed that robot occupations significantly influence the perceived robot-human border. Service robots required a higher degree of human-like features to be recognized as humans compared to robots and security robots. Older adults were more inclined to interpret images as human across all occupational contexts, in contrast to younger adults.

Notably, females displayed a stronger preference for anthropomorphic robot features compared to males. These findings emphasized the need for nuanced robot design catering to diverse occupational expectations, age groups, and gender preferences. This study underscored the pivotal role of robot occupation and participant demographics, serving as a vital reference for future research on robot anthropomorphism.