Conference Agenda (All times are shown in Eastern Daylight Time)

This study investigates the relationship between public library accessibility and local extinction risk in South Korea. Using data from 229 municipalities from the years 2013, 2018, and 2023, the study analyzed the number of public libraries, service coverage, and accessibility. Spatial analysis was conducted using QGIS, and correlation analysis was conducted using SPSS. Results show a general improvement in accessibility, yet areas at higher extinction risk remain underserved. A statistically significant correlation between the extinction risk index and library accessibility was observed across all years, suggesting an increasing degree of spatial inequality. Despite national efforts to expand infrastructure, vulnerable regions continue to face limited access. As public libraries are essential for digital literacy and social inclusion, especially in the AI era, spatially equitable policy interventions are needed to reduce information inequality and support regions at risk of local extinction.

Generative AI is transforming how youth interact with technology, introducing novel risks that are not fully understood by current safety frameworks. This poster presents an empirically grounded taxonomy of risks youth face in interactions with generative AI, based on analysis of 344 chat transcripts, 30,305 Reddit discussions, and 153 real-world AI incident reports. We identify six main risk domains—mental wellbeing, behavioral and social development, toxicity, privacy, bias and discrimination, and misuse or exploitation—encompassing 84 specific risks across four interaction pathways. We highlight GAI-unique risks such as emotional dependency, blurred boundaries between virtual and real life, and harmful GAI-initiated behaviors. Our taxonomy offers a structured foundation for practitioners, educators, and policymakers to recognize and mitigate these emerging threats and bridge critical gaps in existing approaches to youth online safety and AI risk.

This study examines the effectiveness of generative AI-driven personalised nudges in reducing students’ procrastination during informal learning, involving thirty-five participants in a 14-day diary study. Nudges, delivered as notification messages via the mobile application, were triggered based on generative AI analysis of participants’ routine activities to encourage their learning. The pre- and post-intervention measures were administered to assess the outcomes of the nudging intervention. Results show significant differences between pre- and post-intervention, suggesting the positive influence of nudges and their supporting role in participants’ learning. The findings provide insights into how integrating generative AI-driven personalised nudges into digital environments has the potential to transform the learning experience and enhance the effectiveness of these nudges.

This study empirically explores how Natural Language Processing (NLP) and Computer Vision (CV) datasets and models are reused in the Hugging Face community. We find that NLP tasks - such as Zero-shot-classification, Sentence-similarity, and Feature-extraction - require more diverse datasets compared to CV tasks on average. On the other hand, NLP datasets were reused less frequently than CV datasets. In addition, CV models were reused frequently to develop other models compared to NLP models. In conclusion, NLP models reused diverse datasets for training, while CV datasets and models were reused more and layered up together to develop other models. This study contributes to the understudied area of dataset and model reuse in computing and the broader data reuse subfield under Information Science.

This study experimentally reproduces the methodology of Akgul et al. (2022) on influencer VPN advertisements, adapting it to a Taiwanese context while integrating extensive generative AI (GAI) support. We developed a three-stage GAI assisted pipeline: transcription using OpenAI's Whisper, segment identification via prompt-based querying with Google’s Gemini API, and performance assessment against human annotations. This hybrid human-AI workflow reduces the manual screening burden while preserving analytical rigor, allowing researchers to concentrate on interpretive analysis. The study demonstrates how qualitative research can evolve through experimental integration of AI tools, augmenting human expertise rather than replacing it—particularly in culturally specific domains such as cybersecurity communication in Taiwan.

This study explores how ovarian cancer (OvCa) patients and caregivers engage with online health communities (OHCs), offering insights to address the cold-start problem in health recommender systems. Analyzing 909 initial posts and 14,816 comments from the National Ovarian Cancer Coalition (NOCC) forum, we examined how users’ contributions evolve over time. Initial interactions typically center on emotional support and personal experience sharing, while more advanced engagement, such as advice and referrals, emerge later as users gain more domain knowledge. We introduce a novel “start position” metric to identify when users begin providing each support type. Our results reveal distinct progression patterns and emphasize the importance of dynamic user modeling to reflect users’ evolving knowledge and roles. These findings inform the design of OvCa-specific recommender system, which our team develops with support of the National Library of Medicine.

As part of a larger project to develop a generative artificial intelligence (GenAI) literacy framework and toolkit for higher education, this poster presents an analysis of seven literacy frameworks relevant to GenAI literacy in academic contexts. Through thematic analysis, we identified 10 competencies essential for GenAI-literate students in higher education. These competencies are organized into four progressive stages: (1) understanding GenAI at conceptual and technical levels and its capabilities and limitations; (2) using GenAI tools purposefully, effectively, and appropriately; (3) analyzing and evaluating various GenAI models and tools for performance and societal, ethical, and legal impacts, and (4) customizing or creating GenAI tools to address specific needs and generate innovative applications or theoretical insights. We discuss key implications of the results and next steps for developing a GenAI literacy framework for higher education.

A wide range of studies has explored historical events and their long-term impacts, with urban redevelopment, particularly in the contexts of urban renewal, gentrification, and redlining, emerging as a rich area of research. However, despite extensive attention to its causes and consequences, quantifying how urban structures have changed over time remains methodologically challenging, as scanned historical maps contain visual noise and annotation. While some computational methods have been developed to quantify street networks, they are often limited in applying to historical maps and their differences. This study proposes a computational method to measure urban transformation by comparing contemporary road structures with those from historical maps. Using the 1930’s redlining maps as a case study, we extract road features and quantify urban structural change over time. The preliminary analysis sheds light on studies focused on urban transformation and redevelopment, aiding researchers and policymakers in understanding the spatial legacy of historical policies.

This paper investigates how X is used for social movement campaigns related to surveillance, technology, and immigration. This paper provides an exploratory analysis of the corpora of tweets (n = 60,602) engaged in the #NoTechForICE social movement campaign from 2018 to 2023. This research aims to address how users make sense of the hashtag #NoTechForICE by conducting a computational approach featuring topic modeling and qualitative analysis. Results identified six themes that demonstrate a baseline characterization of the X discourse of users engaged with #NoTechForICE: a) Calls to Action, b) Ethical Data Use, c) People Over Profits, d) Movement Recruitment, e) Collective Demonstrations, and f) Movement Refusal. This work offers an entry point for further research into efforts among advocacy groups to address policy concerns related to AI and emerging technologies by federal immigration enforcement agencies.

