Search

The implementation of new robotic technology at workplaces is oftentimes accompanied by social and organizational change processes. A new context-specific questionnaire was developed as a diagnostic tool to measure workers’ attitudes toward mobile cooperative manufacturing robots to provide a basis for managerial decisions and interventions—the “Attitudes toward Cooperative Industrial Robots Questionnaire” (ACIR-Q). Two samples, an online sample of 355 German manufacturing workers and a field sample of 201 workers from 4 local manufacturing companies were collected. For a large item pool, exploratory and confirmatory factor analysis was used to identify the attitudinal factor structure. Data showed a combined affective-behavioral factor and two cognitive factors on task-related and social-related beliefs. Based on this, the 12-item short scale ACIR-Q was derived using ant colony optimization. As attitudes can also be interpreted as networks of evaluative responses, network analysis was used for further insights. The small-world network structure (high clustering and connectivity) allows to hold complex attitudes and centrality measures indicate the most influential evaluative responses. Additionally, we explored relationships between workers’ attitudes and interpersonal variables (perceived competence, perceived control, and general self-efficacy), as well as social/organizational variables (trust in management, support climate, job insecurity and job characteristics). Based on the results, practical implications are suggested to improve workers’ attitudes. Version of record

Understanding the recommendations of an artificial intelligence (AI) based assistant for decision-making is especially important in high-risk tasks, such as deciding whether a mushroom is edible or poisonous. To foster user understanding and appropriate trust in such systems, we assessed the effects of explainable artificial intelligence (XAI) methods and an educational intervention on AI-assisted decision-making behavior in a 2 × 2 between subjects online experiment with N = 410 participants. We developed a novel use case in which users go on a virtual mushroom hunt and are tasked with picking edible and leaving poisonous mushrooms. Users were provided with an AI-based app that showed classification results of mushroom images. To manipulate explainability, one subgroup additionally received attribution-based and example-based explanations of the AI’s predictions; for the educational intervention one subgroup received additional information on how the AI worked. We found that the group that received explanations outperformed that which did not and showed better calibrated trust levels. Contrary to our expectations, we found that the educational intervention, domain-specific (i.e., mushroom) knowledge, and AI knowledge had no effect on performance. We discuss practical implications and introduce the mushroom-picking task as a promising use case for XAI research. Version of record

HOXAI is an interdisciplinary project by the LIT Robopsychology and the Visual Data Science Lab at the Johannes Kepler University Linz in Linz, Austria. The focus of this project is interdisciplinary research on explainable artificial intelligence by conducting a row of user studies. The project is funded by State of Upper Austria and the Austrian Federal Ministry of Education, Science and Research via the LIT – Linz Institute of Technology. For more information visit: https://www.jku.at/en/lit-robopsychology-lab/research/projects/hoxai/. In this OSF component you can find additional materials from study 1.

Based on the theoretical model that the tendency to anthropomorphize non-human agents is influenced by motivational factors (Epley, Waytz, Akalis, & Cacioppo, 2008), we test the correlation between loneliness and the anthropomorphism of an industrial robot. As the results of our previous field study at an art gallery in the city of Graz (Austria) were not sufficiently robust (results unpublished), the correlations between anthropomorphism ratings of visitors for a non-humanoid robot arm that is presented in the art installation “Contact” (Gollob & May, 2021, 2022) and loneliness ratings should be tested again in a different setting and with partly different methods. The art installation “Contact” in which people can spontaneously get in touch with a non-human-looking robot will be presented at the museum of media art Ars Electronica Center in Linz (Austria). Different measures will be used for loneliness and partly different measures will be used for different aspects of anthropomorphism (see variables section). We additionally collect observational data, i.e., if visitors touch the window in order to have “contact” with the robot.

The AI Forest installation is an indoor forest and interactive play environment that was first presented at the Ars Electronica Festival 2021 in Linz, Austria. In this project component you can find illustrated descriptions of the installation setup, a snippet of the sound concept, and the code to the game to replicate the study. If any of the materials is used, please cite this project (see citation information) and the related paper ().

