Rosalina Ribeiro Silva, Zichen Feng, Rikard Olajos, Tim Rolff
原文摘要
We propose an immersive virtual-reality game designed to enable multi-user, multi-language learning during co-play. We hypothesize that our method will be more engaging, leading to better rhythm, pronunciation, memorization, and speaking confidence. We implement the experience as a fantasy role-playing game by introducing vocabulary and phrases to the user as spells to learn. Players then cast more efficient spells depending on their language skills. For evaluation, we compare the game to a single-user baseline that replicates traditional language learning in a non-immersive setting. We measure short- and long-term retention of vocabulary and phrases, as well as scoring rhythm and pronunciation. Furthermore, we collect user data on engagement and mental workload.
Until recently, the biggest obstacles to natural spoken interaction with social robots were speech recognition errors and turn-taking failures. Advances in speech recognition have enabled much better spoken interaction, but turn-taking failures and other misunderstandings continue to be far too prevalent. This short paper looks back at earlier methods for improving speech recognition and turn-taking in interaction with social robots, and then looks at what can be done today, arguing that multimodal bidirectional streaming is needed to reduce the frequency of turn-taking failures.
Solmaz Goodarzi, Arseni Ivanov, David Maruscsák, Diana Quitéria Sousa
原文摘要
Floods are among the most frequent and destructive natural disasters, requiring citizens to make critical decisions under rapidly changing conditions. Most serious games related to flooding emphasize awareness, preparedness, and education in flood management, but there is still a research gap on how to train citizens to act under pressure, uncertainty, and stress. In this paper, we position Virtual Reality (VR) serious games as a promising approach to addressing this gap by providing experiential and immersive training opportunities for citizens. We propose a conceptual VR serious game that enables participants to experience realistic flood scenarios, practice decision-making, rehearse appropriate response behaviors, and develop response-related skills in a safe and controlled environment. In addition, we suggest a research agenda and an evaluation framework focused on flood risk awareness, response capabilities, and stress-related outcomes. The framework thoughtfully integrates subjective, objective, and physiological measures to assess the effectiveness of immersive flood-response training. Finally, we present design, methodological, ethical, and user-experience considerations in VR-based disaster training, and outline future research directions for the development and evaluation of immersive serious games for flood emergency response. The present study aims to contribute to the growing intersection of serious games, extended reality, and disaster risk reduction by establishing citizen-centered response training as a significant yet under-researched field.
Personal email agents decide, for every message, its type and urgency and whether to draft a reply, create a task or ignore it. In production these decisions are delegated to a generative LLM or to rules, and their error rate is rarely measured: the stored labels were written by the system itself. We describe Foyer, a self-hosted personal inbox agent, and a measurement protocol (stratified sampling, frozen snapshots, thread-grouped splits, secret redaction, human labels in the review loop) that produces ground truth before any training. On 231 emails in Italian and English we compare eleven systems: the stored production type, a TF-IDF + logistic-regression baseline, local and hosted typed decision models (Tev1, Nimble, Jeva, Jev) and generative LLMs. Four results. The stored production type scores 28% against a 57% majority class. With about two hundred labels a logistic regression has the highest point accuracy on every field (type 0.77 +/- 0.02 over five seeds) at a thousandth of the latency, but on every field at least one hosted model is statistically indistinguishable from it (exact McNemar, Holm-corrected). A generative model recovers part of the sender's relationship to the owner from the text (0.60); inheritance from the contact is right 98.6% of the time where the sender was seen before: it belongs in the contact graph, not in the prompt. Per-email accuracy is not the product metric: composed into actions, the classifier ignores 18% of the emails that needed action, against 4% for the best generative model; a cost-chosen threshold cuts that to 10%, indistinguishable from the hosted decision model's; on the adversarial corpus the thresholded classifier is the cheapest system above a missed-email price of 4. Two blind annotators agree with the reference at kappa 0.81 and 0.86 on type. v4 corrects threshold results computed on a truncated corpus in v1-v3.