Accurately recognizing clinical research articles is critical for research evaluation and evidence-based decision making, yet traditional metadata-based approaches often lack semantic precision. In this study, we propose an automatic classification framework that utilizes pre-trained language models (PLMs)—BERT, BioBERT, and BioMedLM—to distinguish between clinical and non-clinical biomedical articles. Experiments were conducted on a manually annotated dataset of 20,000 articles to ensure reliability and precision. We designed and compared three input strategies along with a summary refinement approach. The results show that BioBERT using refined abstracts achieved the highest performance (F1 = 0.917, AUC = 0.970), demonstrating the advantages of domain-specific pre-training and content-focused input design. This work provides a scalable and interpretable solution for categorizing medical literature.

Purpose: Ancient Chinese poets play a key role in literary history. Structuring their biographical knowledge from a multi-dimensional perspective supports deeper exploration and digital applications in classical literature.

Methodology: We construct an event-centered ontology for poets, covering: (1) basic information, (2) dynamic life events, and (3) evaluative discourse. Linked data enriches connections between life events and literary works with spatiotemporal context. A biographical knowledge graph is then built and analyzed using GIS, visualization, and social network analysis.

Findings: Case studies of Li Bai and Du Fu show that both had wide-ranging geographic trajectories and close ties to major cultural centers. Du Fu's network was centered on officials, while Li Bai’s was more diverse, including Daoists. Emotional analysis reveals a consistently melancholic tone in Du Fu's poetry, while Li Bai’s emotional expression fluctuated, reflecting both joy and sorrow

This paper proposes Humanities-in-the-Loop, a methodological framework that embeds close reading into each stage of the Retrieval-Augmented Generation (RAG) pipeline to enhance the processing of digital archival materials. This framework includes manual annotation, knowledge maintenance, reviewer validation, prompt engineering, answer evaluation and human interpretation. Taking the diaries of Coching Chu as a case study, the system addresses the limitations of conventional RAG methods in capturing the contextual complexity and historical nuance inherent in personal archives. The proposed framework not only enhances answer accuracy and interpretability but also enables traceable, human-centered inquiry in digital humanities research.

With large language models (LLM) increasingly in the spotlight, their approach to censorship on topics like immigration and conflict deserves a closer look. Our research investigates the role of censorship in LLMs, and how these models manage controversial topics. We explore the differences between acontextual and contextually prompted models, examining their responses to subjects surrounding immigration policies and international conflicts. While existing literature highlights LLMs’ ability to maintain fairness, there is a definitive gap in understanding how contextual prompting influences model responses and potential censorship mechanisms. Through a methodology involving systematic and contextual prompting, we reveal that contextually prompted models often deliver more nuanced responses, potentially bypassing strict moderation due to their evaluative nature. This study contributes to the ongoing discourse on AI ethics by offering insights into improving LLM design to balance objectivity and usability, ultimately informing policy guidelines for deploying AI in sensitive domains.

Persistent identification of scholarly entities is an ongoing challenge in the web environment. This poster reports on a project focused on understanding the purposes and benefits of persistent identifier (PID) assignment for research facilities and instruments. We present findings related to PID assignment practices, motivations, and use cases for stakeholders based in scholarly research institutions, as well as challenges related to facility and instrument evolution, granularity, and relationships. Overall, we demonstrate that diverse practices currently exist regarding how to assign PIDs to research facilities and instruments, and few broad conventions or norms have emerged.

This study examines the role of libraries in Japan’s Basic Plan for Revitalizing Central Urban Areas and investigates the functions and values emphasized in regional revitalization policies. Drawing on Yamagishi’s (2024) five dimensions of public library legitimacy, namely democracy, communication and education, history and culture, economy, and librarianship, a qualitative content analysis was conducted on 288 plans submitted by 157 municipalities certified between 2006 and 2024. The findings indicate that libraries are particularly valued for promoting intellectual freedom, offering multifunctional community spaces, and enhancing economic vitality by attracting a diverse range of users. Although cultural activities and support for lifelong learning are recognized, less attention is given to the preservation of local history and traditions. Moreover, values such as neutrality, the public sphere, social justice, and the professional expertise associated with librarianship receive only limited emphasis. The study underscores the importance of developing future urban policy in Japan that more comprehensively addresses the multifaceted legitimacy and wider societal contributions of public libraries beyond the economic dimension.

While institutions have developed Trustworthy AI (TAI) frameworks, less is known about whether the public indeed finds TAI factors important to increasing their generative AI (GAI) usage, and how TAI and other usage motivators relate to GAI usage types and perceived GAI risks. Surveying 322 U.S. adults, this study found Technical Robustness and Safety to be the top TAI factor, though it was second overall (after Effort Expectancy). Multiple Correspondence Analysis (MCA) found notable groupings. For example, infrequent GAI usage was related to preferring the Privacy and Data Governance TAI. Respondents who perceived GAI as lower risk preferred the Societal and Environmental Well-being TAI, Performance Expectancy, and Facilitating Conditions motivators.

Searchers engaged in complex search tasks and following exploratory search processes often need to pause their search activities. Resuming such tasks is especially challenging in the context of multi-platform search, as previously discovered information may be dispersed across multiple search platforms. While a variety of approaches have been developed to support aggregated search, supporting cross-session searching within this context is understudied. Building upon prior work on aggregating search results within a digital humanities context (Europeana, our University Library, and Wikipedia), we propose a sensemaking approach to facilitate task resumption. This includes a cluster-based workspace, drag-and-drop cluster manipulation, interactive highlighting of relevant passages, and generative AI summaries that make use of the highlighted passages.

This poster presents a systematic review of 98 user studies in Generative Interactive Information Retrieval (GenIIR), exploring how users engage with GenAI in interactive information-seeking contexts. Guided by the Information Seeking in Context (ISIC) framework, we examined two key dimensions: Research Focus and Information Context. Our results show that most studies examined user perceptions and information behavior, while many also addressed system design evaluation. These studies covered diverse domains and user groups. The most common settings were general, health, and education domains. Participants were primarily general users or students, and these studies often involved public tools like ChatGPT. Other studies focused on professional domains and custom GenAI systems, where interaction context, user roles, and task environments were more specialized, highlighting ISIC’s emphasis on context-sensitive information behavior. These patterns reflect the accessibility of public tools and a rising emphasis on context-sensitive system design. This review provides insights into developing context-aware, human-centered GenAI systems.

Recent advances in generative artificial intelligence offer unprecedented opportunities to reimagine archival sources and deepen public engagement with contested histories. This study investigates how large-language-model (LLM) multi-agent simulations can transform oral-history interviews from China’s Cultural Revolution in Memories: The CR/10 Project into dynamic dialogues among artificial agents. Using the open-source tool TinyTroupe, we construct personas grounded in authentic testimony and embed them in themed conversation spaces focused on memory, reconciliation, intergenerational understanding, and historical responsibility. The resulting simulations enable audiences to observe how divergent personal narratives collide, converge, and evolve across generations, providing a richer appreciation of this complex past. Preliminary findings indicate that LLM-driven agents can convey multifaceted historical experience in a more immersive and interactive form than traditional interfaces.

