67471 research outputs found
Sort by
Second Chances: The Impact of Remeasurements on Performance and Trust in Voice User Interfaces for Nuclear Safeguards
Get PDFVoice User Interface (VUI) communications significantly affect user decisions and trust, especially in high-stakes areas like international nuclear safeguards. We investigate these effects through two controlled experiments centered on nuclear safeguards inspection tasks. In Experiment 1, participants evaluated a seal on a nuclear container for tampering, with VUI uncertainty communication manipulated by varying remeasurement suggestions—suggested, required, or none. Results indicated that participants in the suggested remeasure condition exhibited higher trust in the VUI and improved accuracy in identifying true negatives, although they had slower response times for false positives. Experiment 2 involved a nuclear materials measurement task, where the VUI facilitated data collection. While most participants reported similar trust levels, those who perceived differences favored the suggested remeasure condition. Additionally, participants demonstrated quicker responses and higher compliance rates when the VUI recommended remeasurements. Overall, these findings suggest that VUI recommendations can enhance user trust and performance in critical applications
Towards an Understanding of the Impacts of Hybrid Work Technologies on Racialized and Underrepresented People
Get PDFOften promoted as a pillar of the Future of Work, hybrid work is a flexible workplace model combining both in-office and remote work; it is enabled by technologies such as the internet, video-conferencing, enterprise social media (ESM), collaboration platforms, as well as algorithmic management. Despite its growing prevalence, the benefits associated with hybrid work—such as improved work-life balance, more efficient time use, burnout mitigation, and higher productivity—may not be equally distributed. Emerging research suggests that hybrid models risk reinforcing workplace inequities, particularly for racialized and underrepresented workers. This research-in-progress paper theorizes the paradoxical effects of hybrid work technologies on diversity, equity, and inclusion (DEI) in the workplace. Drawing on socio-material theory, affordances, paradox theory, and critical perspectives, this paper develops a conceptual model and set of falsifiable research propositions. Research limitations, and scholarly and practical implications are discussed, as are the proposed next steps in this research project
New Possibilities, New Responsibilities: Ethical Challenges and Guidance for Location Tracking Device-based Research in Supply Chains
Get PDFThis study explores the use of location tracking devices (LTDs) as a method for generating independent, geospatially grounded data to enhance transparency in global supply chains. While LTDs offer promising alternatives to self-reported data, their application introduces complex ethical challenges. Through a developmental literature review, this work synthesizes how LTDs have been empirically employed and how ethical concerns – such as consent, fairness, and data protection – have been addressed. Drawing on these insights, it proposes a practical ethics framework tailored to LTD-based studies in supply chain contexts. The findings contribute to geographic information systems (GIS) research by promoting the integration of geospatial data into opaque domains and support the Green IS agenda by advancing responsible, sustainability-oriented research design
Too Many LLMs to Choose? Design and Trust in Multi-AI Collaboration Systems
Get PDFAlthough generative artificial intelligence (GAI) has demonstrated potential in many fields, it still struggles with issues such as limited knowledge and hallucinations. To address these issues, we present a multi-AI collaboration system. Our study takes a multimethod approach, combining design science research (DSR) with quantitative methods. Our aim is to explore the design of effective multi-AI systems and how these affect team trust compared to single-AI setups. We also aim to investigate the differences in users' collaboration experiences between the two models. Our results reveal that multi-AI systems can significantly enhance users' trust in AI, particularly with regard to perceived ability. Users also reported a higher willingness to use multi-AI collaboration and greater satisfaction with it. However, they also noticed more perceived conflict when working with multiple AIs. These findings will help to create more reliable and user-friendly AI collaboration tools, improving the way in which humans and AI can work together effectively
Experience Over Explanation: Perceived Transparency in AI-Based Skin Cancer Detection
Get PDFArtificial intelligence (AI) is increasingly integrated into everyday life and holds great potential for high-stakes domains such as healthcare, for example in early skin cancer detection. However, user trust remains a major barrier to adoption, prior research has largely treated explainable AI (XAI) approaches as universally applicable rather than accounting for individual differences. In this study, we investigate how three XAI formats (mechanism–modality pairings) shape user trust through perceived transparency in AI-powered skin cancer diagnostics. Using a between-subjects online experiment with 15 dermoscopic images, we show that the effect of XAI format on trust is fully mediated by perceived transparency, and this is significantly moderated by users’ AI experience. Notably, AI experience can reverse the effect, underscoring the importance of tailoring explanations to user backgrounds. These findings advance the understanding of how trust in AI can be more appropriately calibrated and provide guidance for designing personalized XAI in healthcare