Evaluations of Wikipedia's programs for new editors usually ask whether a program works on average; we ask which newcomers benefit when their question to a mentor gets a reply. English Wikipedia's Growth mentorship system assigns every new account a volunteer mentor, and about a quarter of the questions newcomers send never get a reply. Using the first questions of 34,073 newcomers (2021 to 2026), we compare newcomers who received a reply with newcomers who did not, within groups that are similar on 164 characteristics recorded before the question (propensity score stratification). (1) Does a reply help newcomers keep editing? Yes, in the short term: in the 14 days after the reply, newcomers who received one were 1.8 percentage points (pp) more likely to edit an article (22% vs. an estimated 20% without a reply) and 1.8pp more likely to edit on two or more days. Later in the first year, the only difference that remains significant after correction for multiple testing is in the share who edit an article at least once between six months and one year after the reply (+1.5pp). (2) Does the effect depend on what the newcomer asked? Yes. We sort questions by Morrison's (1993) information-seeking types, which earlier studies measured mostly through self-reports, using a language model checked against human labels. A reply has significant effects for questions about rules and norms (+3.4pp), questions that show the newcomer's earlier editing (+3.4pp) and specific how-to questions (+2.4pp). For newcomers who do not know what to work on, a reply has no detectable effect, and the effect is 4.2pp smaller than for newcomers with other questions. (3) Which replies are followed by more editing? Among newcomers who received a reply, those whose reply came sooner or stayed closer to the topic of their question edited more.
Riya Sahni, Aryan Kaul, Anushka Bhatnagar, Xuanming Zhang, Vivian Liu, Reya Vir, Lydia B. Chilton
原文摘要
Coding agents let developers modify software rapidly through natural language. However, as agents take on larger changes, developers can lose track of how the codebase is evolving. These edits may look correct on the surface while restructuring the system in unexpected ways: duplicating logic, changing modules outside the requested scope, removing functionality developers meant to keep, or introducing architectural drift. We investigate how visual diffs can help developers understand and evaluate the structural changes that coding agents make. We present Rasoi, a design probe that maps how a codebase's architecture evolves after each agent-made edit, letting developers inspect which architectural elements and relationships were added, modified, or removed across multiple levels of abstraction. We deployed Rasoi in a week-long study with 12 developers working on their own codebases across varied technical domains. We found that visual diffs helped developers see whether changes stayed within scope, notice unexpected restructuring, and understand how modified parts related to the system as a whole. Our findings suggest that the value of oversight tools for coding agents lies less in making developers trust the agent and more in helping developers feel confident in their own evaluation of agent-made changes. We discuss implications for visual tools that support code comprehension and human oversight in AI-assisted software development.
As AI capabilities advance, early-stage teams need better ways to judge whether historical incidents transfer to new products, demanding support beyond static risk checklists or one-off ideation tools. We present RiskLens, a human-AI workflow with an interactive system built around a live crawl, refreshable ranker, and extensible, human-informed schema. RiskLens analyzes 2,034 AI incidents and 10,443 AI products to rank transferable incidents using a feature schema spanning technical, operational, and user-facing attributes and supports an interface that presents ranked incident evidence, generated scenarios, risk metrics, and mitigation suggestions. In a study with 21 AI-familiar participants, users surfaced evaluation criteria that we added as 13 human-informed dimensions, including harm categories, failure mechanisms, and vulnerable groups. The feature-based model outperformed a generic baseline with an 86.2% win rate, and remained stable on a twenty-month future holdout; after human-informed schema expansion, the updated model outperformed the same baseline on 91.4% of pairs.
Sam Yu-Te Lee, Yan To Linus Lam, Manami Nakagawa, Kwan-Liu Ma
原文摘要
Graph Retrieval-Augmented Generation (GraphRAG) grounds Large Language Model (LLM) responses in structured, domain-specific knowledge. However, verification remains difficult because errors can arise at multiple layers: an answer may not be faithful to its retrieved context, the retrieval pipeline may select irrelevant evidence, and the knowledge graph may omit relevant facts or contain incorrect triplets. We present VArify, a visual analytics system for food-science researchers to inspect and repair these layers in a GraphRAG workflow. A formative study with five participants using a conventional chat-and-table interface revealed behaviors reflecting answer anchoring, over-trust, and difficulty tracing multi-hop relationships. In response, VArify combines an evidence-first answer reveal with coordinated chat and hierarchical graph views, provenance inspection, and subquery editing. It connects the generated answer, the retrieval state, and the provenance-linked graph so that researchers can trace discrepancies across them and repair the retrieved context. From a qualitative system-use study on six food science researchers, we characterize how they scrutinize GraphRAG evidence with progressive visual inspection. Our study also reveals an important verification boundary: information absent from both retrieval and a user's knowledge remains unverifiable. We discuss the broader role of visualization in supporting oversight of LLM responses and outline future research directions.
