Vulnerability of LLM Outputs to Heuristics-Inducing Prompt StructuresLarge Language Models (LLMs) have become indispensable tools in daily life. Although LLM applications have rapidly expanded across various domains, distorted outputs (specifically, bias and hallucination) are unresolved problems that threaten the reliability of LLM-based artificial intelligence (AI) agents. Focusing on internal mechanisms and social biases, prior research has rarely considered the possibility of distortion-induction even from non-malicious, ordinary prompts with specific input patterns. This study empirically investigates whether LLMs exposed to certain prompt structures can induce biases commonly elicited in humans, namely, representativeness heuristics, anchoring heuristics, and framing heuristics. To this end, we constructed a test set that triggers one of the three heuristics and evaluated the outputs of state-of-the-art LLMs. We further examined the effectiveness of prompt engineering and debiasing interventions. The LLMs continued to produce heuristic-derived biased outputs under certain prompt conditions. Anchoring heuristics were observed at rates significantly above chance, whereas the representativeness and framing heuristics depended on the model and prompt structure. Debiasing interventions notably reduced the representativeness heuristics but exerted limited impact on anchoring and framing heuristics. This study highlights the need for enhanced awareness of vulnerabilities in LLM outputs against particular prompts. It also reveals that typical prompt-engineering strategies offer insufficient protection against such prompt structures. These results will contribute to the safe and effective use of LLMs in human–computer interactions and AI deployment.2026TKToshiki Kuramoto et al.Bridgestone CorporationHuman-LLM CollaborationExplainable AI (XAI)AI Ethics, Fairness & AccountabilityIUI
EyeXRciser: Guiding Eye Exercises without Task Interruption in Virtual WorkspacesEye strain presents a significant challenge in human-information interaction in virtual reality (VR), as prolonged exposure contributes to various eye problems. This study introduces a new gaze redirection method called EyeXRciser, which passively activates eye movements to help prevent eye muscle stiffness during VR reading. This method achieves gaze redirection by slowly shifting the relative position of the text window within the user’s field of view through a head-bound coordinate system. We implemented our method with two different redirection speeds (i.e., unnoticeable speed at 0.03 $rad/s$; noticeable speed at 0.12 rad/s) and conducted a user study (N=24) comparing with a baseline using a fixed text window. Results show that both our methods successfully minimized the decline in accommodative ability caused by prolonged reading, without negatively impacting reading comprehension. Results also show that the unnoticeable redirection speed produced less subjective discomfort, eye fatigue, and reading distraction than the noticeable speed.2026HXHongyue Xu et al.Tohoku UniversityImmersion & Presence ResearchDigital Reading ExperienceEye Tracking & Gaze InteractionCHI
Screen-Directed Stretching: Supporting Ergonomic Neck Stretching in XR WorkspacesHabitual static stretching is recommended in many ergonomics guidelines, but continuing this habit remains challenging to workers. We address two factors causing this difficulty: most workers lack an understanding of the appropriate mechanisms of stretching, and stretching sessions require the interruption of ongoing tasks. We propose Screen-Directed Stretching, a novel technique that supports neck stretching in extended reality (XR) workspaces without interrupting work. Our technique temporarily repositions the virtual screen that the user is focusing on, guiding them to rotate their head in yaw, pitch, and roll directions and to maintain a posture for a specific duration, thus facilitating muscle extension. Through two preliminary user studies, we developed a prototype that seamlessly switches the screen's coordinate system between world-bound and body-bound frames of reference, balancing practical workability and good stretching guidance. A user study (N=16) demonstrates that our technique effectively induces stretching movements while minimizing loss of task efficiency.2026RSRyo Sasaki et al.Tohoku UniversityMixed Reality WorkspacesImmersion & Presence ResearchWorkplace Wellbeing & Work StressCHI
A Framework for Efficient Development and Debugging of Role-Playing Agents with Large Language ModelsWe propose a framework that leverages large language models (LLMs) to semi-automate the development and debugging of role-playing agents, reducing the need for extensive manual effort. Role-playing agents powered by LLMs offer scalable solutions that enhance communication and interaction in various applications, such as employee training, healthcare, and software development. However, creating prompts manually is a time-consuming process, and sequential debugging increases the difficulty of anticipating conversation flow, resulting in increased cognitive load. Our framework addresses these challenges by generating and summarizing dialogue examples, providing a clearer overview of conversation flow and reduce mental workload. It also enhances role-playing quality by mitigating LLMs’ tendency to produce generic or vague responses. In a user study, the proposed method significantly improved perceived workload and five of the six NASA-TLX dimensions. Moreover, it can generate agents comparable to those created with expertly crafted prompts. This framework is model-agnostic, enabling integration of advancements in LLM capabilities and prompting techniques, and is applicable to diverse domains.2025HTHirohane Takagi et al.Agent Personality & AnthropomorphismHuman-LLM CollaborationAI-Assisted Creative WritingIUI
