Chuxuan Zhang, Bermet Burkanova, Lawrence H. Kim, Lauren Yip, Ugo Cupcic, Stéphane Lallée, Angelica Lim
How do people use their faces and bodies to test the interactive abilities of a robot? Making lively, believable agents is often seen as a goal for robots and virtual agents but believability can easily break down. In this Wizard-of-Oz (WoZ) study, we observed 1169 nonverbal interactions between 20 participants and 6 types of agents. We collected the nonverbal behaviors participants used to challenge the characters physically, emotionally, and socially. The participants interacted freely with humanoid and non-humanoid forms: a robot, a human, a penguin, a pufferfish, a banana, and a toilet. We present a human behavior codebook of 188 unique nonverbal behaviors used by humans to test the virtual characters. The insights and design strategies drawn from video observations aim to help build more interaction-aware and believable robots, especially when humans push them to their limits.
{"title":"React to This! How Humans Challenge Interactive Agents using Nonverbal Behaviors","authors":"Chuxuan Zhang, Bermet Burkanova, Lawrence H. Kim, Lauren Yip, Ugo Cupcic, Stéphane Lallée, Angelica Lim","doi":"arxiv-2409.11602","DOIUrl":"https://doi.org/arxiv-2409.11602","url":null,"abstract":"How do people use their faces and bodies to test the interactive abilities of\u0000a robot? Making lively, believable agents is often seen as a goal for robots\u0000and virtual agents but believability can easily break down. In this\u0000Wizard-of-Oz (WoZ) study, we observed 1169 nonverbal interactions between 20\u0000participants and 6 types of agents. We collected the nonverbal behaviors\u0000participants used to challenge the characters physically, emotionally, and\u0000socially. The participants interacted freely with humanoid and non-humanoid\u0000forms: a robot, a human, a penguin, a pufferfish, a banana, and a toilet. We\u0000present a human behavior codebook of 188 unique nonverbal behaviors used by\u0000humans to test the virtual characters. The insights and design strategies drawn\u0000from video observations aim to help build more interaction-aware and believable\u0000robots, especially when humans push them to their limits.","PeriodicalId":501541,"journal":{"name":"arXiv - CS - Human-Computer Interaction","volume":"4 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142252160","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This research dives into exploring the dark mode effects on students of a university. Research is carried out implementing the dark mode in e-Learning sites and its impact on behavior of the users. Students are spending more time in front of the screen for their studies especially after the pandemic. The blue light from the screen during late hours affects circadian rhythm of the body which negatively impacts the health of humans including eye strain and headache. The difficulty that students faced during the time of interacting with various e-Learning sites especially during late hours was analyzed using different techniques of HCI like survey, interview, evaluation methods and principles of design. Dark mode is an option which creates a pseudo inverted adaptable interface by changing brighter elements of UI into a dim-lit friendly environment. It is said that using dark mode will lessen the amount of blue light emitted and benefit students who suffer from eye strain. Students' interactions with dark mode were investigated using a survey, and an e-learning site with a dark mode theme was created. Based on the students' comments, researchers looked into the effects of dark mode on HCI in e-learning sites. The findings indicate that students have a clear preference for dark mode: 79.7% of survey participants preferred dark mode on their phones, and 61.7% said they would be interested in seeing this feature added to e-learning websites.