Sam Yu-Te Lee, Jingya Chen, Albert Calzaretto, Richard Lee, Samir Passi, Alice Ferng, Mihaela Vorvoreanu
原文摘要
Enterprise AI chatbots show promise in supporting knowledge workers in information synthesis tasks by retrieving context from large, heterogeneous databases before generating answers. However, when the retrieved context misaligns with user intentions, these chatbots often produce ``irrelevantly right'' responses that raise the risk of overreliance. To foster appropriate reliance, we need a different approach than asking users to verify chatbot outputs. We posit that human oversight and modification of the context retrieval process can improve the quality of chatbots' outputs and users' awareness, resulting in more appropriate reliance and higher accountability. In this work, we explore this approach through a Research-through-Design study using VizCopilot, a chatbot extended with corpus visualization techniques. Through a within-subject study comparing VizCopilot and a simplified pure-text chatbot, we evaluate the role of visualization in enterprise chatbots and surface future design opportunities for fostering appropriate reliance through context engineering. Our findings show that visualization not only helps users detect and correct misaligned context, but also enhances user agency and system transparency. At the same time, the study reveals limitations regarding verification support and trust in AI summaries. Based on these findings, we discuss how visualization can foster appropriate reliance, the associated design trade-offs, and our vision towards transparent and sustainable human-AI interaction.
Garrick Chan, Yun-Hsin Kuo, Sam Yu-Te Lee, Yinuo Tang, Kwan-Liu Ma
原文摘要
Communicative visualizations can support multiple chart-grounded takeaways, including divergent or conflicting ones; we refer to this variation as interpretive ambiguity. Such ambiguity can undermine a visualization's communicative goal when the supported takeaways diverge from its intended message. Evaluating visualizations with target audiences can surface this variation, but such evaluations can be costly and difficult to scale. In this work, we explore persona-conditioned Vision Language Models (VLMs) as a complementary evaluation tool for interpretive ambiguity. For a given visualization, our pipeline uses topic-relevant personas to elicit candidate contextual takeaways supported by interpretations of perceived visual patterns, such as trends and clusters. We derive a Target--Pattern--Scope framework for comparing these interpretations, use LLM judges to assign Compatible, Disjoint, or Opposing ambiguity labels to statement pairs, and summarize their distribution in an ambiguity profile. To improve chart grounding, we benchmark candidate VLMs on chart comprehension and select the highest-performing model for each chart type. We evaluate statement-pair labeling against human-authored annotations and apply the approach to real-world, public-facing visualizations curated from established datasets. Through case studies, we examine how perceived visual patterns relate to contextual takeaways and how a visualization redesign changes the sampled interpretations. Our findings show how persona-conditioned VLMs can complement human feedback by generating candidate contextual takeaways and surfacing potential ambiguity for further inspection.
Visual analytics systems traditionally combine analytical methods and interactive visualizations into a fixed application specified at the design phase. Coding agents offer a flexible approach: experts could conduct analysis through natural-language instructions. We envision an agentic dashboard as a persistent visual analytics workspace where an agent makes decisions about analytical methods, views, and their coordination within a designer-defined decision space. This approach shifts part of the visualization designer's role from engineering a fixed interface to defining the decisions an agent may make at runtime. We illustrate how such a dashboard could operate with current capabilities and propose criteria for evaluating its behavior and potential failures. As analytical workflows become more fluid and agent-mediated, visualization can help experts maintain awareness, coordinate with agents, and exercise judgment.
Jaewook Lee, Judith Amores, Donghoon Shin, Bo-Eun Choi, Jason Kim, Davin Win Kyi, Seok-Young Kim, Jon E. Froehlich 等 9 位作者
原文摘要
What if a single face-facing camera could sense your surroundings---through your eyes? We introduce EyeQ, which explores corneal reflection as a modality for context-aware mobile interaction. Traditional vision-based context sensing relies on world-facing cameras that demand explicit aiming or dual-camera setups; EyeQ instead shows that an off-the-shelf VLM can read scene context from corneal reflections captured with a smartphone camera. To recover usable detail from mobile corneal crops, EyeQ pairs on-device iris-guided capture with server-side AI enhancement, then leverages VLMs to extract scene types, names, and objects from warped, unfamiliar eye-image sequences. In an N=10 study, we characterize how capture conditions shape reflection fidelity, evaluate what VLMs detect, misclassify, and hallucinate, and demonstrate interaction opportunities, including detecting a steering wheel to distinguish drivers from passengers. This work reframes the eye in HCI---from a target for gaze estimation to a passive mirror of one's surroundings---and examines the privacy tensions it raises.