Beyond Click to Cognition: Effective Interventions for Promoting Examination of False Beliefs in MisinformationIn the digital information ecosystem, clicks serve as a crucial gateway to fact-checking, yet the essential challenge extends beyond this to fostering cognitive shifts that update entrenched false beliefs. This study investigates effective interventions aimed at encouraging users vulnerable to misinformation, particularly those who tend to avoid incongruent facts, to examine their false beliefs. We conducted an online experiment with 627 participants, comparing metacognitive and ranking interventions. Both interventions successfully improved click behavior, with the metacognitive intervention increasing belief-examining clicks by 14 percentage points and the ranking intervention by 33 percentage points. However, only the metacognitive intervention significantly promoted users' examination of misinformation. This finding underscores the importance of interventions that go beyond merely influencing easily measurable clicks to facilitating thoughtful engagement with fact-checking content. We discuss implications for designing strategies to enhance online fact-checking engagement and mitigate misinformation's societal impact.2025YTYuko Tanaka et al.Nagoya Institute of Technology, Graduate School of EngineeringExplainable AI (XAI)Misinformation & Fact-CheckingAlgorithmic Fairness & BiasCHI
LoopBot: Representing Continuous Haptics of Grounded Objects in Room-scale VRIn room-scale virtual reality, providing continuous haptic feedback from touching grounded objects, such as walls and handrails, has been challenging due to the user's walking range and the required force. In this study, we propose LoopBot, a novel technique to provide continuous haptic feedback from grounded objects using only a single user-following robot. Specifically, LoopBot is equipped with a loop-shaped haptic prop attached to an omnidirectional robot that scrolls to cancel out the robot's displacement, giving the user the haptic sensation that the prop is actually fixed in place, or ``grounded.'' We first introduce the interaction design space of LoopBot and, as one of its promising interaction scenarios, implement a prototype for the experience of walking while grasping handrails. A performance evaluation shows that scrolling the prop cancels $77.5\%$ of the robot's running speed on average. A preliminary user test ($N=10$) also shows that the subjective realism of the experience and the sense of the virtual handrails being grounded were significantly higher than when the prop was not scrolled. Based on these findings, we discuss possible further development of LoopBot.2024TITetsushi Ikeda et al.Force Feedback & Pseudo-Haptic WeightFull-Body Interaction & Embodied InputSmart Home Privacy & SecurityUIST
InflatableBots: Inflatable Shape-Changing Mobile Robots for Large-Scale Encountered-Type Haptics in VRWe introduce InflatableBots, shape-changing inflatable robots for large-scale encountered-type haptics in VR. Unlike traditional inflatable shape displays, which are immobile and limited in interaction areas, our approach combines mobile robots with fan-based inflatable structures. This enables safe, scalable, and deployable haptic interactions on a large scale. We developed three coordinated inflatable mobile robots, each of which consists of an omni-directional mobile base and a reel-based inflatable structure. The robot can simultaneously change its height and position rapidly (horizontal: 58.5 cm/sec, vertical: 10.4 cm/sec, from 40 cm to 200 cm), which allows for quick and dynamic haptic rendering of multiple touch points to simulate various body-scale objects and surfaces in real-time across large spaces (3.5 m x 2.5 m). We evaluated our system with a user study (N = 12), which confirms the unique advantages in safety, deployability, and large-scale interactability to significantly improve realism in VR experiences.2024RGRyota Gomi et al.Tohoku UniversityShape-Changing Interfaces & Soft Robotic MaterialsSocial & Collaborative VRImmersion & Presence ResearchCHI
SwapVid: Integrating Video Viewing and Document Exploration with Direct ManipulationVideos accompanied by documents---\textit{document-based videos}---enable presenters to share contents beyond videos and audience to use them for detailed content comprehension. However, concurrently exploring multiple channels of information could be taxing. We propose SwapVid, a novel interface for viewing and exploring document-based videos. SwapVid seamlessly integrates a video and a document into a single view and lets the content behaves as both video and a document; it adaptively switches a document-based video to act as a video or a document upon direct manipulation (\textit{e.g.,} scrolling the document, manipulating the video timeline). We conducted a user study with twenty participants, comparing SwapVid to a side-by-side video/document views. Results showed that our interface reduces time and physical workload when exploring slide-based documents based on video referencing. Based on the study findings, we extended SwapVid with additional functionalities and demonstrated that it further extends the practical capabilities.2024TMTaichi Murakami et al.Tohoku UniversityInteractive Data VisualizationData StorytellingContext-Aware ComputingCHI