In this laboratory experiment in a local museum, the effect of introductory information on the evaluation of a collaborative robot (e.g., trustworthiness, predictability and understandability) as well as on the performance of the human-robot team in collaborative game tasks (time to completion and task success) will be explored. We use a mixed reality game specifically designed for this study called "CoBot Studio Clean-Up – Deep Space Edition" in which human participants have to collaborate with a real industrial mobile robot on three different tasks in a fictional virtual reality underwater research station presented in a large-format projection space allowing for highly immersive 3D-experiences. The mixed reality game "CoBot Studio Clean-Up – Deep Space Edition" is a follow-up to the virtual reality game “CoBot Studio Clean-Up” and the mixed-reality study is a replication and extension of the corresponding virtual reality study in a mixed reality contrary to a virtual reality scenario.

When users interact with autonomous systems based on artificial intelligence (AI) for decision-making it is often unclear how the AI-based system comes to its results. However, researchers as well as policymakers are increasingly demanding that such systems be explainable so that human users can understand how they come to their results. It is expected that such an explainable AI (XAI) will improve team performance and avoid users’ overtrust or distrust in AI-based systems. A group of XAI methods to achieve such an appropriate level of understanding is visual explanations. But which of the wide variety of visual XAI methods work better than others in which situations? In combination with another study (see related OSF project), participants will be introduced to an online game with a virtual mushroom hunt and will be tasked with 1) identifying pictures of mushrooms as edible or inedible/poisonous and 2) picking only edible mushrooms but leaving inedible/poisonous ones. For this task, participants will additionally receive recommendations from an AI-based app showing classification results of mushroom images. In order to understand the effects of different visual explanations, the AI-based app’s interface will be manipulated in terms of explainability with participants receiving 1) no explanations, 2) example-based nearest neighbour explanations, 3) attribution-based Grad-CAM explanations, and 4) concept-based semantic explanations (network dissection). As dependent variables we will assess 1) participants’ identification performance, 2) the correctness of their picking intentions for each mushroom item (i.e., mushroom-picking performance), 3) users’ trust in the system after playing the game, 4) the users’ evaluation of the system, and 5) their intention to use the system. Furthermore, we investigate the relationships of performance and trust with AI knowledge, domain-specific knowledge (mushroom knowledge), as well as propensity to trust.

In a laboratory experiment the influence of room size and mental workload on human comfort distance towards an industrial robot prototype is investigated. It is hypothesized that this effect is mediated through arousal / stress and perceived control over the situation.

There are many situations in which human users consider the recommendations of an artificially intelligent assistant system for decision-making, even in high-risk decision tasks. But who do we blame in cases of wrong decisions, and why? In combination with another study (see related OSF project), participants will be introduced to an online game with a virtual mushroom hunt and will be tasked with picking only edible mushrooms but leaving poisonous ones. For this task, participants will additionally receive recommendations from an AI-based app showing classification results of mushroom images. In the study, the AI-based app’s interface will be manipulated in terms of explainability with participants receiving 1) no explanations, 2) example-based nearest neighbour explanations, 3) attribution-based Grad-CAM explanations, and 4) concept-based semantic explanations (network dissection). After the game, participants will read a text vignette describing a scenario in which a friend gets poisoned by a mushroom that was a gift from the mushroom hunt and declared safe by the AI. Subsequently, we investigate the amount of blame attributed to various involved agents (i.e., oneself, the AI, the developer, and the friend) as well as a justification of the attributed blame. Further, we investigate various possible mediators for the blame attributions (i.e., mind attribution, attributed agency, perceived free will, responsibility, intentionality, and whether the misclassification by the AI could have been detected by the user) or moderators (i.e., perceived severity of poisoning).

Based on the theoretical model that the tendency to anthropomorphize non-human agents is influenced by motivational factors such as a sociality motivation (Epley, Waytz, Akalis, & Cacioppo, 2008), we test if loneliness is correlated with the anthropomorphism of a non-human-looking robot. We therefore measure trait and state loneliness levels, as well as several aspects of anthropomorphism of visitors of the art installation “Contact” in which people can spontaneously get in touch with a non-human-looking robot in a shop window on the streets (Gollob & May, 2021). We additionally collect observational data, i.e., if visitors touch the window in order to have “contact” with the robot.

