Generative Muscle Stimulation: Providing Users with Physical Assistance by Constraining Multimodal-AI with Embodied KnowledgeElectrical-muscle-stimulation (EMS) can support physical-assistance (e.g., shaking a spray-can before painting). However, EMS-assistance is highly-specialized because it is (1) fixed (e.g., one program for shaking spray-cans, another for opening windows); and (2) non-contextual (e.g., a spray-can for cooking dispenses cooking-oil, not paint—shaking it is unnecessary). Instead, we explore a different approach where muscle-stimulation instructions are generated considering the user’s context (e.g., pose, location, surroundings). The resulting system is more general—enabling unprecedented EMS-interactions (e.g., opening a pill-bottle) yet also replicating existing systems (e.g., Affordance++) without task-specific programming. It uses computer-vision/large-language-models to generate EMS-instructions, constraining these to a muscle-stimulation knowledge-base & joint-limits. In our user-study, we found participants successfully completed physical-tasks while guided by generative-EMS, even when EMS-instructions were (purposely) erroneous. Participants understood generated-gestures and, even during forced-errors, understood partial-instructions, identified errors, and re-prompted the system. We believe our concept marks a shift toward more general-purpose EMS-interfaces.2026YHYun Ho et al.University of ChicagoElectrical Muscle Stimulation (EMS)Generative AI (Text, Image, Music, Video)Human-LLM CollaborationCHI
Digital Companionship: Overlapping Uses of AI Companions and AI AssistantsLarge language models are increasingly used for both task-based assistance and social companionship, yet research has typically focused on one or the other. Drawing on a survey (N = 202) and 30 interviews with high-engagement ChatGPT and Replika users, we characterize digital companionship as an emerging form of human-AI relationship. With both systems, users were drawn to humanlike qualities, such as emotional resonance and personalized responses, and non-humanlike qualities, such as constant availability and inexhaustible tolerance. This led to fluid chatbot uses, such as Replika as a writing assistant and ChatGPT as an emotional confidant, despite their distinct branding. However, we observed challenging tensions in digital companionship dynamics: participants grappled with bounded personhood, forming deep attachments while denying chatbots "real" human qualities, and struggled to reconcile chatbot relationships with social norms. These dynamics raise questions for the design of digital companions and the rise of hybrid, general-purpose AI systems.2026AMAikaterina Manoli et al.Max Planck Institute for Human Cognitive and Brain SciencesHuman-LLM CollaborationAgent Personality & AnthropomorphismAffective Human-Computer DialogueCHI
Increasing Input Accuracy of Embodied Devices via Electrical Muscle StimulationThis paper evaluates interaction techniques to increase input accuracy with embodied devices—an emergent type of interactive system where the user's body serves as both the input and output medium (e.g., gestural input via cameras/IMUs; gestural output via motors/muscle stimulation). A shortcoming of existing embodied devices is their failure to enforce alignment between users' proprioceptive inputs and interface state. Thus, we present and evaluate interaction techniques that use muscle stimulation to enable embodied devices to: (1) recall previous interface states; (2) provide confirmation cues on state transitions; and (3) constrain inputs to valid ranges. In our study, participants performed pairs of interactions with an embodied slider, separated by a distraction task. The results showed that, compared to the same embodied slider without EMS, the combination of our techniques increased users': (1) absolute input accuracy; (2) relative input accuracy; and (3) confidence.2026LCLonnie Chien et al.University of ChicagoElectrical Muscle Stimulation (EMS)Hand Gesture RecognitionForce Feedback & Pseudo-Haptic WeightCHI
Attention Nooks: Situated Frictions to Foster Intentional Technology UseAs persuasive technologies weave themselves deeper into the fabric of domestic life, the challenge of sustaining digital wellbeing grows increasingly entangled with the spaces and rhythms of everyday living. Conventional Digital Self-Control Tools (DSCTs), while offering momentary reprieve, often falter under sustained use, revealing a gap between device-centric interventions and the situated nature of technology habits. In resistance, we present Attention Nooks: a set of spatial interventions that deploy "situated frictions" within the home. Attention Nooks recast digital wellbeing as a lived negotiation of spatial boundaries in the home. Developed through an autobiographical design process, we surface design events that shaped our making and living with our prototypes. We discuss the teleological nature of interventions, implications for ubiquitous computing, and the subversion of ethically ambiguous technologies. Our contribution lies in reframing digital wellbeing as a design opportunity that calls for pluralistic situated encounters in the home.2026ASAnup Sathya et al.University of ChicagoUbiquitous ComputingSmartphone Addiction & Digital WellbeingAging-in-Place Assistance SystemsCHI