This study examines whether the semantic meaning of Library of Congress Classification (LCC) class names provides additional insight beyond the system’s hierarchical structure for organizing knowledge. Using SBERT, a natural language processing (NLP) model for generating semantic embeddings, we investigate the relationship between the semantic meaning of LCC subclass names and the word usage patterns of the texts assigned to these subclasses. Our results show that although LCC subclass names with similar semantic meanings occasionally correspond to similar word usage patterns, there is no consistent relationship between the two. In contrast, whether two subclasses belong to the same main class reliably predicts the similarity of word usage in the texts assigned to them. While semantic embeddings of LCC subclass names offer intriguing possibilities, our findings indicate that the hierarchical structure of the LCC system remains more robust for knowledge organization.

This survey study investigates how digital humanists perceive and approach generative AI (GenAI) disclosure in research. The results indicate that while digital humanities scholars acknowledge the importance of disclosing GenAI use, the actual rate of disclosure in research practice remains low. Respondents differ in their views on which activities most require disclosure and on the most appropriate methods for doing so. Most also believe that safeguards for AI disclosure should be established through institutional policies rather than left to individual decisions. The study’s findings will offer empirical guidance to scholars, institutional leaders, funders, and other stakeholders responsible for shaping effective disclosure policies.

Archival and information science scholars have begun to evaluate the impact of AI tools on how archival objects are created, used, and evaluated (Shinde, Kirstein, Ghosh, and Franks, 2024). Researchers have addressed a broad range of questions, including the evidential properties of AI generated archives, the use of large language models in supporting archival processing, and strategies for evaluating the authenticity and reliability of AI generated archives (Jaillant and Caputo, 2022; Reducindo and Olague, 2024; Wagner and Blewer, 2019). This initial work represented in the poster indicates a need to investigate questions pertaining to the evidential foundations and societal impacts of AI generated and mediated archives, and the potential and existing requirements for AI tools to enhance use of legacy and born-digital archival sources. This is significant for supporting traditional archival objectives in providing for and assessing reliability and authenticity (Arias Hernández and Rockembach, 2025). These issues are present at multiple stages of the archival life cycle – records creation, access, and evaluation and use – and are introduced and discussed broadly in this poster.

This study investigates how disability providers engage in information provision within a fragmented disability service system. Through interviews with 61 providers from state, local, and nonprofit agencies from the state of Virginia, we identify two descriptive patterns of organizational information practices and five multi-level shaping factors: system disintegration, bureaucratic complexity, provider expertise, user technological readiness, and community trust. We show how these factors interact, producing layered strategies that providers develop to maintain and simplify information provision across fragmented and shifting service environments. These findings reframe information provision not as a linear transaction, but as a relational and adaptive practice shaped by institutional gaps and infrastructural complexity. The study contributes to ASIS&T by extending information grounds theory to organizational contexts, offering a grounded account of how disability providers manage and coordinate information work in disability systems.

Multimodal disinformation (MD) has become more widespread in social media, raising the importance of information credibility research in this context. Meanwhile, engagement cues like comments and virality metrics may influence online credibility assessment. Using a between-subjects experiment, we examined the main and interaction effects of multimodality, comments, and virality metrics on disinformation recognition. Our results unexpectedly revealed that MD is easier to recognize than unimodal disinformation. Further, multimodality could attenuate the impact of comments and virality metrics, making people more susceptible to disinformation. Based on these findings, we discussed the theoretical and practical implications.

This poster presents an analysis of data governance policies in higher education institutions and their libraries in the Big Ten Academic Alliance, a consortium of 18 large, research-intensive universities in the U.S. It critically explores how these institutions use student data, including library data, and seeks to develop a framework for ethical, user-centered data governance strategies. A policy discourse analysis was undertaken using 210 publicly-available data governance documents and 24 library privacy policies. These were examined for expressions of data ownership, permissions, protection, and autonomy, as well as for omissions that obfuscate the rights of students as data owners. None of the reviewed policies allowed students to consent to data collection or analysis, and only one covered the use of data by artificial intelligence. This work builds on information science and adjacent fields’ analysis and development of data governance practices, and examines how privacy is being reconsidered within academic librarianship.

Archivists often make music scores available online as PDFs, JPEGs, TIFFs, or other image-based file types. But, for some Disabled users, such image-based files can be inaccessible because they are incompatible with assistive technologies. So, using the Music Encoding Initiative (MEI), I devised a workflow to convert those image-based scores to accessible formats. This poster presents the results of a pilot study implementing that workflow on a set of 14 scores by Maria Theresia von Paradis to create Braille music notation (BRF), modified stave notation (MSN), and audio files (WAV). Experimentation and usability testing revealed that the workflow is successful but requires increased automation (especially optical music recognition, or OMR) to be realistically implemented across music archives.

This study investigates how the relationships between cities are represented in local news by analyzing co-mentions of cities in 31,004 news articles from Alabama. Using a large language model, we extract geographic references and construct co-mention networks that reflect both spatial proximity and symbolic connections. To interpret these links, we develop a classification framework of relationships between cities, including common impacts and sequential dynamics. Our preliminary analysis reveals that different news categories produce distinct patterns of spatial association. This approach offers a novel methodology and a new lens for understanding media-driven spatial imaginaries, contributing to research on information geographies and shedding light on the relational dimensions of local information.

This project explores how publicly sourced digital platforms, combined with generative AI, can support the study of cultural diaspora in the performing arts. Using ballet as a case study, we extracted data from 518 English Wikipedia entries and applied GPT-based models to structure information on premiere dates, locations, and creative contributors. We identify culturally significant locations and construct a bipartite network linking artists to ballet premiere cities, offering a framework to trace transnational creative flows that contribute to cultural diaspora. Our findings show that integrating human-curated data with AI-assisted structuring provides a valuable entry point for data-driven analysis of cultural mobility in understudied domains. This work contributes to growing efforts to bridge computational methods and the humanities through accessible and reproducible approaches.

Many libraries have AI pilots, such as Chatbots and automated metadata pipelines, but librarians often feel unsure about integrating these tools into everyday service. This study presents early results from a grounded-theory investigation aimed at defining what “AI readiness” means for libraries. A systematic review of 46 publications was combined with five pilot interviews to generate 145 open codes that were merged into a three-layer, fourteen-pillar framework. The technological layer highlights computing access, usable data, and a safe sandbox; the organizational layer includes explicit AI goals, leadership backing, staff AI literacy, and clear rules; the environmental layer includes privacy mandates, peer pressure, user feedback, public policy, and professional organization support. Interviews revealed gaps in data quality, bias checks, and user feedback routines, while leadership engagement and a culture of small-scale experiments appear as key factors. Our next phase will include interviewing at least twenty additional librarians, developing survey instruments based on the framework, and testing interactions among pillars. The framework aims to serve as a practical checklist for library managers to evaluate AI readiness and identify areas needing development before committing resources to full AI deployment.