Andrea Green, Jane L. E, Sneha Jain, Parker Ruth, Matthew Jörke, Sarah L. Billington, James Landay, Andrea Cuadra
原文摘要
While ambient assisted living (AAL) sensing systems offer promise for aging in place, their impacts on professional caregivers are overlooked. We conducted two collective speculation workshops (N=24) with personal health aides (PHAs, n=18) and certified nursing assistants (CNAs, n=6) to examine the ethical implications and value tensions of a care system with ambient intelligence (CSAI), our conceptualization of an AAL. Participants anticipated different implications across care roles and settings. PHAs anticipated additional responsibilities from CSAI, including learning how it operates and supporting older adults in its use. CNAs, who had greater access to workplace resources and protections, similarly noted that CSAI could create additional work. Participants also identified tensions around visibility, as sensing could support caregivers by documenting difficult working conditions while simultaneously making their behavior more vulnerable to misinterpretation. Through their modifications and surfaced discussions, we propose design guidelines for AAL systems, such as integrating context-aware sensing.
Governments are making decisions about AI before they can measure its public effects, and those decisions determine whether a person can learn that AI influenced an outcome, challenge it, and see it corrected. Research has documented what AI governance produces: principles, strategies, registers, and audits. However, we lack shared language for understanding AI policy while it is still being made. We introduce the Society-Centered AI Governance Framework, which gives policymakers, researchers, and the public a way to assess instruments in any jurisdiction by whether a person can get notice, challenge a decision, and see it corrected. It reads a jurisdiction's instruments together by date and binding status, separates an agency duty from a right a person can invoke, and traces which later instrument must enforce each commitment. We evaluated it in North Carolina using interviews with nine AI Leadership Council members, conducted before and after the state's 2026 AI roadmap, and a provision-level reading, with human-reviewed model-assisted coding, of 49 AI governance documents from 19 states. For the state's own AI systems, we find visibility-first governance: visibility duties are in effect, notice exists as an agency duty with no right a person can invoke, and AI-specific challenge and correction remain advisory. The roadmap assigns owners and dates to the missing protections, so the gaps are documented but not yet closed. Recent federal preemption proposals focus on AI development, so recourse for a state's own systems is likely to remain a state decision. We identify "legitimacy before protection" as a testable risk of visibility-first governance and name the milestones that would test it. More broadly, judging AI policy by who can act on what it makes visible lets the field assess recourse while policy is still forming, before harm occurs.
Yang Zhan, Yuxuan Huang, Yuchen Cai, Yichen Andy Yu, Xiaoran Yang, Ye Yuan, Qiao Jin
原文摘要
Mixed reality (MR) cross-device experiences often combine MR headsets with non-immersive devices such as laptops, tablets, and phones. Yet such cross-device design remains challenging because users must coordinate workflows across heterogeneous devices with different roles. This motivates our focus on usage patterns: recurrent ways in which people distribute and coordinate work across devices. Through a systematic analysis of 238 MR cross-device papers, we developed MRxD, a usage-pattern framework organized around three dimensions: user attention, task relatedness, and cognitive symmetry. Using these dimensions, we identified six usage patterns and synthesized the design strategies commonly used within each pattern. We further investigated the framework's application in design practice through a workshop (n=6) and the results suggest that MRxD can guide pattern-based design exploration and identify blind spots. We further discuss how usage patterns may shift or combine, and outline research gaps and design implications for future MR cross-device systems.
Nicole Sullivan, Ella Cao, Leilani Battle, Magdalena Balazinska
原文摘要
While LLMs are increasingly used in various contexts, we lack a fundamental understanding of how they impact SQL query debugging. In response, we conducted a user study with 16 participants of varying SQL expertise, examining how they debug complex SQL queries with and without LLM support. Initial benchmarking indicated that our selected models are already capable SQL debuggers. Qualitative analyses of our study data revealed that while participants trusted the LLM, they varied widely in how and whether they delegated debugging tasks to the LLMs, which we characterized through a new three-state model of participants' SQL reasoning processes. Additionally, participants enforced specific collaborative roles for the LLM such as TA or sounding board, which participants controlled by withholding context from the LLM. These findings suggest that LLM-based tools could enable preset role selection, broader context management, and flexible automation features, including low automation options, to support SQL debugging moving forward.