Modeling Perceived Force of Electrical Muscle Stimulation to Improve User’s RecallInteractive electrical-muscle-stimulation (EMS) supports motor- skills by actuating the user’s muscles. However, existing EMS- interfaces exclusively focus on demonstrating movements/sequences (e.g., which fingers to actuate to play a piano melody) and have not investigated EMS for skills requiring precise force application (e.g., playing musical instruments, practicing culinary techniques, operating force-sensitive tools). Our user study found that when EMS-interfaces demonstrate a force, participants trying to recall this force, overshoot by a median 19%; with especially larger over- shoots at lower target-forces (e.g., produce a∼1.2 kg force, after a 1 kg demonstration). This force mismatch renders EMS-interfaces unable to accurately demonstrate forces—drastically limiting the growing potential of EMS for HCI. To significantly improve on this, we modeled users’ recall of EMS-demonstrated forces. This model allows to adjust EMS-interfaces to render a target force that, when recalled, matches the intended force best—in our study, this improved median force recall by∼35%.2026MGMithil Guruvugari et al.University of ChicagoElectrical Muscle Stimulation (EMS)Force Feedback & Pseudo-Haptic WeightVibrotactile Feedback & Skin StimulationCHI
Underreporting of AI Use: The Role of Social Desirability BiasRapid integration of artificial intelligence (AI) into work and educational settings challenges organizations to gauge and respond to adoption rates. However, most measures of AI adoption come from self-reported surveys, producing estimates of AI use that disagree by up to 40 percentage points within the same setting. We investigate whether social desirability bias—the tendency to answer surveys in ways that would be viewed favorably by an outside party—can explain this discrepancy. Surveying 338 university students, we assess potential social desirability bias using a method from psychology, indirect questioning: students report both their own AI use and that of their peers. We find a significant gap, with approximately 60% of students reporting that they use AI compared to 90% of their peers. Through qualitative analysis of student explanations for this gap, we conclude that social desirability bias is a key driver of mis-measurement, causing underestimates of AI adoption in educational settings.2026YLYier Ling et al.University of ChicagoHuman-LLM CollaborationAI Ethics, Fairness & AccountabilityUser Research Methods (Interviews, Surveys, Observation)CHI
Beyond PII: How Users Attempt to Estimate and Mitigate Implicit LLM InferenceLarge Language Models (LLMs) such as ChatGPT can infer personal attributes from seemingly innocuous text, raising privacy risks beyond memorized data leakage. While prior work has demonstrated these risks, little is known about how users estimate and respond. We conducted a survey with 240 U.S. participants who judged text snippets for inference risks, reported concern levels, and attempted rewrites to block inference. We compared their rewrites with those generated by ChatGPT and Rescriber, a state-of-the-art sanitization tool. Results show that participants struggled to anticipate inference, performing a little better than chance. User rewrites were effective in just 28\% of cases - better than Rescriber but worse than ChatGPT. We examined our participants' rewriting strategies, and observed that while paraphrasing was the most common strategy it is also the least effective; instead abstraction and adding ambiguity were more successful. Our work highlights the importance of inference-aware design in LLM interactions.2026SWSynthia Qia Wang et al.University of ChicagoExplainable AI (XAI)Privacy by Design & User ControlPrivacy Perception & Decision-MakingCHI
Understanding Parents’ Desires in Moderating Children’s Interactions with GenAI Chatbots through LLM-Generated ProbesThis paper studies how parents want to moderate children’s interactions with Generative AI Chatbots, with the goal of informing the design of future GenAI parental control tools. We first used an LLM to generate synthetic Child--GenAI Chatbot interaction scenarios and worked with four parents to validate their realism. From this dataset, we carefully selected 12 diverse examples that evoked varying levels of concern and were rated the most realistic. Each example included a prompt and GenAI Chatbot response. We presented these to parents (N=24) and asked whether they found them concerning, why, and how they would prefer to modify the responses and be informed. Our findings reveal three key insights: (1) parents express concern about interactions that current GenAI Chatbot parental controls neglect; (2) parents want fine-grained transparency and moderation at the conversation level; and (3) parents need personalized controls that adapt to their desired strategies and children's ages.2026JDJohn Driscoll et al.University of California San DiegoConversational ChatbotsMental Health Technology for YouthChildren's AI Literacy & Data LiteracyCHI