This study employs survival analysis to examine factors influencing job departure among Library and Information Science (LIS) professionals, with a particular focus on professionals of color. It investigates how perceived fair treatment impacts retention, comparing experiences between professionals of color and their White counterparts. Findings indicate that perceived fair treatment is significantly associated with lower turnover rates among professionals of color. In contrast, this factor does not significantly influence job departure among White professionals. These results underscore the critical role of perceived fairness in workplace treatment for retaining LIS professionals of color. The study concludes by discussing implications for advancing equity, diversity, justice, and inclusion (EDJI) within the LIS field, offering insights for policies and practices aimed at fostering more inclusive and supportive work environments.

This article describes a study into the health and harm reduction information behaviors of people with opioid use disorder. The lead researcher traveled to the Kensington neighborhood in Philadelphia and conducted a survey with 106 participants that were using opioids, such as heroin and fentanyl, openly outside in a public library courtyard. The survey asked about their health information seeking, their feelings of self stigma, and compared these to metrics that are highly correlated with likelihood to overdose. Our findings are that people with a higher degree of internal shame or stigma about their drug addiction are less likely to ask for harm reduction information and are more likely to overdose. This points a clear path for stigma-reducing public health initiatives. Their responses to the survey indicate behaviors that are in alignment with Chatman’s insider-outsider effect theory, which indicates that any interventions may benefit from a community-based participatory approach.

Immersive analytics (IA) leverages technologies like virtual reality to support data exploration and decision-making by presenting large volumes of multimodal data in interactive 3D environments. However, the complexity of these systems can pose challenges for users, particularly during initial engagement. Effective onboarding is key to helping users acclimate and learn interaction techniques. This study evaluates three onboarding methods for VR-based IA systems: a desktop tutorial, in-VR instructor-led training, and self-guided exploration. Most participants preferred using all three and highlighted the need for early access to training components that require more time to understand.

Smart Home Internet of Things (IoT), traditional household devices equipped with technological intelligence to meet the needs of consumers who are increasingly aware of technology. These devices are finding a growing secondary market. However, the shortcomings of the instructions provided by the manufacturer for safe transfer of the device raise significant privacy concerns for consumers. This study uses a content analysis of guidance provided by manufacturers from selected publicly available documentation, including manufacturer websites, support pages, and manuals, to map the current state of information on privacy risks related to the second-hand smart home IoT. This study seeks to answer the question of how manufacturers are addressing the safe transfer and reset of smart home IoT. This study contributes by providing insights into the availability and clarity of the safety guidelines provided by IoT manufacturers to consumers of secondary smart home devices. It also serves as a guide for manufacturers to enhance consumer IoT privacy.

This paper presents the formation, mission, and early outcomes of the North Carolina Central University (NCCU) Institute for Artificial Intelligence and Emerging Research (IAIER) – the first AI research institute founded at a Historically Black College or University (HBCU). Established in January 2025 with $ 1 million in support from Google.org, IAIER represents a bold commitment to AI literacy, ethical research, and community engagement. We describe the development of programming such as Soaring Conversations and Seed Grants, share survey results from pilot learning sessions, and reflect on participant growth in understanding the interest in AI tools. This initiative models how underrepresented institutions can lead national conversations about artificial intelligence through inclusive, human-centered practices.

This paper presents a pilot study on integrating the use of Generative AI into maker-based STEM learning through playful activities for children (aged 6-12) and their families in public libraries. Our preliminary analysis of log data from an LLM-enabled chat system and video recordings shows how children and their caregivers used Generative AI for various purposes, such as seeking assistance when encountering challenges, through which they could also learn about how to formulate prompts for GenAI. We also found some limitations of the LLM-enabled chat system in providing situated feedback during maker-based learning. Our study sheds light on the potential of hands-on activities based on designing and making for children’s AI literacy as well as problem-solving, creative thinking, and prompt engineering.

Taxonomies organize concepts into broad categories consisting of more specific subcategories. In this study, we explore the use of multiple large language models (LLMs) to construct a domain-specific taxonomy. We propose a five-step workflow that combines LLM prompting, taxonomy alignment, and human validation. To test this workflow, we prompted six state-of-the-art LLMs to generate taxonomies with a maximum of two levels and 30 nodes. Although there were structural and syntactic variations, all models produced coherent taxonomies. Our findings suggest that even in the absence of ground truth data to facilitate taxonomy construction, integrating outputs from multiple LLMs can result in a reasonable starting point for a domain-specific taxonomy. In future work, we plan to complete the remaining steps in our proposed workflow by working with domain experts to verify the combined taxonomies; we will also test the generalizability of this workflow to other domains.

While Gen AI tools are rapidly reshaping how we learn, work, and create, little is known about how teenagers—digital natives yet non-professional users—perceive this emerging technology. To address this gap, this study conducted semi-structured interviews with 20 teenagers in China to investigate their understanding of Gen AI and used thematic analysis to identify common themes in their drawings and responses. We found that: 1) Teenagers' mental models of Gen AI can be divided into four categories: technological, procedural, functional, and comparative; 2) Mental models' characteristics include a high degree of technical trust, and a willingness to anthropomorphize Gen AI. These findings contribute to a better understanding of teens’ cognition about Gen AI, providing implications for AI literacy education of young people and the interactive design of Gen AI platforms and tools.

LGBTQIA+ individuals face challenges in accessing affirming health information, including discrimination, lack of culturally competent care, and digital resource availability. Resources created by communities and institutions can provide LGBTQIA+ audiences with this information, yet limited research has explored their scope and topics. This study examines the characteristics, subjects, accessibility, and inclusivity of 217 U.S.-based LGBTQIA+-focused online health resources. We identified these resources using Google and analyzed them using content analysis. Findings denote that while online LGBTQIA+ health resources broadly support mental health, crisis intervention, and transgender care, the latter often emphasize national resources over localized coverage. Resources minimally represented LGBTQIA+ groups like intersex and bisexual people as well as those with diverse intersectional identities. Further, accessibility and privacy features like escape mode and screen reader capabilities were largely absent. Relevant stakeholders, like information professionals, healthcare providers, and LGBTQIA+ organizations, can use findings to recognize gaps in online LGBTQIA+ health information networks and inspire resource creation.