Computer-use agents learn to operate mobile interfaces from corpora of captured screens and accessibility trees. The same trees are what a screen reader consumes, so an agent trained on them inherits whatever labelling gaps the underlying applications carry. This paper measures one of those gaps, content-description coverage, in two Android corpora: Rico, collected in 2017, and AndroidControl, collected in 2024. Across 17,373,614 interface elements, 17.74% of the Rico elements that need a content description under our coding rule carry one, against 57.27% in AndroidControl. The difference of 39.5 percentage points holds under five alternative coding rules and when whole apps are resampled. It describes two datasets that also differ in app selection, capture method and platform version; it does not measure change over time. Input controls are the least labelled role in both corpora (2.20% and 27.21%, or 30.99% in AndroidControl if hint text counts as a name). Descriptions in the older corpus are also five times more likely to be placeholders (10.25% against 1.95%), so the difference is not an artefact of developers typing filler. Resolving 26,205 task actions to the elements they touched shows that, among tappable graphics with no text, demonstrations operate elements described 9.0 points more often than the others on the screens they visit, about 3.5 points of it from the choice of element within a screen. That is a selection effect inside the benchmark. A two-coder reliability check (kappa = 0.91 and 0.96) and the full pipeline accompany the paper.
Children who learn braille have few illustrated books to read with their fingers, and almost none about the things they ask for. Existing approaches either adapt a published picture book by hand, with a specialist in the loop, or generate a tactile image without the braille page around it. We present Tact, an open system that turns a one-sentence idea into a printable page of braille with a raised picture, in Italian or English, on a consumer 3D printer, with no account and no cost beyond the plastic. Its central idea is to treat the physical page as a constraint on generation: the cell budget of a fixed tactile layout decides how long a story the language model is asked to write, and the story is measured and requested again until it fills two cards. The page itself follows requirements traced to braille standards, the tactile-graphics literature, and formative feedback from a teacher of students with visual impairments and a blind reader who printed and read a page. In an evaluation of 60 stories generated live, every story fit exactly two cards and 57 filled the second card to at least 11 of its 13 rows; a final targeted round recovered 5 of 6 retries of the remaining cases. The picture chosen matched the story's subject on 105 of 105 illustrated pages, against 63 for first-match keyword selection. Uncontracted Italian and English braille, and contracted English produced by liblouis compiled to WebAssembly, matched native liblouis cell for cell on 106 texts, and the interface showed no WCAG 2.2 AA violations. Cards are 150 mm square on a 2 mm plate, printed standing up, and take about 57 g of PETG; test prints are under evaluation.
Jeffrey M. Girard, Antony D'Avirro, Jason Z. Zheng, Jacqueline R. Vertino, Benjamin Peloquin
原文摘要
Listeners can judge an agent's speech on how human it sounds (human-likeness) and on how much they would want to talk to it (conversational preference). Listening tests rarely ask about conversational preference, so the relation between the two judgments is unclear. We compared them in paired-choice tests on 120 turns that actors improvised as customer-service representatives. On each trial, two agents spoke the same turn. In Study 1, both agents came from six commercial text-to-speech systems, each with eight voices. In Study 2, one was the actor's original recording and the other a system. Some systems ranked higher on human-likeness, and others on conversational preference. Listeners chose systems over actors in 19% of human-likeness trials but 36% of conversational preference trials (and 49% for the most preferred system). On both judgments, variation among a system's voices was about three-quarters of that between systems. Of three automatic naturalness scores, two predicted lower human-likeness, and none predicted conversational preference. In this setting, listeners tended to favor faster speech and fewer pauses between clauses on both judgments, and more pitch variation for conversational preference. Thus, measuring only human-likeness or testing one voice per system can misjudge which agents listeners would rather talk to.