{"title":"An Exploration of Effects of Dark Mode on University Students: A Human Computer Interface Analysis","authors":"Awan Shrestha, Sabil Shrestha, Biplov Paneru, Bishwash Paneru, Sansrit Paudel, Ashish Adhikari, Sanjog Chhetri Sapkota","doi":"arxiv-2409.10895","DOIUrl":"https://doi.org/arxiv-2409.10895","url":null,"abstract":"This research dives into exploring the dark mode effects on students of a\u0000university. Research is carried out implementing the dark mode in e-Learning\u0000sites and its impact on behavior of the users. Students are spending more time\u0000in front of the screen for their studies especially after the pandemic. The\u0000blue light from the screen during late hours affects circadian rhythm of the\u0000body which negatively impacts the health of humans including eye strain and\u0000headache. The difficulty that students faced during the time of interacting\u0000with various e-Learning sites especially during late hours was analyzed using\u0000different techniques of HCI like survey, interview, evaluation methods and\u0000principles of design. Dark mode is an option which creates a pseudo inverted\u0000adaptable interface by changing brighter elements of UI into a dim-lit friendly\u0000environment. It is said that using dark mode will lessen the amount of blue\u0000light emitted and benefit students who suffer from eye strain. Students'\u0000interactions with dark mode were investigated using a survey, and an e-learning\u0000site with a dark mode theme was created. Based on the students' comments,\u0000researchers looked into the effects of dark mode on HCI in e-learning sites.\u0000The findings indicate that students have a clear preference for dark mode:\u000079.7% of survey participants preferred dark mode on their phones, and 61.7%\u0000said they would be interested in seeing this feature added to e-learning\u0000websites.","PeriodicalId":501541,"journal":{"name":"arXiv - CS - Human-Computer Interaction","volume":"208 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142252426","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Emotions, shaped by past experiences, significantly influence decision-making and goal pursuit. Traditional cognitive-behavioral techniques for personal development rely on mental imagery to envision ideal selves, but may be less effective for individuals who struggle with visualization. This paper introduces Emotional Self-Voice (ESV), a novel system combining emotionally expressive language models and voice cloning technologies to render customized responses in the user's own voice. We investigate the potential of ESV to nudge individuals towards their ideal selves in a study with 60 participants. Across all three conditions (ESV, text-only, and mental imagination), we observed an increase in resilience, confidence, motivation, and goal commitment, but the ESV condition was perceived as uniquely engaging and personalized. We discuss the implications of designing generated self-voice systems as a personalized behavioral intervention for different scenarios.
{"title":"Leveraging AI-Generated Emotional Self-Voice to Nudge People towards their Ideal Selves","authors":"Cathy Mengying Fang, Phoebe Chua, Samantha Chan, Joanne Leong, Andria Bao, Pattie Maes","doi":"arxiv-2409.11531","DOIUrl":"https://doi.org/arxiv-2409.11531","url":null,"abstract":"Emotions, shaped by past experiences, significantly influence decision-making\u0000and goal pursuit. Traditional cognitive-behavioral techniques for personal\u0000development rely on mental imagery to envision ideal selves, but may be less\u0000effective for individuals who struggle with visualization. This paper\u0000introduces Emotional Self-Voice (ESV), a novel system combining emotionally\u0000expressive language models and voice cloning technologies to render customized\u0000responses in the user's own voice. We investigate the potential of ESV to nudge\u0000individuals towards their ideal selves in a study with 60 participants. Across\u0000all three conditions (ESV, text-only, and mental imagination), we observed an\u0000increase in resilience, confidence, motivation, and goal commitment, but the\u0000ESV condition was perceived as uniquely engaging and personalized. We discuss\u0000the implications of designing generated self-voice systems as a personalized\u0000behavioral intervention for different scenarios.","PeriodicalId":501541,"journal":{"name":"arXiv - CS - Human-Computer Interaction","volume":"116 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142252435","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Interactive Task Learning (ITL) systems acquire task knowledge from human instructions in natural language interaction. The interaction design of ITL agents for hierarchical tasks stays uncharted. This paper studied Verbal Apprentice Learner(VAL) for gaming, as an ITL example, and qualitatively analyzed the user study data to provide design insights on dialogue language types, task instruction strategies, and error handling. We then proposed an interface design: Editable Hierarchy Knowledge (EHK), as a generic probe for ITL systems for hierarchical tasks.
{"title":"Improving Interface Design in Interactive Task Learning for Hierarchical Tasks based on a Qualitative Study","authors":"Jieyu Zhou, Christopher MacLellan","doi":"arxiv-2409.10826","DOIUrl":"https://doi.org/arxiv-2409.10826","url":null,"abstract":"Interactive Task Learning (ITL) systems acquire task knowledge from human\u0000instructions in natural language interaction. The interaction design of ITL\u0000agents for hierarchical tasks stays uncharted. This paper studied Verbal\u0000Apprentice Learner(VAL) for gaming, as an ITL example, and qualitatively\u0000analyzed the user study data to provide design insights on dialogue language\u0000types, task instruction strategies, and error handling. We then proposed an\u0000interface design: Editable Hierarchy Knowledge (EHK), as a generic probe for\u0000ITL systems for hierarchical tasks.","PeriodicalId":501541,"journal":{"name":"arXiv - CS - Human-Computer Interaction","volume":"42 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142252428","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jonas Hein, Jan Grunder, Lilian Calvet, Frédéric Giraud, Nicola Alessandro Cavalcanti, Fabio Carrillo, Philipp Fürnstahl
Virtual Reality technology, when integrated with Surgical Digital Twins (SDTs), offers significant potential in medical training and surgical planning. We present SurgTwinVR, a VR application that immerses users within an SDT and enables them to navigate a high-fidelity virtual replica of the surgical environment. SurgTwinVR is the first VR application to utilize a dynamic 3D environment that is a clone of a real surgery, encompassing the entire surgical scene, including the surgeon, anatomy, and instruments. Our system utilizes a SDT with important improvements for real-time rendering and features to showcase the potential benefits of such an application in surgical education.