AiCommentator: A Multimodal Conversational Agent for Embedded Visualization in Football ViewingTraditionally, sports commentators provide viewers with diverse information, encompassing in-game developments and player performances. Yet young adult football viewers increasingly use mobile devices for deeper insights during football matches. Such insights into players on the pitch and performance statistics support viewers’ understanding of game stakes, creating a more engaging viewing experience. Inspired by commentators’ traditional roles and to incorporate information into a single platform, we developed AiCommentator, a Multimodal Conversational Agent (MCA) for embedded visualization and conversational interactions in football broadcast video. AiCommentator integrates embedded visualization, either with an automated non-interactive or with a responsive interactive commentary mode. Our system builds upon multimodal techniques, integrating computer vision and large language models, to demonstrate ways for designing tailored, interactive sports-viewing content. AiCommentator’s event system infers game states based on a multi-object tracking algorithm and computer vision backend, facilitating automated responsive commentary. We address three key topics: evaluating young adults’ satisfaction and immersion across the two viewing modes, enhancing viewer understanding of in-game events and players on the pitch, and devising methods to present this information in a usable manner. In a mixed-method evaluation (n=16) of AiCommentator, we found that the participants appreciated aspects of both system modes but preferred the interactive mode, expressing a higher degree of engagement and satisfaction. Our paper reports on our development of AiCommentator and presents the results from our user study, demonstrating the promise of interactive MCA for a more engaging sports viewing experience. Systems like AiCommentator could be pivotal in transforming the interactivity and accessibility of sports content, revolutionizing how sports viewers engage with video content.2024PAPeter Andrews et al.Intelligent Voice Assistants (Alexa, Siri, etc.)Social & Collaborative VRInteractive Data VisualizationIUI
Use of an AI-powered Rewriting Support Software in Context with Other Tools: A Study of Non-Native English SpeakersAcademic writing in English can be challenging for non-native English speakers (NNESs). AI-powered rewriting tools can potentially improve NNESs' writing outcomes at a low cost. However, whether and how NNESs make valid assessments of the revisions provided by these algorithmic recommendations remains unclear. We report a study where NNESs leverage an AI-powered rewriting tool, Langsmith, to polish their drafted academic essays. We examined the participants' interactions with the tool via user studies and interviews. Our data reveal that most participants used Langsmith in combination with other tools, such as machine translation (MT), and those who used MT had different ways of understanding and evaluating Langsmith's suggestions than those who did not. Based on these findings, we assert that NNESs' quality assessment in AI-powered rewriting tools is influenced by the simultaneous use of multiple tools, offering valuable insights into the design of future rewriting tools for NNESs.2023TITakumi Ito et al.Generative AI (Text, Image, Music, Video)AI-Assisted Creative WritingUIST
Who Does Not Benefit from Fact-checking Websites? A Psychological Characteristic Predicts the Selective Avoidance of Clicking Uncongenial FactsFact-checking messages are shared or ignored subjectively. Users tend to seek like-minded information and ignore information that conflicts with their preexisting beliefs, leaving like-minded misinformation uncontrolled on the Internet. To understand the factors that distract fact-checking engagement, we investigated the psychological characteristics associated with users’ selective avoidance of clicking uncongenial facts. In a pre-registered experiment, we measured participants’ (N = 506) preexisting beliefs about COVID-19-related news stimuli. We then examined whether they clicked on fact-checking links to false news that they believed to be accurate. We proposed an index that divided participants into fact-avoidance and fact-exposure groups using a mathematical baseline. The results indicated that 43% of participants selectively avoided clicking on uncongenial facts, keeping 93% of their false beliefs intact. Reflexiveness is the psychological characteristic that predicts selective avoidance. We discuss susceptibility to click bias that prevents users from utilizing fact-checking websites and the implications for future design.2023YTYuko Tanaka et al.Nagoya Institute of TechnologyVoice AccessibilityMisinformation & Fact-CheckingCHI