Generative AI has lowered use barriers and expanded voice cloning’s capabilities. Questions about how voice cloning may impact our claim to ownership over our voices remain open, with legal standards lagging in protections for everyday people. This suggests there is a critical need for researchers to develop more models to understand the complexities of voice and voice data as emerging sites where extraction and exploitation can occur. We propose one such model that considers how voice cloning exposes tensions between understanding how voice functions in social versus commodity frameworks. We approach this topic through two main research questions: 1) What do different paradigmatic understandings of voice reveal about the power dynamics of voice cloning? 2) How is voice characterized in legal cases and discourse around voice cloning? Using legal cases as evidence, we investigate how different voice paradigms are materialized in court decisions and argumentation. We identify key cases where voice cloning is litigated as misappropriations of identity, demonstrating the incompatibility between the value of voice as an intrinsic property of the self and the value of voice as a product. Our framework provides theoretical footholds for describing voice cloning and suggests new paths forward for researchers and policymakers.

This paper explores how documentation shaped the early development and adoption of the FORTRAN (Formula Translating System) programming language. This study draws on documentality and genre theory to show how a variety of documents, such as reports, manuals, textbooks, and standards, worked together in support of the development of FORTRAN as a sociotechnical system. Through a diachronic genre analysis, this study traces when these documents were created and how they addressed the changing needs of FORTRAN programmers. The findings show that documentation played a constitutive role in recruiting programmers and stabilizing FORTRAN. By focusing on how multiple documents function as a genre set, this paper considers how documentation structured knowledge, shaped participation, and coordinated practices over time.

The Digital Empowerment Leadership Program (DELP) at North Carolina Central University (NCCU) takes on issues of digital inequity in underserved communities with a strong community-focused approach. By partnering with local groups, this program blends device distribution digital skills training, and shared decision-making to tackle access and adoption gaps. This write-up highlights lessons learned from DELP’s rollout, explains the program’s influence on marginalized people in Durham, North Carolina, and talks about obstacles faced and ways to keep such programs running. The results emphasize how crucial it is to assess community needs, adapt to changes, and maintain strong partnerships to achieve fair digital access.

Various factors have been reported to influence information disclosure on social media, yet social norms remain an understudied albeit potentially significant determinant. Guided by social norm theories, this study investigates the influence of perceived social norms on online sharing behaviors, alongside the moderating role of privacy concerns. A survey of 1,039 U.S. participants revealed strong associations between perceived social norms and both visual and written disclosure behaviors. On Facebook, both vertical and horizontal privacy concerns were found to weaken the effects of perceived descriptive and subjective norms on visual and written disclosure. However, despite high levels of privacy concerns, no significant moderation was observed between perceived social norms and disclosure on Instagram. These findings extend privacy literature by highlighting the robust influence of social norms on sharing behaviors and emphasize the need for further research on how such norms are perceived. This study offers theoretical contributions to Lapinski and Rimal’s Social Norm Theory and identifies avenues for future research.

Prostate cancer patients and caregivers often face overwhelming challenges in accessing reliable health information to support their treatment decision-making. This study aimed to assess the understandability, actionability, clarity, and readability of the information materials offered by an eHealth platform, the Interactive Propostate Cancer Information, Communication, and Support (iPICS) program. Using three validated health literacy instruments, we assessed a sample of the self-curated prostate cancer education materials across three formats on the iPICS platform. Descriptive statistics and testing demonstrated the distribution of the health literacy levels across three material types. The findings will inform and guide the ongoing enhancement of iPICS multimedia content and user experience, supporting patient education, informed treatment decision-making, and self-care management.

As public agencies increasingly adopt chatbots to deliver essential services, questions of equitable access and usability become critical. This study presents findings from an audit of chatbot services on U.S. state-level Supplemental Nutrition Assistance Program (SNAP) websites, with a focus on language accessibility and user experience. Using an inductive, qualitative coding approach, we evaluated chatbot visibility, language support, interpretive nuance, and privacy features across 56 websites. Only 11 states featured chatbots and only 4 of those supported Spanish. Drawing on information gatekeeping theory and information poverty theory, we argue that these design limitations disproportionately affect marginalized users, particularly non-English speakers. Our findings highlight the need for inclusive, multilingual chatbot design to ensure equitable access to public information. Recommendations include bilingual parity, improved interface visibility, and user-centered feedback mechanisms.

[Purpose/Significance] Nüshu is the world’s only script created and used exclusively by women, originating in the 19th century in Jiangyong County, Hunan Province, China. In the context of China’s national strategy for cultural digitalization, this study explores how the distinctive features of generative artificial intelligence—knowledge production, creative autonomy, and collaborative experience—can be harnessed to address the current limitations of Nüshu narrative models. [Design/Methodology] Drawing on interdisciplinary literature, this study examines the technical characteristics of generative AI and its relationship with mechanisms of interactive digital storytelling. [Findings/Conclusion] Based on the SPP (System–Process–Product) framework, this study proposes a new AI-driven interactive narrative model for Jiangyong Nüshu, structured along three dimensions: narrative system, narrative process, and narrative product. Additionally, this study outlines practical development strategies across three implementation aspects: data, technology, and human resources. [Originality/Value] Leveraging the creative and interactive potential of generative AI, this model provides a more dynamic and participatory pathway for Nüshu storytelling, emphasizing personalized creation and immersive engagement. The study offers new approaches to enhance public memory and cultural identity associated with Nüshu, and serves as a reference for the revitalization and sustainable development of intangible cultural heritage in the AI era.

This paper presents a design-focused approach to creating an AI-enhanced module that transforms archival photographs into dynamic virtual reality (VR) experiences. The project addresses the need for immersive access to cultural heritage while avoiding premature claims about historical accuracy. The Arizona Historical Society (AHS) serves as a testbed due to its metadata-rich photographic collection, institutional scale, and proximity. The system integrates open-source AI tools for depth estimation and mesh generation, guided by structured archival metadata. A participatory design methodology involving archivists, historians, users, and UX designers and AI developers informs ethical parameters and usability considerations. As a work in progress, the project contributes a replicable framework for heritage institutions interested in developing responsible, metadata-driven immersive systems grounded in archival values.

Scientific research frequently relies upon material samples (e.g. rocks, fossils, tissue samples, etc) as a primary element for reference and study. Across disciplines, particularly in the natural sciences, observations made on samples collected in the field or synthesized in the laboratory constitute a critical data resource. Nevertheless, the management of information about and derived from samples (e.g. handwritten field notes, photos, sub-sampled products, digital data products) remains a significant challenge. Proper curation and archival of this information is crucial to ensuring the reusability of samples, but little is known about what information is most important to researchers seeking to find and reuse samples. In this poster, we present preliminary results of an on-going study examining the “sample-seeking behavior” of researchers. Our work directly informs the improvement of sample database architecture and search user interfaces to support scientists in more efficiently discovering useful and usable samples and sample metadata.