{"title":"Virtual Reality for Immersive Education in Orthopedic Surgery Digital Twins","authors":"Jonas Hein, Jan Grunder, Lilian Calvet, Frédéric Giraud, Nicola Alessandro Cavalcanti, Fabio Carrillo, Philipp Fürnstahl","doi":"arxiv-2409.11014","DOIUrl":"https://doi.org/arxiv-2409.11014","url":null,"abstract":"Virtual Reality technology, when integrated with Surgical Digital Twins\u0000(SDTs), offers significant potential in medical training and surgical planning.\u0000We present SurgTwinVR, a VR application that immerses users within an SDT and\u0000enables them to navigate a high-fidelity virtual replica of the surgical\u0000environment. SurgTwinVR is the first VR application to utilize a dynamic 3D\u0000environment that is a clone of a real surgery, encompassing the entire surgical\u0000scene, including the surgeon, anatomy, and instruments. Our system utilizes a\u0000SDT with important improvements for real-time rendering and features to\u0000showcase the potential benefits of such an application in surgical education.","PeriodicalId":501541,"journal":{"name":"arXiv - CS - Human-Computer Interaction","volume":"6 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142252423","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Community health workers (CHWs) provide last-mile healthcare services but face challenges due to limited medical knowledge and training. This paper describes the design, deployment, and evaluation of ASHABot, an LLM-powered, experts-in-the-loop, WhatsApp-based chatbot to address the information needs of CHWs in India. Through interviews with CHWs and their supervisors and log analysis, we examine factors affecting their engagement with ASHABot, and ASHABot's role in addressing CHWs' informational needs. We found that ASHABot provided a private channel for CHWs to ask rudimentary and sensitive questions they hesitated to ask supervisors. CHWs trusted the information they received on ASHABot and treated it as an authoritative resource. CHWs' supervisors expanded their knowledge by contributing answers to questions ASHABot failed to answer, but were concerned about demands on their workload and increased accountability. We emphasize positioning LLMs as supplemental fallible resources within the community healthcare ecosystem, instead of as replacements for supervisor support.
{"title":"ASHABot: An LLM-Powered Chatbot to Support the Informational Needs of Community Health Workers","authors":"Pragnya Ramjee, Mehak Chhokar, Bhuvan Sachdeva, Mahendra Meena, Hamid Abdullah, Aditya Vashistha, Ruchit Nagar, Mohit Jain","doi":"arxiv-2409.10913","DOIUrl":"https://doi.org/arxiv-2409.10913","url":null,"abstract":"Community health workers (CHWs) provide last-mile healthcare services but\u0000face challenges due to limited medical knowledge and training. This paper\u0000describes the design, deployment, and evaluation of ASHABot, an LLM-powered,\u0000experts-in-the-loop, WhatsApp-based chatbot to address the information needs of\u0000CHWs in India. Through interviews with CHWs and their supervisors and log\u0000analysis, we examine factors affecting their engagement with ASHABot, and\u0000ASHABot's role in addressing CHWs' informational needs. We found that ASHABot\u0000provided a private channel for CHWs to ask rudimentary and sensitive questions\u0000they hesitated to ask supervisors. CHWs trusted the information they received\u0000on ASHABot and treated it as an authoritative resource. CHWs' supervisors\u0000expanded their knowledge by contributing answers to questions ASHABot failed to\u0000answer, but were concerned about demands on their workload and increased\u0000accountability. We emphasize positioning LLMs as supplemental fallible\u0000resources within the community healthcare ecosystem, instead of as replacements\u0000for supervisor support.","PeriodicalId":501541,"journal":{"name":"arXiv - CS - Human-Computer Interaction","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142252424","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
One application area of long-term memory (LTM) capabilities with increasing traction is personal AI companions and assistants. With the ability to retain and contextualize past interactions and adapt to user preferences, personal AI companions and assistants promise a profound shift in how we interact with AI and are on track to become indispensable in personal and professional settings. However, this advancement introduces new challenges and vulnerabilities that require careful consideration regarding the deployment and widespread use of these systems. The goal of this paper is to explore the broader implications of building and deploying personal AI applications with LTM capabilities using a holistic evaluation approach. This will be done in three ways: 1) reviewing the technological underpinnings of LTM in Large Language Models, 2) surveying current personal AI companions and assistants, and 3) analyzing critical considerations and implications of deploying and using these applications.