BirdViewAR: Surroundings-aware Remote Drone Piloting Using an Augmented Third-person PerspectiveWe propose BirdViewAR, a surroundings-aware remote drone-operation system that provides significant spatial awareness to pilots through an augmented third-person view (TPV) from an autopiloted secondary follower drone. The follower drone responds to the main drone's motions and directions using our optimization-based autopilot, allowing the pilots to clearly observe the main drone and its imminent destination without extra input. To improve their understanding of the spatial relationships between the main drone and its surroundings, the TPV is visually augmented with AR-overlay graphics, where the main drone's spatial statuses are highlighted: its heading, altitude, ground position, camera field-of-view (FOV), and proximity areas. We discuss BirdViewAR's design and implement its proof-of-concept prototype using programmable drones. Finally, we conduct a preliminary outdoor user study and find that BirdViewAR effectively increased spatial awareness and piloting performance.2023MIMaakito Inoue et al.Tohoku UniversityContext-Aware ComputingDrone Interaction & ControlCHI
WaddleWalls: Room-scale Interactive Partitioning System using a Swarm of Robotic PartitionsWe propose WaddleWalls, a room-scale interactive partitioning system using a swarm of robotic partitions that allows occupants to interactively reconfigure workspace partitions to satisfy their privacy and interaction needs. The system can automatically arrange the partitions' layout designed by the user on demand. The user specifies the target partition's position, orientation, and height using the controller's 3D manipulations. In this work, we discuss the design consideration of the interactive partition system and implement WaddleWalls' proof-of-concept prototype assembled with off-the-shelf materials. We demonstrate the functionalities of WaddleWalls through several application scenarios in an open-planned office environment. We also conduct an initial user evaluation that compares WaddleWalls with conventional wheeled partitions, finding that WaddleWalls allows effective workspace partitioning and mitigates the physical and temporal efforts needed to fulfill ad hoc social and privacy requirements. Finally, we clarify the feasibility, potential, and future challenges of WaddleWalls through an interview with experts.2022YOYuki Onishi et al.Domestic RobotsHuman-Robot Collaboration (HRC)UIST
ModularHMD: A Reconfigurable Mobile Head-Mounted Display Enabling Ad-hoc Peripheral Interactions with the Real WorldWe propose ModularHMD, a new mobile head-mounted display concept, which adopts a modular mechanism and allows a user to perform ad-hoc peripheral interaction with real-world devices or people during VR experiences. ModularHMD is comprised of a central HMD and three removable module devices installed in the periphery of the HMD cowl. Each module has four main states: occluding, extended VR view, video see-through (VST), and removed/reused. Among different combinations of module states, a user can quickly setup the necessary HMD forms, functions, and real-world visions for ad-hoc peripheral interactions without removing the headset. For instance, an HMD user can see her surroundings by switching a module into the VST mode. She can also physically remove a module to obtain direct peripheral visions of the real world. The removed module can be reused as an instant interaction device (e.g., touch keyboards) for subsequent peripheral interactions. Users can end the peripheral interaction and revert to a full VR experience by re-mounting the module. We design ModularHMD’s configuration and peripheral interactions with real-world objects and people. We also implement a proof-of-concept prototype of ModularHMD to validate its interactions capabilities through a user study. Results show that ModularHMD is an effective solution that enables both immersive VR and ad-hoc peripheral interactions.2021IEIsamu Endo et al.Mixed Reality WorkspacesImmersion & Presence ResearchUIST
Can Playing with Toy Blocks Reflect Behavior Problems in Children?Although children’s behavioral and mental problems are generally diagnosed in clinical settings, the prediction and awareness of children’s mental wellness in daily settings are getting increased attention. Toy blocks are both accessible in most children’s daily lives and provide physicality as a unique non-verbal channel to express their inner world. In this paper, we propose a toy block approach for predicting a range of behavior problems in young children (4-6 years old) measured by the Child Behavior Checklist (CBCL). We defined and classified a set of quantitative play actions from IMU-embedded toy blocks. Play data collected from 78 preschoolers revealed that specific play actions and patterns indicate total problems, internalizing problems, and aggressive behavior in children. The results align with our qualitative observations, and suggest the potential of predicting the clinical behavior problems of children based on short free-play sessions with sensor-embedded toy blocks.2021XWXiyue Wang et al.Tohoku UniversityHuman Pose & Activity RecognitionSpecial Education TechnologyBiosensors & Physiological MonitoringCHI