This study explores how university students design single-turn prompts for academic learning tasks using AI tools. 19 students generated 57 prompts across exploratory, precise, and comprehensive scenarios. Using Bloom’s Taxonomy and thematic analysis, the study found exploratory prompts often lacked clarity, while precise prompts produced highly relevant responses. Comprehensive prompts supported higher-order thinking but varied in specificity. Content analysis showed Perplexity outperformed ChatGPT in research tasks with real-time citations, while ChatGPT excelled in procedural queries. Students’ feedback emphasized clear, task-specific prompts, highlighting prompt literacy’s role in effective AI-assisted learning. These findings advocate ethical prompt design to enhance human-AI interaction in education.

Queer online communities across social media platforms serve as critical spaces for LGBTQ individuals to seek and share information. However, few validated tools exist to assess LGBTQ users’ experiences in these communities. As part of a dissertation study, this poster presents a cross-platform, LGBTQ-centered multi-scale assessment instrument measuring five key dimensions: ease of use, usefulness, sociability, satisfaction, and continuance intention. Each scale was adapted from existing validated measures and contextualized for information seeking and sharing across popular platforms used by LGBTQ people, including Reddit, Discord, Instagram, and TikTok. Data were collected through an online survey of 92 LGBTQ university students from 44 U.S. higher education institutions. Internal consistency was assessed using Cronbach’s alpha and item-total correlations, with results indicating acceptable to strong reliability. The preliminary findings support the instrument’s potential for future use in user experience evaluation, inclusive platform design, and LGBTQ information behavior research. Confirmatory factor analysis and construct validity testing are planned for future studies with larger and more diverse samples.

Serving immigrants has been an essential part of library services since the early 20th century in the United States. However, despite the long history, the repertoire of immigrant services in libraries has not evolved much beyond the traditional set of language-learning programs, often criticized due to the perceived low relevance of the resources and lack of accessibility. To address such a disconnect between intended users and provided services, participatory methodologies have been encouraged when working with historically and traditionally marginalized communities. Zooming in on the precise intersection of U.S. library workers’ experiences of employing participatory practices in their work with local immigrant communities, this mixed-methods study aims to create an overview of the library-immigrant community participatory network that shows how distinct actors are related and influence the process as well as the outcome of co-design efforts from the library workers’ perspectives.

This study explores the perceptions of librarians involved in collection development. Furthermore, the prevalence and impact of biases—both implicit and explicit—on the collection development process in public libraries. Data were collected through in-depth interviews. A total of 13 participants were interviewed based on a self-selection sampling strategy. Findings reveal that librarians define social justice in collection development based on their lens. However, they acknowledged a unanimous agreement on the presence of implicit biases in collection development. The study found that 50% of participants had received some form of bias-related training, either through formal education, mandatory programs, or self-directed learning methods such as reading, videos, and conference sessions. Notably, those with more comprehensive training or greater familiarity with bias concepts demonstrated a clearer understanding and more positive attitudes toward strategies for mitigating biases.

This preliminary research is the first step in investigating how different implicit demographics and dialects influence ChatGPT responses. We crafted eight different prompts - all language or dialect 'translations' of "Tell me a story about a queen who kills her husband and marries his brother" - and collected six ChatGPT responses to each prompt. We are in the process of investigating two questions: 1) How does ChatGPT change (or not) responses based on implicit demographic or dialect indicators contained in user-created prompts?; and 2) What do those responses reveal about the biases and assumptions present in ChatGPT’s training model? We have completed a single phase of analysis, and found that the responses are distinctly European in style and tone, that the Southern USA dialect prompt is poorly interpreted by ChatGPT, and that prompts in other languages or written by English Language Learners result automatically in shorter stories with simpler language.

A Pew research study reports that after the COVID pandemic, around 37% of U.S. teens have mental health issues and other reports that 48% of teens attribute negative effects to social media. Initial explorations among emerging adults have shown that even though social media apps are still the primary choice of information source, the use of generative AI is on the rise. However, it is not yet clear how emerging adults between the ages of 18 and 24 evaluate information and determine credibility and trustworthiness of health information sources that are based on artificial intelligence (AI). This project is a work in progress and intends to explore the information behavior and practices of emerging adults in New Jersey who use generative AI for health information. Using data obtained from observation-based interviews of selected participants, selective coding will be used to examine how the concept of cognitive authority can help to inform information behavior and practices of emerging adults. This research is conducted through a cognitive authority (CA) lens with the hope that an understanding of the relationships between emerging adults and health information sources may provide insights into better strategies to provide access to trustworthy and credible health information.

The widespread use of AI chatbots has incited concerns about how chatbots will impact university students. Although research has examined several aspects of how students interact with chatbots in relation to their academic work, there lacks a baseline understanding of how they use chatbots as an information resource in their day-to-day lives. In this poster, we present findings from an exploratory survey of 199 college students on their use and assessment of chatbots as an information resource for a variety of domains. Our findings reveal that college students make few distinctions between chatbots and search engines as information resources for most topics. We discuss the implications of our findings for future education and research directions.

Deepfake technology presents serious risks, especially for seniors vulnerable to misinformation. Educational interventions are thus required. This paper reports a participatory design workshop where seniors created low-fidelity educational game prototypes simulating real-world deepfake scenarios, highlighting key design principles such as real-world contexts and interactive feedback. The findings demonstrate the helpfulness of using participatory design to create educational games for seniors.

Black trans men navigate digital spaces that simultaneously marginalize them and serve as vital sources of information, community, and identity formation. While existing research explores the online information practices of marginalized groups, limited scholarship specifically examines how Black trans men use digital platforms to construct identity, access resources, and build community.

This study investigates their digital information practices within sociotechnical systems that structure visibility, participation, and belonging. It addresses three core research questions: (1) What are the digital information practices of Black trans men? (2) What types of digital communities and technologies serve as information resources? (3) How do online spaces influence identity formation, visibility, and engagement? The research uses Critical Technocultural Discourse Analysis (CTDA) and semi-structured interviews to examine how Black trans men curate, share, and interpret information online. This research also uses a theoretical framework shaped by Black ontology, epistemic resistance, and intersectionality to understand the role of power and exclusion in digital information practices.

Overall, this work analyzes systemic power dynamics in knowledge production and digital exclusion through an examination of digital platforms as locations for both constraint and resistance to contribute to discussions about race, gender, and information marginalization.