Amama Mahmood, Shiye Cao, Laura Fleig, Junan Jiang, Theodore Jensen, Yee-Yin Choong, Kristen Greene, Chien-Ming Huang
原文摘要
Ambient artificial intelligence (AI) scribes have moved rapidly from experimental pilots to enterprise-wide deployment across health systems, promising reduced documentation burden and improved clinician experience. While research evaluating these tools has expanded quickly, there are no agreed-upon protocols or standardized metrics for assessing their performance, making it difficult to compare findings or cumulatively synthesize evidence across studies. Across 54 recent evaluation studies, we identify ten evaluation dimensions and 40 distinct metrics, operationalized in 96 different ways-35 of them (36%) used by only a single study. Assessment is dominated by documentation efficiency and time, measured in 44 of 54 studies. The other dimensions appear at broadly similar rates (16-24 studies each), and safety and clinical outcome least consistently of all. Ripple effects on the broader care system, such as impact on patients' care outcomes, redistribution of work, changes in billing and coding intensity, and downstream use of notes by other clinicians are rarely assessed. Moreover, the metrics used rely heavily on self-report and are fragmented and inconsistently operationalized, which limits comparability and hinders synthesis across studies. As an effort to support more rigorous and cumulative evaluation, in this paper we review the study designs, data sources, and metrics commonly used to evaluate ambient AI scribes in healthcare settings, discuss the limitations of current evaluation practice, and argue for a more holistic approach grounded in a standardized, multi-dimensional set of metrics.
Megan Wei, Melanie Subbiah, Audrey Lee, Annya Dahmani, Dave Edwards, Helen Edwards, Ellie Pavlick
原文摘要
When does work done with AI still feel like ours? As AI becomes woven into everyday tasks, we must examine what happens to our sense of ownership and contribution when a machine shares in producing what we make. We report an exploratory qualitative survey in which participants were asked to describe two recent, self-selected tasks completed with AI: one that felt like their own and one that did not. We find that felt ownership depends on the process of collaboration: people disown work when they merely approve AI's suggestions, but retain ownership when they lead, iterate, or rewrite. Ownership can also extend to settings where people own the vision for a project but not the execution; respondents reported high ownership on tasks they could not have completed without AI. Loss of personal voice and a lack of comprehension of the output both erode ownership. Finally, willingness to disclose AI use is often decoupled from actual pride or ownership, and instead shaped by community norms and fear of credit erasure. We propose several research directions as a result of these findings to promote AI development that supports people's sense of authorship over their own lives.
Pranav Goyal, Andrew Stratton, Christoforos Mavrogiannis
原文摘要
Legibility in human-robot interaction (HRI) is often described as the property of robot motion that enables an observer to confidently and quickly infer the robot's intent. Mature frameworks exist for generating legible motion in front of static observers, typically by making the robot's destination apparent as early as possible. Social robot navigation in public spaces, however, presents a different challenge: pedestrians are themselves in motion, often with divided attention, and must coordinate with the robot in real time. In such encounters, where the robot is ultimately headed may matter less to a pedestrian than how it intends to pass them. This raises two questions: which aspect of intent should legible motion convey, and does legibility remain beneficial when pedestrians are not fully attentive to the robot? We address these questions in the ubiquitous and demanding scenario of hallway navigation, through two controlled user studies comparing multiple legibility formulations implemented within a shared model predictive control framework. Study 1 (N = 45) compares destination-based and passing-side intent representations, showing that the latter, particularly when adaptively updated, allows for smoother human motion and is perceived as more competent and less mentally and physically demanding than the former. Study 2 (N = 45) examines the effect of pedestrian attention, demonstrating that legible motion supports smooth human motion even under distraction, although this benefit is not consistently reflected in subjective ratings. Together, these findings suggest that effective legible motion in social robot navigation benefits from interaction-level intent representations that support coordination, with some effects persisting even when human attention is divided.
Eunkyu Park, Vasudha Varadarajan, Wesley Hanwen Deng, Anisha Reddy, Gunhee Kim, Motahhare Eslami, Jana Schaich Borg, Maarten Sap
原文摘要
When people seek advice from AI on social situations, the system may recommend a decision they fully agree with but justify it through reasoning they disagree with. Conversely, it may offer reasoning they agree with, and yet reach a decision they reject. We distinguish decision alignment, whether an advisor recommends the option favored by a user's measured values, from reason alignment, whether its stated reasons align with the user's values. In this work, we ask how these two forms of alignment shape people's preferences for AI advisors and their willingness to use them in different roles. We conducted a within-subjects study (N=121) in which participants evaluated three AI advisors varying in decision and reason alignment across eight everyday moral dilemmas. Participants then chose which advisors they would consult, allow to draft messages for review, or allow to act on their behalf in new tasks. When seeking advice, participants most often preferred an advisor that reasoned through their prioritized values while recommending the opposing decision (40%). However, when participants had to pick advisors for new tasks, their preferences shifted toward advisors that also recommended decisions aligned with their values, rather than those that only shared their reasons. Also, as participants considered tasks that gave AI advisors greater autonomy, they became more likely to opt out of using any advisor, from 12% when consulting to 38% when allowing an advisor to act on their behalf. Our findings suggest that decision and reason alignment shape advisor preferences differently across contexts of AI use, and that alignment with a user's values does not necessarily imply willingness to delegate.