Evaluation of Information Extraction Algorithms for Preserving Analogical Semantics within Knowledge Graphs
Get PDFAnalogical reasoning is a promising, lightweight solution to inferencing on novel data without prior training. However, algorithms with this methodology have historically relied on strict, human-defined schemas. To address this issue, this work proposes using information extraction algorithms to transform textual analogies into knowledge graphs (KGs) for a more machine-friendly format. We compare the knowledge graphs created by four relation extractors, three co-reference resolvers, and two embedding models via Pearson’s coefficient, root mean squared error (RMSE), and the Wilcoxon Signed-rank Test. We observe that the Wilcoxon Signed-rank Test provided the most streamlined result for algorithm and embedding selection, and that extractors were more influential than the resolver in creating KGs. From this, OpenIE was the best-performing extractor, and the SBERT embeddings yielded the KGs that best preserved analogical structure. Future work should focus on additional statistical tests and a greater range of information extraction algorithms and embeddings
How Do Scientists Communicate on Social Media During Crisis?
Get PDFThis study explores how scientists utilize social media for science communication during crises, with a focus on the COVID-19 pandemic. Grounded in philosophical stance of sociomateriality and in affordance theory, it investigates the relational dynamics between scientists, social media platforms, and crisis contexts. In this paper, we employ a mixed-method approach that combines content analysis and network analysis to understand both what scientists communicated and with whom they communicated. Analyzing 1,761 social media activities from eight German virologists over six pandemic phases, our research reveals how scientists use social media to facilitate change during times of crisis. Also, we show that despite social media platforms afford scientists to connect with various types of groups, they primarily interacted with accounts of scientific entities. This study contributes theoretically by illustrating context-dependent, relational uses of social media affordances during crises, and practically as it enables organizations and decision-makers develop strategies for crisis contexts
Introduction to the Minitrack on Human-Like Conversational Agents: Chatbots, Digital Humans, and Virtual Influencers
No full textUnderstanding the Impact of AI Label Disclosure on Engagement with Deepfakes Videos
Get PDFThe rapid proliferation of AI-generated content on digital platforms—particularly deepfakes involving synthetic audio and videos—has intensified concerns about misinformation, content authenticity, and user trust. In response, major platforms have implemented AI labeling policies to enhance transparency and mitigate the societal risks posed by deepfakes. Drawing on the Elaboration Likelihood Model, this study empirically examines the effectiveness of such interventions by analyzing the relationship between video inauthenticity and user engagement, and how AI label disclosure moderates this effect. We analyzed a dataset of 787 TikTok videos, generated by 30 unique content creators between November 12, 2024, and April 14, 2025. We found that video inauthenticity negatively impacts user engagement, such as liking and sharing. The AI label disclosure has a positive moderating role in the negative relationship between video inauthenticity and user engagement. Our research informs policy and design strategies for platform governance in an era increasingly shaped by generative AI
Signals that Matter: Gender, Content Framing, and Engagement in Online Communities of Practice
Get PDFAs social media evolves into a core venue for professional interaction and knowledge exchange, understanding what drives engagement in online communities of practice remains a crucial area of inquiry. This study examines how gender and content framing jointly influence peer engagement in an online physician community (i.e., pathology community) on X. Analyzing 2,467 patient case tweets using matched sampling and negative binomial random-effects models, we find that content authored by female physicians receives more favorable engagement than content authored by male physicians. Engagement also varies by content framing: diagnostic challenges and curbside consultations elicit more replies than mere shares. Female physicians benefit more from sharing less cognitively demanding content, while gender differences diminish for complex cases. These findings highlight the decision dilemma professionals face when framing knowledge contributions and reveal how identity signals and content strategies jointly shape engagement in online communities of practice, offering both theoretical and practical insights