PinpointFly: An Egocentric Position-control Drone Interface using Mobile ARAccurate drone positioning is challenging because pilots only have a limited position and direction perception of a flying drone from their perspective. This makes conventional joystick-based speed control inaccurate and more complicated and significantly degrades piloting performance. We propose PinpointFly, an egocentric drone interface that allows pilots to arbitrarily position and rotate a drone using position-control direct interactions on a see-through mobile AR where the drone position and direction are visualized with a virtual cast shadow (i.e., the drone's orthogonal projection onto the floor). Pilots can point to the next position or draw the drone's flight trajectory by manipulating the virtual cast shadow and the direction/height slider bar on the touchscreen. We design and implement a prototype of PinpointFly for indoor and visual line of sight scenarios, which are comprised of real-time and predefined motion-control techniques. We conduct two user studies with simple positioning and inspection tasks. Our results demonstrate that PinpointFly makes the drone positioning and inspection operations faster, more accurate, simpler and fewer workload than a conventional joystick interface with a speed-control method.2021LCLinfeng Chen et al.Tohoku University, Tohoku UniversityAR Navigation & Context AwarenessDrone Interaction & ControlCHI
TiltChair: Manipulative Posture Guidance by Actively Inclining the Seat of an Office ChairWe propose TiltChair, an actuated office chair that physically manipulates the user's posture by actively inclining the chair's seat to address problems associated with prolonged sitting. The system controls the inclination angle and motion speed with the aim of achieving manipulative but unobtrusive posture guidance. To demonstrate its potential, we first built a prototype of TiltChair with a seat that could be tilted by pneumatic control. We then investigated the effects of the seat's inclination angle and motions on task performance and overall sitting experience through two experiments. The results show that the inclination angle mainly affects the difficulty of maintaining one's posture, while the motion speed affected the conspicuousness and subjective acceptability of the motion. However, these seating conditions did not affect objective task performance. Based on these results, we propose a design space for facilitating effective seat-inclination behavior using the three dimensions of angle, speed, and continuity. Furthermore, we discuss promising applications.2021KFKazuyuki Fujita et al.Tohoku UniversityForce Feedback & Pseudo-Haptic WeightWorkplace Wellbeing & Work StressCHI
ZoomWalls: Dynamic Walls that Simulate Haptic Infrastructure for Room-scale VR WorldWe focus on the problem of simulating the haptic infrastructure of a virtual environment (i.e. walls, doors). Our approach relies on multiple ZoomWalls—autonomous robotic encounter-type haptic wall-shaped props—that coordinate to provide haptic feedback for room-scale virtual reality. Based on a user’s movement through the physical space, ZoomWall props are coordinated through a predict-and-dispatch architecture to provide just-in-time haptic feedback for objects the user is about to touch. To refine our system, we conducted simulation studies of different prediction algorithms, which helped us to refine our algorithmic approach to realize the physical ZoomWall prototype. Finally, we evaluated our system through a user experience study, which showed that participants found that ZoomWalls increased their sense of presence in the VR environment. ZoomWalls represents an instance of autonomous mobile reusable props, which we view as an important design direction for haptics in VR.2020YYYAN YIXIAN et al.Shape-Changing Interfaces & Soft Robotic MaterialsMixed Reality WorkspacesUIST
Third-Person Piloting: Increasing Situational Awareness using a Spatially Coupled Second DroneWe propose Third-Person Piloting, a novel drone manipulation interface that increases situational awareness using an interactive third-person perspective from a second, spatially coupled drone. The pilot uses a controller with a manipulatable miniature drone. Our algorithm understands the relationship between the pilot’s eye position and the miniature drone and ensures that the same spatial relationship is maintained between the two real drones in the sky. This allows the pilot to obtain various third-person perspectives by changing the orientation of the miniature drone while maintaining standard the primary drone control using the conventional controller. We design and implement a working prototype with programmable drones and propose several representative operation scenarios. We gather user feedback to obtain initial insights of our interface design from novices, advanced beginners, and experts. Result shows that our interface was positively evaluated by all of them, and their feedback suggests the additional interactive third-person perspective increases spatial awareness and helps their primary drone manipulation.2019RTRyotaro Temma et al.Drone Interaction & ControlTeleoperation & TelepresenceUIST
Asian CHI Symposium: Emerging HCI Research CollectionThis symposium showcases the latest work from Asia on interactive systems and user interfaces that address under-explored problems and demonstrate unique approaches. In addition to circulating ideas and sharing a vision of future research in human-computer interaction, this symposium aims to foster social networks among academics (researchers and students) and practitioners and create a fresh research community from Asian region.2018SSSaki Sakaguchi et al.The University of TokyoDeveloping Countries & HCI for Development (HCI4D)User Research Methods (Interviews, Surveys, Observation)CHI