Improving disability representation in AI-generated media requires community-specific datasets, yet openly sharing such data can expose communities to misuse, decontextualization, and exploitation. We examine how disability communities can participate in dataset governance through context-sensitive licensing. Working with three disability advocacy organizations, we conducted semi-structured interviews, an educational webinar and modular license activity to promote community participation in data stewardship. Our findings show that governance in these high-context communities relies on organizational memory (especially situated harms); both formal authority and informal relational practices; communication; and resulting trust. The modular license activity made organizational data choices more accessible and values-led, allowing communities to adapt Creative Commons clauses to express permitted uses, sector-specific restrictions, safeguarding requirements, and commitments to respectful representation. Yet, concerns remain about monitoring and enforcing appropriate data use. We contribute (1) an empirical account of governance tensions, (2) a practical modular approach, and (3) design implications for AI data platforms that support relational, context-sensitive governance beyond standardized, one-size-fits-all frameworks.
Atharva Naik, Shounok Kar, Varnika Sharma, Ashwin Rajadesingan, Koustuv Saha
原文摘要
Social and personal decisions in relational domains such as matchmaking are deeply entwined with cultural norms and historical hierarchies, and can potentially be shaped by algorithmic and AI-mediated assessments of compatibility, acceptance, and stability. In South Asian contexts, caste remains a central aspect of marital decision-making, yet little is known about how contemporary large language models (LLMs) reproduce or disrupt caste-based stratification in such settings. In this work, we conduct a controlled audit of caste bias in LLM-mediated matchmaking evaluations using real-world matrimonial profiles. We vary caste identity across Brahmin, Kshatriya, Vaishya, Shudra, and Dalit, and income across five buckets, and evaluate five LLM families (GPT, Gemini, Llama, Qwen, and BharatGPT). Models are prompted to assess profiles from the perspective of a caste-identified prospective bride or groom along dimensions of social acceptance, marital stability, and cultural compatibility. Our analysis reveals consistent hierarchical patterns across models: same-caste matches are rated most favorably, with average ratings up to about 30% higher (on a 10-point scale) than inter-caste matches, which are further ordered by distance along the traditional caste hierarchy, with higher-caste candidates generally favored at equal distance and an additional penalty for Dalit candidates. These findings highlight how existing caste hierarchies are reproduced in perspective-conditioned LLM evaluations and underscore the need for culturally grounded evaluation and intervention strategies in AI systems deployed in socially sensitive domains, where such systems risk reinforcing historical forms of exclusion.
Xuan-Quy Dao, Ngoc-Bich Le, Xuan-Dung Phan, Le-Minh Nguyen
原文摘要
Engineering courses spread their materials across slides (PPTX), typeset notes (LaTeX/PDF), and MATLAB, leaving students to connect them. We describe an instructor-in-the-loop workflow in which generative AI (GenAI) converts these assets into single-file HTML learning objects that combine presentation, mathematics, visualization, simulation, and assessment; a page is released only after its simulations pass numerical verification. Six open demonstrators implement 18 tasks from six engineering courses. In the eight tasks with quantitative cross-checks, every quantity met its tolerance at default settings (1\%, or 2\% for the finite-difference time-domain simulation) and agreed with a separate Python re-computation and a value-blinded MATLAB re-computation run by a colleague not involved in development; the checks also flagged under-resolved settings. In a perception survey, students (19 after post-hoc screening of 28 responses) reported that the pages linked equations with simulation, and 74\% preferred them to separate files, but usability was marginal (System Usability Scale 56.8). Twelve lecturers rated the approach as useful (item means 4.25--4.58) but were hesitant to adopt it; their median ratings on a 12-criterion capability rubric agreed moderately with the first author's ($κ_w=0.60$; individual median 0.31). The results show technical feasibility and preliminary positive perceptions; learning outcomes were not measured.