Based on the theoretical model that the tendency to anthropomorphize non-human agents is influenced by motivational factors such as a sociality motivation (Epley, Waytz, Akalis, & Cacioppo, 2008), we test if loneliness is correlated with the anthropomorphism of a non-human-looking robot. We therefore measure trait and state loneliness levels, as well as several aspects of anthropomorphism in an online study using the video documentation of an art installation in which a human visitor interacts with a robot arm seeking to contact the human through the window of an art gallery (Gollob & May, 2021).

In this laboratory experiment, the effect of introductory information on the evaluation of a collaborative robot (e.g., trustworthiness, predictability and understandability) as well as on the performance of the human-robot team in collaborative game tasks (time to completion and task success) will be explored. We use a virtual reality game specifically designed for this study called "CoBot Studio: Clean-Up" in which human participants have to collaborate with a robot on three different tasks in a fictional underwater research station.

In this online experiment, it is tested how an AI-based voice assistant’s (VA) response strategy to gender-stereotypical insults affects the tolerance of sexism toward women and user preferences. In the online experiment, participants are introduced to an AI-based voice assistant system. This fictitious VA called SARA is presented as a new voice assistant system that can be used at home or on a person’s mobile phone. It is able to answer questions and interacts in a human-like manner. In the experiment, participants are confronted with five gender-stereotypical insults of potential users toward the VA and hear the responses of the VA in the form of an audio file, while also seeing a visual representation of the VA. For every presented situation of insult and response, participants are tasked to evaluate the VA's response. As a manipulation, participants will be randomly assigned to one of three different VA response-style groups (avoiding, submissive and educational). In the second part of the experiment, participants will be confronted with a real-life scenario. A picture of a situation in an office that shows a man and a woman will be presented to the participants. They are tasked to imagine that the five subsequent gender-stereotypical insults are said by a man to a woman in such an office scenario. Participants are then prompted to evaluate how appropriate or inappropriate they perceive these gender-stereotypical insults in the described situation (sexism tolerance). In addition, out of three possible answers, participants will have to choose what they think is the women's most adequate response to the men's comment (preference of response to sexism). Research has shown that behavior towards a digital system, can transmit to interpersonal relationships in real life (Lee et al., 2021; Whitby 2008; Chin et al., 2020). Therefore, we assume that the VA's answers to gender-stereotypical insults affect sexism tolerance and their preferences in how to respond toward sexism.

A recent literature review on personal space in human-robot interaction identified a research gap for the influence of contextual factors. At the same time, psychological research on interpersonal distancing and theoretical considerations based on compensatory control models suggest the importance of considering these factors in robot path planning. To address this gap, we tested the effect of room size and working memory load on participants’ comfort distance toward an approaching robot. In a preregistered 3 × 2 within-subject design, N = 72 participants were approached by a mobile manufacturing robot in a corridor with varying room size and with and without a cognitive secondary task. As dependent variables, comfort distance, arousal, and perceived control were measured. While room size and working memory load had no significant direct effect on comfort distance, participants felt higher arousal and lower control in smaller rooms and in conditions with high working memory load, which in turn caused larger comfort distances (indirect effect). With experience, comfort distances decreased. Based on the indirect effects, future studies should test the effect of more extreme manipulations on comfort distances. Robots should adapt their path planning by keeping larger distances toward human workers in stressful environments to avoid discomfort. Version of record

Frontiers in robotics and AI 9, 838116 (2022). doi:10.3389/frobt.2022.838116, Published by Frontiers, Lausanne