{"title":"Towards Ethical Personal AI Applications: Practical Considerations for AI Assistants with Long-Term Memory","authors":"Eunhae Lee","doi":"arxiv-2409.11192","DOIUrl":"https://doi.org/arxiv-2409.11192","url":null,"abstract":"One application area of long-term memory (LTM) capabilities with increasing\u0000traction is personal AI companions and assistants. With the ability to retain\u0000and contextualize past interactions and adapt to user preferences, personal AI\u0000companions and assistants promise a profound shift in how we interact with AI\u0000and are on track to become indispensable in personal and professional settings.\u0000However, this advancement introduces new challenges and vulnerabilities that\u0000require careful consideration regarding the deployment and widespread use of\u0000these systems. The goal of this paper is to explore the broader implications of\u0000building and deploying personal AI applications with LTM capabilities using a\u0000holistic evaluation approach. This will be done in three ways: 1) reviewing the\u0000technological underpinnings of LTM in Large Language Models, 2) surveying\u0000current personal AI companions and assistants, and 3) analyzing critical\u0000considerations and implications of deploying and using these applications.","PeriodicalId":501541,"journal":{"name":"arXiv - CS - Human-Computer Interaction","volume":"25 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142252433","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Aditya Raikwar, Lucas Plabst, Anil Ufuk Batmaz, Florian Niebling, Francisco R. Ortega
Implementing visual and audio notifications on augmented reality devices is a crucial element of intuitive and easy-to-use interfaces. In this paper, we explored creating intuitive interfaces through visual and audio notifications. The study evaluated user performance and preference across three conditions: visual notifications in fixed positions, visual notifications above objects, and no visual notifications with monaural sounds. The users were tasked with cooking and serving customers in an open-source Augmented-Reality sandbox environment called ARtisan Bistro. The results indicated that visual notifications above objects combined with localized audio feedback were the most effective and preferred method by participants. The findings highlight the importance of strategic placement of visual and audio notifications in AR, providing insights for engineers and developers to design intuitive 3D user interfaces.
在增强现实设备上实现视觉和音频通知是直观易用界面的关键要素。本文探讨了通过视觉和音频通知创建直观界面的问题。研究评估了用户在三种情况下的表现和偏好：固定位置的视觉通知、物体上方的视觉通知以及无视觉通知的单声道声音。用户的任务是在名为 ARtisan Bistro 的开源增强现实沙盒环境中烹饪并为顾客提供服务。结果表明，物体上方的视觉通知与本地化音频反馈相结合是最有效且最受参与者青睐的方法。研究结果强调了在 AR 中战略性地放置视觉和音频通知的重要性，为工程师和开发人员设计直观的 3D 用户界面提供了启示。
{"title":"Ping! Your Food is Ready: Comparing Different Notification Techniques in 3D AR Cooking Environment","authors":"Aditya Raikwar, Lucas Plabst, Anil Ufuk Batmaz, Florian Niebling, Francisco R. Ortega","doi":"arxiv-2409.11357","DOIUrl":"https://doi.org/arxiv-2409.11357","url":null,"abstract":"Implementing visual and audio notifications on augmented reality devices is a\u0000crucial element of intuitive and easy-to-use interfaces. In this paper, we\u0000explored creating intuitive interfaces through visual and audio notifications.\u0000The study evaluated user performance and preference across three conditions:\u0000visual notifications in fixed positions, visual notifications above objects,\u0000and no visual notifications with monaural sounds. The users were tasked with\u0000cooking and serving customers in an open-source Augmented-Reality sandbox\u0000environment called ARtisan Bistro. The results indicated that visual\u0000notifications above objects combined with localized audio feedback were the\u0000most effective and preferred method by participants. The findings highlight the\u0000importance of strategic placement of visual and audio notifications in AR,\u0000providing insights for engineers and developers to design intuitive 3D user\u0000interfaces.","PeriodicalId":501541,"journal":{"name":"arXiv - CS - Human-Computer Interaction","volume":"65 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142252422","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Drill tool positioning in dental implantology is a challenging task requiring 5DOF precision as the rotation around the tool axis is not influential. This work improves the quasi-static visual elements of the state-of-the-art with a novel Augmented Collimation Widget (ACW), an interactive tool of position and angle error visualization based on the gestalt reification, the human ability to group geometric elements. The user can seek in a quick, pre-attentive way the collimation of five (three positional and two rotational) error component widgets (ECWs), taking advantage of three key aspects: component separation and reification, error visual amplification, and dynamic hiding of the collimated components. We compared the ACW with the golden standard in a within-subjects (N=30) user test using 32 implant targets, measuring the time, error, and usability. ACW performed significantly better in positional (+19%) and angular (+47%) precision accuracy and with less mental demand (-6%) and frustration (-13%), but with an expected increase in task time (+59%) and physical demand (+64%). The interview indicated the ACW as the main preference and aesthetically more pleasant than GSW, candidating it as the new golden standard for implantology, but also for other applications where 5DOF positioning is key.