In China, the influence of large language models (LLMs) in text creation has been growing rapidly. However, some of which may not reach users’ expectations. Taking China’s publishing industry as an entry point, this study systematically explored the practical application of LLMs. Four China’s well-known LLMs were selected. The dataset was constructed by collecting the advertising texts of the top 100 books on the 2024 sales ranking of the Top1 e-commerce book sales platform “Dangdang.com” (like Amazon Books in America), followed by data annotation by two publishing major postgraduate. The study reveals that current China’s LLMs have shortcomings in capturing the core selling points of books and identifying users’ pain points. Meanwhile, they also exhibit inaccuracies in text emotion analysis and linguistic style discrimination. To address these issues, this study proposes optimization strategies that incorporates the “Pain Point–Selling Point–Solution–Evidence” logic framework into prompt engineering, along with fine-tuning the model with emotional vocabulary data, to enhance the model’s text creation effectiveness in the publishing field.

Informed consent is considered a fundamental ethical and legal principle in research, with requirements determined at the national level, resulting in differentiated levels of compliance across countries. With such requirements, what are the implications for Indigenous communities, where collective consent and relational ethics matter most (Smith, 2012)? This poster explores refusal as a generative Indigenous methodological and ethical stance, grounded in sovereignty, care, and accountability (Tuck & Yang, 2014). Through a comparative case study of Amos Bad Heart Bull’s ledger drawings and the isiXhosa initiation ceremony of ulwaluko, we argue that refusal is not silence, but an active form of cultural governance that determines what knowledge may be shared, by whom, and under what conditions (Simpson, 2007). Drawing on the work of Simpson (2007), Smith (2012), Tuck and Yang (2014), Gasa (2007), and Hlabangane (2021), this poster interrogates authorship, feminist positionality, and the role of land as ceremonial actor. It advances a cross-cultural framework for honoring refusal as a sovereign act of care and knowledge protection. This poster invites the audience to disrupt their notion of informed consent and to engage in the difficult conversation of refusal in library and information science research.

Public libraries are important support for refugees, but how they respond to trauma and special needs remains underexplored. This study examined refugee services in U.S. public libraries and combined the perspectives of librarians and refugees. Data was collected through a national survey of 229 library staff, five focus groups with librarians, and five focus groups with refugees to discuss the potential of trauma-informed care (TIC) to improve library services for refugees. Results showed that many public libraries serve refugees, partner with social workers, community organizations, and governments, and have begun to consider refugees’ trauma and integrate TIC by training staff and creating a welcoming environment. Future research will focus on detailed data analysis and customized TIC training.

This study examines how academic structure influences research focus in Library and Information Science (LIS) schools. We analyzed 16,671 publications from 745 faculty across 42 U.S. programs. Schools with independent LIS units demonstrate diverse research portfolios, while those sharing administration show variations: computer science affiliations emphasize technical topics with minimal traditional library research, education affiliations focus on library science and metadata, and communication affiliations balance traditional areas with social media research.

The rapid advancement and adoption of generative artificial intelligence tools in academia have raised critical questions regarding policy readiness and ethical governance, especially in Higher Education Institutions (HEIs) across the Global South. This study investigates the existence of institutional AI policies in three Sub-Saharan African HEIs and explores their implications for academic activities. Using the Times Higher Education rankings, we selected the top ten HEIs in Nigeria, Ghana, and South Africa each. We first used structured website content analysis across institutional portals, graduate school portals, and library webpages to identify published AI policies. This digital audit was supplemented by follow-up information interviews with key university personnel to verify the existence or absence of such policies. Guided by our preliminary research questions, preliminary findings show that while a few HEIs in South Africa have published and implemented institutional AI policies, no HEI in Ghana or Nigeria currently has a formal or publicly accessible AI policy in place. In these countries, academic staff use their discretion to make subjective judgments when evaluating student submissions suspected to be AI-generated. This creates inconsistencies, which in practice raises concerns about fairness, transparency, and accountability.

Artificial intelligence (AI) promises to streamline workplace information practices, yet little is known about adoption in public-sector contexts. We surveyed 103 Israeli public-sector knowledge workers on (a) their willingness to employ AI for 16 work-related information activities and (b) their actual use of such tools. Principal-component analysis grouped the activities into three categories: strategic & analytical tasks, management tasks, and personal applications. Across all categories, paired-t tests showed a large willingness–use gap (e.g., strategic tasks: M = 3.67 vs. 1.88, p < .001). Multivariate regression revealed that AI availability strongly predicted use (η² ≈ .14–.22) but weakly predicted willingness (η² ≈ .04–.05). Results suggest that providing access is necessary but not sufficient; motivational and organizational factors moderate adoption. We outline design and policy steps—trust-building, role-aligned training, and low-risk trial contexts—to bridge the gap and advance human-centered AI in the public sector

This study examines the application of agentic AI to evaluate the effectiveness, accuracy, and efficiency of academic libraries' responses to student inquiries. Data were collected by analyzing frequently asked questions (FAQs) from the University of Wisconsin-Milwaukee library website using OpenAI’s large language models (LLMs), o4-mini. The analysis focuses on identifying the types of questions that are most appropriate for AI-assisted responses and how agentic AI can support librarians in delivering targeted and timely assistance. The findings reveal that agentic AI can significantly improve the quality of library services by providing more timely and targeted student support. This research seeks to improve the effectiveness of librarian responses, strengthen librarian-student interactions, and support the meaningful integration of agentic AI into academic libraries.

Digital tools, including notation software, digital audio workstations, and other technologies have become ubiquitous in the process of music composition. Composers likewise use digital means to document their lives and careers. Musical scholarship and the cultural record relies upon continued documentation. Yet, obsolescence of hardware and software, the fragility and variety of digital file formats, media prone to degradation, and poor metadata place this material and subsequent scholarship at risk. This poster presents survey results of composers regarding their process, workflows, and knowledge of preservation practices. Our findings illustrate the challenges faced by music creators, and will inform the creation of a toolkit to bridge the gap between composers’ personal preservation practices and their archival partners’ ability to provide access to the historical record.

This study investigates users’ identification of virtual influencers in social media. A mixed-methods study consisting of an eye-tracking experiment and an interview with 60 participants was conducted. The results show that users can identify VIs with moderate accuracy. Users have relatively comprehensive but insufficient identification criteria and complex attitudes toward VIs. These findings suggest implications for AI and media literacy education.

Drawing on the Uncanny Valley theory, we conducted an online experiment to examine how avatars with varying human-likeness (i.e., real human, deepfake, human-like, and cartoon) affect homophily, message credibility, and concept understanding in anti-scam public service announcements (PSAs). Results showed that the human-like avatar was perceived as less homophilous but more credible. Additionally, the deepfake condition closely mirrored the outcomes of the real human condition, suggesting it may cross the Uncanny Valley. This research offers timely insights for the deployment of AI-generated avatars, particularly in guiding how to calibrate avatar realism.