Next Generation Wearable Haptics Should Balance Virtual & Real-world FidelityProviding tactile-feedback when users contact virtual-interfaces has been a seminal advance. However, we posit these advances have been explored in isolation from considerations of users’ physical interactions with surrounding-objects. Most touch-interfaces were designed to optimize virtual interfaces, but rarely consider that users also need to feel physical interfaces (e.g., tools, putting on/off headsets). We argue against this being the sole design-objective driving haptic-interfaces; instead, we propose also to optimize the fidelity of the real-world sensations that users feel while wearing a haptic device. We propose a framework to classify touch-devices by measuring not only their abilities to deliver virtual-feedback but also how much they impair physical-feedback—we argue this balancing act is an urgent mainstream need, given the success of Mixed-Reality. Thus, to accelerate the research in this area, we synthesize existing techniques into new conceptual-categories: feel-through, on-demand, relocated, and remote actuators. Finally, we present their pros/cons and discuss a possible roadmap.2026STShan-Yuan Teng et al.National Taiwan UniversityMid-Air Haptics (Ultrasonic)Haptic WearablesImmersion & Presence ResearchCHI
Exploring Texture-Level Creative Decisions with penPal, a Novel Handheld Actuated Drawing ToolThis paper looks at texture---middle-level components---as an important aspect of drawing. We present a hardware tool, penPal, that is designed to support dynamic mark-making and direct creative actions at this level. By incorporating a tendon-driven continuum robot, penPal’s tip can move independently, giving the user a new axis of creative control. Combined with the user’s own manipulations, penPal allows for emergent combinations of computer and manual control over the rapid generation of diverse textures. Through a 10-participant study and a professional artist commission, we examine how users negotiate control by integrating multiple coordinate systems (their body, the paper, and penPal’s tip) as they construct compositions. We suggest some benefits of supporting users at the texture level, such as the ability to shift the primary focus of their activity, the ability to selectively defamiliarize the creative process for generative potential, and for pleasure.2026TRTucker Rae-Grant et al.University of ChicagoShape-Changing Interfaces & Soft Robotic Materials3D Modeling & AnimationTangible User Interface DesignCHI
Writing with AI Can Reduce Gender Bias in Hiring EvaluationsWomen remain underrepresented in the workplace, partly due to stereotypes associating competence traits with men rather than women. Efforts to change such stereotypes often yield mixed results. As language models become integrated into daily life, AI writing assistants offer an opportunity to shift gender images. In a preregistered experiment (N=672), participants evaluated résumés for a female ("Jennifer") and a male ("John") candidate applying to a financial analyst role. They wrote evaluations using AI-generated suggestions in one of three conditions: suggestions for Jennifer integrated stereotypically male, female, or neutral traits. Suggestions for John remained neutral. Participants exposed to male-trait suggestions evaluated Jennifer as more competent, selected her as the leader, and offered higher salaries. However, we also observed signs of backlash: participants were less willing to work with competent Jennifer. We discuss implications for designing AI writing assistants to mitigate gender bias in hiring contexts.2026ALAlicia T.H. Liu et al.University of ChicagoGenerative AI (Text, Image, Music, Video)AI Ethics, Fairness & AccountabilityAlgorithmic Fairness & BiasCHI
Overcoming Translation Delays: Towards Better Subtitle Design for Foreign Language Conversations in Extended RealityIn multilingual conferences, translation support should not compromise non‑verbal cues or social interaction. Prior work on eXtended Reality (XR) subtitles aids comprehension but rarely examines translation latency. We conducted a VR-simulated conference, testing latencies of 0, 1.5, 3, 4.5, and 6 seconds to measure overall comprehension and attribution of verbal and non‑verbal information. Results showed that latencies beyond 3 seconds significantly increased subjective difficulty and affected accuracy, while shorter latencies showed no significant effects. Furthermore, participants noted that very low delay drew attention to subtitles, reducing opportunities to observe the speaker. Guided by these insights, we designed and evaluated four VR subtitle interfaces, including one traditional and three novel designs. Across delay conditions, Merged Subtitles improved opportunities to observe the speaker and resulted in better emotion attribution and user experience than other designs. We also proposed design guidelines for XR subtitle interfaces based on different levels of translation latency.2026ZLZiming Li et al.Hong Kong University of Science and Technology (GZ)Multilingual & Cross-Cultural Voice InteractionAR Navigation & Context AwarenessImmersion & Presence ResearchCHI