There is a confidence crisis in many scientific disciplines, in particular disciplines researching human behavior, as many effects of original experiments have not been replicated successfully in large-scale replication studies. While human-robot interaction (HRI) is an interdisciplinary research field, the study of human behavior, cognition and emotion in HRI plays also a vital part. Are HRI user studies facing the same problems as other fields and if so, what can be done to overcome them? In this article, we first give a short overview of the replicability crisis in behavioral sciences and its causes. In a second step, we estimate the replicability of HRI user studies mainly 1) by structural comparison of HRI research processes and practices with those of other disciplines with replicability issues, 2) by systematically reviewing meta-analyses of HRI user studies to identify parameters that are known to affect replicability, and 3) by summarizing first replication studies in HRI as direct evidence. Our findings suggest that HRI user studies often exhibit the same problems that caused the replicability crisis in many behavioral sciences, such as small sample sizes, lack of theory, or missing information in reported data. In order to improve the stability of future HRI research, we propose some statistical, methodological and social reforms. This article aims to provide a basis for further discussion and a potential outline for improvements in the field. Version of record

Augsburg : Fraunhofer-Institut für Gießerei-, Composite- und Verarbeitungstechnik IGCV 196 Seiten (2021). doi:10.24406/IGCV-N-624794, Published by Fraunhofer-Institut für Gießerei-, Composite- und Verarbeitungstechnik IGCV, Augsburg

Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2021; Aachen : RWTH Aachen University 1 Online-Ressource : Illustrationen, Diagramme (2021). = Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2021, The successful implementation of new robotic technology in contexts with human users requires a socially acceptable human-robot interaction and needs to consider the socio-technical system holistically. Human-robot interaction is embedded in a social system and shapes but also is shaped by this social context. Research in human-robot interaction thus needs to take into consideration the social matrix in which the interaction is embedded on multiple levels of analysis. One scientific field that can enhance human-robot interaction research in achieving socially acceptable designs of systems is social psychology. Based on these assumptions, this dissertation has three major contributions: First, it is conceptually shown how socially acceptable human-robot interaction can be studied holistically by applying social-psychological theories, methods, and research practices on multiple levels of analysis. In doing so, a framework with two layers is proposed, a layer of phenomena, on which the human-robot interaction as a social phenomenon is analyzed, and a meta-scientific layer, on which philosophical foundations, theories, methods, and research practices of human-robot interaction as a discipline are critically reflected. Furthermore, on the layer of phenomena, human-robot interaction is analyzed on three different levels, an individual level focusing on social thinking, a situational level focusing on social relations and contextual effects, and a higher system level describing social impact emerging from social system dynamics.Second, in order to contribute to a better understanding of human-robot interaction on each level of analysis, four empirical studies are presented addressing research gaps on different topics of human-robot interaction. Social thinking is analyzed in a laboratory experiment on the social desirability bias with N = 107 participants. This experiment is a conceptual replication of a classic study by Nass, Moon, and Carney (1999) stating that technological devices are perceived as social actors. However, the results failed to show such an effect indicating that the social desirability effect is overestimated. Social relations are analyzed by studying personal space in human robot interaction. A systematic literature review and meta-analysis including k = 27 studies with N = 1299 participants, as well as theoretical considerations, show that knowledge on personal space in human-robot interaction is limited due to mixed effects and many problems in theory, methods, and research practices. An additional laboratory experiment with N = 72 participants on the effect of contextual factors on personal space is presented to address a research gap. However, neither room size nor working memory load had significant effect on users’ comfort distance toward an approaching manufacturing robot. In order to understand the social impact of robots, attitudes toward the implementation of a new cooperative industrial robot were analyzed based on data of N = 556 workers from manufacturing companies using factor and network analysis. In doing so, a new, context-specific 12-item questionnaire (ACIR-Q) had been developed. This study shows how workers’ attitudes toward robots are influenced by organizational variables and differ between organizations.A third contribution of this dissertation is based on meta-scientific conclusions. Throughout the dissertation, it is demonstrated how human-robot interaction research can be improved in terms of theory, methods, and scientific practices. Human-robot interaction is facing the same problems that have been discussed in context of social psychology’s replicability crisis. Therefore, the reforms in social psychology are also recommended for human-robot interaction research.Human-robot interaction is still in its infancy and the results in this dissertation show that it is facing many challenges. Social-psychology can not only help to design social-acceptable human-robot interaction on different levels of analysis, but also enhance the human-robot interaction discipline’s research methods and theory., Published by RWTH Aachen University, Aachen