{"title":"Gestalt driven augmented collimator widget for precise 5 dof dental drill tool positioning in 3d space","authors":"Mine Dastan, Antonio E. Uva, Michele Fiorentino","doi":"arxiv-2409.10960","DOIUrl":"https://doi.org/arxiv-2409.10960","url":null,"abstract":"Drill tool positioning in dental implantology is a challenging task requiring\u00005DOF precision as the rotation around the tool axis is not influential. This\u0000work improves the quasi-static visual elements of the state-of-the-art with a\u0000novel Augmented Collimation Widget (ACW), an interactive tool of position and\u0000angle error visualization based on the gestalt reification, the human ability\u0000to group geometric elements. The user can seek in a quick, pre-attentive way\u0000the collimation of five (three positional and two rotational) error component\u0000widgets (ECWs), taking advantage of three key aspects: component separation and\u0000reification, error visual amplification, and dynamic hiding of the collimated\u0000components. We compared the ACW with the golden standard in a within-subjects\u0000(N=30) user test using 32 implant targets, measuring the time, error, and\u0000usability. ACW performed significantly better in positional (+19%) and angular\u0000(+47%) precision accuracy and with less mental demand (-6%) and frustration\u0000(-13%), but with an expected increase in task time (+59%) and physical demand\u0000(+64%). The interview indicated the ACW as the main preference and\u0000aesthetically more pleasant than GSW, candidating it as the new golden standard\u0000for implantology, but also for other applications where 5DOF positioning is\u0000key.","PeriodicalId":501541,"journal":{"name":"arXiv - CS - Human-Computer Interaction","volume":"25 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142252425","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Large language models (LLMs) are being increasingly integrated into everyday products and services, such as coding tools and writing assistants. As these embedded AI applications are deployed globally, there is a growing concern that the AI models underlying these applications prioritize Western values. This paper investigates what happens when a Western-centric AI model provides writing suggestions to users from a different cultural background. We conducted a cross-cultural controlled experiment with 118 participants from India and the United States who completed culturally grounded writing tasks with and without AI suggestions. Our analysis reveals that AI provided greater efficiency gains for Americans compared to Indians. Moreover, AI suggestions led Indian participants to adopt Western writing styles, altering not just what is written but also how it is written. These findings show that Western-centric AI models homogenize writing toward Western norms, diminishing nuances that differentiate cultural expression.
{"title":"AI Suggestions Homogenize Writing Toward Western Styles and Diminish Cultural Nuances","authors":"Dhruv Agarwal, Mor Naaman, Aditya Vashistha","doi":"arxiv-2409.11360","DOIUrl":"https://doi.org/arxiv-2409.11360","url":null,"abstract":"Large language models (LLMs) are being increasingly integrated into everyday\u0000products and services, such as coding tools and writing assistants. As these\u0000embedded AI applications are deployed globally, there is a growing concern that\u0000the AI models underlying these applications prioritize Western values. This\u0000paper investigates what happens when a Western-centric AI model provides\u0000writing suggestions to users from a different cultural background. We conducted\u0000a cross-cultural controlled experiment with 118 participants from India and the\u0000United States who completed culturally grounded writing tasks with and without\u0000AI suggestions. Our analysis reveals that AI provided greater efficiency gains\u0000for Americans compared to Indians. Moreover, AI suggestions led Indian\u0000participants to adopt Western writing styles, altering not just what is written\u0000but also how it is written. These findings show that Western-centric AI models\u0000homogenize writing toward Western norms, diminishing nuances that differentiate\u0000cultural expression.","PeriodicalId":501541,"journal":{"name":"arXiv - CS - Human-Computer Interaction","volume":"15 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142252421","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：