BloomBeacon: Blooming Physical Touch Display Surfaces via Persistence-of-Vision MotionWe explore how a display surface can physically emerge on demand to support both mid-air visualization and direct touch interaction. We introduce blooming, a concept that repurposes Persistence-of-Vision (POV) motion to deploy a large, touchable surface from a compact, relocatable device. Using a soft, rotating line with arch-shaped electrodes, our system renders dynamic mid-air visuals while enabling direct touch input on the manifested surface. Realizing this concept requires addressing the unique challenges of touching spinning elements, including ensuring safety, minimizing disturbances to rotation caused by touch, and detecting brief unstable touches during spinning. We present a safety-oriented device design, special blades effective in minimizing finger disturbance, and optimization techniques tailored to transient, noisy touches. We also reveal how rotation speed and electrode height significantly affect sensing accuracy and user experience. Finally, we demonstrate applications that show how blooming touch displays can flexibly augment everyday objects and environments.2026WYWilla Yunqi Yang et al.University of ChicagoPhysical-Digital Hybrid InteractionTangible User Interface DesignShape-Changing Interfaces & Soft Robotic MaterialsCHI
Myo Action: Accelerating Voluntary Actions via Electromyography and Muscle StimulationWe propose a technique for accelerating users’ action without overriding intention, thereby preserving agency. In our approach, it is the user’s muscle signals, detected via electromyography (EMG), that trigger electrical-muscle-stimulation (EMS) without external sensors or stimulation-timing calibration. The key to enable this “agentic speedup” is a synergy between EMG and EMS: EMG can detect an early-onset of the neural-response; EMS can contract a muscle faster than a typical voluntary-contraction. This–coupled with our low-latency system (~290 us)–results in an accelerated reaction-time, even though the haptic-assistance is initiated after the muscle-signal. In our study, we confirmed that our novel approach: (1) accelerated users’ reaction-time by ~23 ms compared to voluntary-action; (2) preserved agency in decision-involving actions (i.e., go/no-go trials), which existing muscle-stimulation techniques cannot achieve; and (3) participants felt it augmented their performance in physical-tasks. This puts forward embodied-assistance that aligns with users’ decisions/intentions, which we demonstrate in exemplary applications.2026YTYudai Tanaka et al.University of ChicagoElectrical Muscle Stimulation (EMS)Force Feedback & Pseudo-Haptic WeightEmotion-Sensing WearablesCHI
Investigating the Effects of LLM Use on Critical Thinking Under Time Constraints: Access Timing and Time AvailabilityThe impact of large language models (LLMs) on critical thinking has provoked growing attention, yet this impact on actual performance may not be uniformly negative or positive. Particularly, the role of time---the temporal context under which an LLM is provided---remains overlooked. In a between-subjects experiment (n=393), we examined two types of time constraints for a critical thinking task requiring participants to make a reasoned decision for a real-world scenario based on diverse documents: (1) LLM access timing---an LLM available only at the beginning (early), throughout (continuous), near the end (late), or not at all (no LLM), and (2) time availability---insufficient or sufficient time for the task. We found a temporal reversal: LLM access from the start (early, continuous) improved performance under time pressure but impaired it with sufficient time, whereas beginning the task independently (late, no LLM) showed the opposite pattern. These findings demonstrate that time constraints fundamentally shape whether an LLM augments or undermines critical thinking, making time a central consideration when designing LLM support and evaluating human-AI collaboration in cognitive tasks.2026JZJiayin Zhi et al.University of ChicagoHuman-LLM CollaborationAI-Assisted Decision-Making & AutomationUser Research Methods (Interviews, Surveys, Observation)CHI
Aesthetics of Felt AsymmetryOur bodies mediate every interaction with technology, yet—as soma design and feminist HCI remind us—the body is not a neutral canvas. We introduce and examine felt asymmetries—somaesthetic experiences of difference in the body—as a site for generative and critical engagement in interaction design. Through an autobiographical design exploration, and a series of somatic explorations with nine designers including individual inquiries and workshops, we sensitised to, articulated, and shared personal experiences of asymmetry. We draw from these explorations to contribute: (1) Opening a design space exploring the aesthetics of felt asymmetries; (2) Reflections on engaging with asymmetry in design, e.g. as a design material, an estrangement activity or doorway into intimate experience; (3) Considerations for creating technologies that resonate with, rather than erase, the asymmetries of lived experience. We argue that bodily asymmetries are not only to be accommodated in design, but embraced as aesthetic resources—sources of joy, tension, and creativity.2026AHAlice C Haynes et al.KTH Royal Institute of TechnologyTechnology Ethics & Critical HCIParticipatory DesignPrototyping & User TestingCHI