This submission indexes contributions to the CSCW 2026 workshop, "Women's Health as Cooperative Work: Participatory AI Across Care, Community, and Policy," held on October 11, 2026. The workshop examines women's health as cooperative work spanning patients, caregivers, clinicians, communities, institutions, and digital technologies. Contributions explore participatory AI, coordination breakdowns, the distribution of care labor, governance, decision-making authority, and accountability across care, community, and policy contexts. The index provides paper titles, authors, affiliations, thematic areas, and unique report numbers to support discovery of the corresponding individual arXiv submissions. The workshop proposal is separately published under DOI 10.1145/3785651.3818821.
Emma Ricci-De Lucca, Saanvi Chilakapati, Jillian Berch, Ashley Kim, Sahithi Medikondla, Iha Rastogi, Margaret R. Emerson, Elizabeth L. Murnane 等 9 位作者
原文摘要
Doulas are trained, non-medical professionals who provide emotional, physical, and informational support to pregnant individuals before, during, and after childbirth. Doulas provide extensive relational and coordination labor for pregnant and postpartum patients, spending far more time with clients than traditional providers do, yet they operate with almost no doula-specific digital infrastructure. This gap is especially acute in rural communities, where postpartum depression affects about 25% of women, well above the national average, and where geographic isolation and provider shortages leave many without access to behavioral health care. DoulaConnect began as a 12-week evidence-based postpartum texting curriculum, piloted separately in a randomized controlled trial in rural New England. Building on that work, we present DoulaConnect's development into an AI-enabled platform co-designed with doulas and rural mothers through participatory human-centered design, as a case for surfacing where coordination breaks down in doula practice and who currently absorbs the resulting burden. Our community-based participatory study, conducted across the platform's Discover and Define phases, combines qualitative sessions with doulas and mothers to identify these breakdowns. We close with tensions this approach surfaces and open questions for the workshop about how these findings will inform future participatory AI infrastructure for doula care in rural and other under-resourced communities.
Annalisa Degenhard, Annika Stampf, Michael Rietzler, Enrico Rukzio
原文摘要
Collision prevention systems (CPS) help VR users navigate their physical surroundings safely. Most approaches overlay conceptually distinct grids or add solid virtual counterparts of physical obstacles; both can reveal the physical space, as they either stand out visually or behave inconsistently during locomotion. We explore sight restriction as an alternative: a standard particle system visualizes the play area boundary as a fog-like cloud whose intensity scales with proximity and viewing direction. A pre-study (N=10) determined the distance at which the visualization must appear. In a main study (N=11), we compared this particle-based CPS with a grid-based solution, focusing on locomotion behavior: across 253 approach episodes, users interrupted their approach after a comparable distance once either visualization appeared. Although the particle boundary was visible three times as often, participants reported higher presence. We release our implementation as open source to support further work on immersive collision prevention.
Evan Pačes-Wiles, Bryen Alvarez, Sara Gulabani, Akshara Kolipaka, Snehalkumar 'Neil' S. Gaikwad
原文摘要
Communities in the United States and around the world are contesting data centers, the physical infrastructure of AI. These disputes pose environmental justice issues: residents must weigh unevenly distributed burdens using evidence scattered across metrics, corporate claims, policy documents, and public testimony. A formative study with 10 users showed that dashboards and single-agent AI can settle questions before people examine that evidence. Here we present EJAI, a source-grounded system that links an interactive map, retrieved evidence, missing-data statements, and two stakeholder personas that allow people to inspect evidence and form their own policy recommendations. In a small functional benchmark of 21 queries, retrieved evidence supported 98.4% of factual claims and the system acknowledged missing evidence in 32 of 33 cases. Persona alignment carried a grounding cost: the Corporate Executive omitted some evidence that conflicted with its assigned priorities. In a within-subjects study, 15 participants acting as residents prepared for a simulated town hall. Some checked cited AI claims while treating map metrics as official, some made conditional recommendations with safeguards such as water-use disclosure, and some found the personas simplified or biased. Our results suggest that civic AI can support preparation for public reasoning when it makes evidence, gaps, and disagreement open to inspection, and that corpus imbalance and two-sided personas remain open justice concerns.