A Systematic Review of User Experiments on the Effects of Dark PatternsDeceptive/Manipulative Patterns (DMP) are interface designs, also known as "dark patterns," that manipulate user behavior. While considerable attention has been paid to their ethical and legal implications, empirical evidence about their real-world effects remains diffuse. This review synthesizes up-to-date experimental studies, focusing on works that quantify how (or whether) DMPs influence users. We also aggregate findings on interventions aimed at reducing DMP effects. Our synthesis highlights the experimental agreement that DMPs do significantly alter user behavior (with large variance in effect size) and that external interventions have been mostly unsuccessful in mitigating their effects. Lastly, we show that significant correlations between DMP effects and personal characteristics (e.g., age or political affiliation) are uncommon, indicating DMPs similarly affected nearly all populations tested. By summarizing the experimental evidence, we clarify the effects of DMPs, highlight gaps and tensions in the existing experimental literature, and help inform ongoing research and policy directions.2026BSBrennan Schaffner et al.Georgetown UniversityDark Patterns RecognitionPrivacy Perception & Decision-MakingCHI
Mental Models of Autonomy and Sentience Shape Reactions to AINarratives about artificial intelligence (AI) entangle autonomy, the capacity to self-govern, with sentience, the capacity to sense and feel. AI agents that perform tasks autonomously and companions that recognize and express emotions may activate mental models of autonomy and sentience, respectively, provoking distinct reactions. To examine this possibility, we conducted three pilot studies (N = 374) and four preregistered vignette experiments describing an AI as autonomous, sentient, both, or neither (N = 2,702). Activating a mental model of sentience increased general mind perception (cognition and emotion) and moral consideration more than autonomy, but autonomy increased perceived threat more than sentience. Sentience also increased perceived autonomy more than vice versa. Based on a within-paper meta-analysis, sentience changed reactions more than autonomy on average. By disentangling different mental models of AI, we can study human-AI interaction with more precision to better navigate the detailed design of anthropomorphized AI and prompting interfaces.2026JPJanet V.T. Pauketat et al.Sentience InstituteAgent Personality & AnthropomorphismExplainable AI (XAI)AI Ethics, Fairness & AccountabilityCHI
Water On My Block: Reflections on Building A Participatory Artificial Intelligence System For Precision Weather With Scientists and An Urban CommunityCreating accurate hyper-local climate Artificial Intelligence (AI) models requires neighborhood-level weather measurements and community partnerships. In this paper, we describe a three year case study of using a participatory approach to support the creation of hyper-local climate AI models, or what we term ``precision weather.'' Using participatory design to involve stakeholders in the climate AI pipeline design process i.e., ``participatory AI,'' we collaborated with a national laboratory and a community organization in a major metropolitan area in the United States, working with community members and scientists. We held interviews, co-design workshops (``Community Cafes''), and created an app for the community to collect flood reports in their neighborhood for advocacy and to contribute data to the AI model pipeline. We discuss our findings, lessons learned, and implications for future participatory projects to support hyper-local climate AI.2026KWKelly B. Wagman et al.University of ChicagoGenerative AI (Text, Image, Music, Video)Participatory DesignSmart Cities & Urban SensingCHI
Evaluating Peer Fact-Checking on WhatsAppPrivate messaging platforms hinder public oversight, making misinformation hard to counter. Meanwhile, platforms are pivoting to crowdsourced verification amid waning trust in institutional fact-checkers. This raises a critical question: how do peer corrections compare with local journalists or fact-checking tiplines? We tested this via a privacy-preserving randomized field study on participants' real WhatsApp group messages in India, complemented by interviews. Fact-checks from a close contact significantly improved accuracy over the control group, while corrections from the local journalist and national tipline did not reach statistical significance. However, none of the interventions improved participants' ability to identify novel misinformation on similar themes, suggesting corrections on WhatsApp are context-bound rather than skill-building. We contribute: (1) the first ecologically valid randomized test of peer-led fact-checking on WhatsApp, benchmarked against journalists and tiplines; (2) an empirical account of how participants make sense of corrections in closed messaging environments; and (3) design implications for community-based fact-checking, including training high-social-capital individuals as embedded verifiers.2026SHSudhamshu Hosamane et al.Rutgers UniversityMisinformation & Fact-CheckingCommunity Collaboration & WikipediaCHI