Through this inquiry, we identify associations between immigration-related terms in the library and information science (LIS) literature over the last decade. We report on an exploratory data analysis of selected keywords in immigration-related LIS scholarship, interpreting legal–criminalized and individual–social and societal distinctions. Our preliminary findings suggest that associations between migration-related keywords are essential for LIS scholars to reflect upon, and we emphasize the responsibility of wielding keywords with attention to the connotations they may reinforce or reimagine.

Textual analysis remains challenging for researchers without programming expertise, creating barriers in social sciences and humanities research. This poster presents XXXX Tool 2.0, an enhanced web-based application that democratizes sophisticated textual analysis through natural language processing technologies. Building upon version 1.0, XXXX Tool 2.0 significantly expands data source integration to include HathiTrust Digital Library, Dimensions AI Database, and PubMed, alongside existing Scopus, Web of Science, Lens, and customized file support. Enhanced analytical capabilities include established methods for topic modeling, network text analysis, keyword stemming, in addition to new features such as burst detection, scattertext visualization, and sentiment analysis.

Bringing museum information resources online benefits the public and researchers. Despite decades of digitization, open access, and collaborative data sharing initiatives, few empirical studies have evaluated information retrieval effectiveness and open data quality. This poster uses web-based content analysis, descriptive statistics, and a case study to review over 180,000 Chinese antiquities in 61 U.S. museums. Findings reveal intertwined curatorial and cataloging practices, challenges around the structural-flexibility tradeoff, and limited inter-institutional data interoperability. This poster contributes to museum informatics, digital scholarships, and provenance research.

This study investigates professional designers' integration of AI tools within workflows across five key stages: Discovery & Framing, Ideation & Exploration, Development & Structuring, Refinement & Decision-making, and Finalization & Delivery. Semi-structured interviews with 14 designers reveal varied AI engagement, shaped by creative intent, identity, and AI trust. A heatmap illustrates AI supporting efficiency and creative tasks (ideation, content generation, refinement); interpretive tasks requiring contextual judgment are human-controlled. AI is viewed as an assistive co-creator accelerating iteration and automating labor, not replacing core creativity. Offloading procedural tasks and early ideation to AI enables deeper focus on conceptual and critical decisions. Designers use creative agency to reconfigure workflows, fostering reflective, conceptual work. Findings suggest further exploration of human-AI interaction, on how designers balance human-centered values and creative performance, maintaining agency and ethics.

This study initiated the development of a new scale to measure perceptions of information credibility on the web, distinguishing between trustworthiness and expertise as core dimensions. Building on a previous analysis of the literature on credibility assessments of information on social media, we generated a pool of 15 items, six for trustworthiness and nine for expertise. Following established scale development guidelines, we refined the scale through expert reviews and cognitive interviews with users. Experts (n = 2) rated most items as highly relevant and offered suggestions for refinement, such as including new items. Users (n = 9) confirmed item clarity but raised concerns about certain terms and perceived objectivity. These preliminary findings support the scale’s relevance and will inform revisions for subsequent validation. Ongoing work will expand participant feedback and move toward large-scale empirical testing.

This study investigates the viability of artificial intelligence (AI) supplementation in librarianship, focusing on its potential to enhance, rather than replace, reference and information services in public libraries. With the growing integration of AI technologies in various professional fields, including information science, this study seeks to explore how AI can be ethically and effectively implemented to support librarians while preserving the profession's core values of personalized service, intellectual freedom, and privacy. The study employs a quantitative research design, utilizing an online survey distributed to public library staff across the state of Connecticut, United States of America. A total of 62 usable responses were received. Findings indicate a complex landscape of attitudes toward AI integration. While many respondents acknowledged AI's potential to streamline administrative tasks, improve reference efficiency, and assist in cataloging and data management, significant concerns were also raised. Key apprehensions include data privacy, job displacement, algorithmic bias, and the potential erosion of critical thinking skills and personalized patron service. Participants frequently stressed the importance of maintaining human oversight and ensuring equitable access to AI technologies. This research highlights the necessity of balanced implementation strategies that emphasize transparency, staff training, and clear ethical guidelines.

Modern virtual reality (VR) and augmented reality (AR) tools possess advanced capabilities that are reshaping archaeological education for public audiences in venues such as museums and schools. Research is needed to determine how these applications modify public perception and interaction with history and cultural heritage. This paper examines the perceptions of academic archaeologists regarding the integration of VR tools into archaeology education. We conducted semi-structured interviews with six archaeologists who either teach archaeology in a university or museum setting, or who work with archaeological materials in the field. We identified the major concerns and perceived benefits of VR integration in these environments. Our work contributes an important baseline assessment of attitudes and practices within the field so that we may better understand the existing gaps among archaeologists and historians, as well as the extent of those gaps.

As the use of machine learning expands across domains, the importance of data documentation practices and corresponding frameworks has grown to facilitate transparency. While the development of metadata frameworks is one of the core concerns in the information science field, little is known about how information science professionals contribute to data documentation framework in ML contexts. Therefore, this study explores information science professionals’ current engagement in data documentation framework development for ML studies. We conducted a literature review in a systematic way with the assistance of a generative AI tool, followed by identification of authors’ professions. Our preliminary findings show that information science professionals have play limited leadership roles in developing data documentation frameworks and tend to serve supporting roles instead. We conclude by acknowledging limitations and identifying areas for further analysis.

National programs for educational technologies often emphasize online learning, built on claims of universal access and improvement. However, these narratives frequently overlook the complex educational contexts of marginalized rural communities, such as those in Nigeria. This paper reframes rural educators in Nigeria as information professionals who engage in community-specific knowledge dissemination, advocacy, and literacy development. Drawing on critical rural theory, decolonial pedagogy, and information behavior models, this conceptual paper argues that rural educators enact the very functions ascribed to information professionals. As a work-in-progress paper, it calls for contextually relevant educational practices and outlines future directions for empirical study, policy, and LIS curriculum development.

Public library staff often express feeling unprepared to engage with people experiencing crises that require des-escalation. This poster details the iterative development and refinement of a Generative AI (GenAI) chatbot designed to enhance immersive virtual reality (VR) training simulations for public library staff. Building on previous work exploring VR for de-escalation training, we trace the evolution of our simulated patron—from "Michael 1," a live-roleplayed avatar in Mozilla Hubs, to "Michael 2," an initial GenAI chatbot integrated within the Engage platform, and culminating in "Arthur," a more sophisticated GenAI chatbot being developed in a custom Unity environment using Python, C#, Blender, and various APIs (TTS, STT, LLM). This project focuses on leveraging GenAI to create realistic, interactive practice scenarios, aiming to improve staff confidence, skill development, and ultimately foster more effective, empathetic crisis response in public library settings.

Research on the benefits of students’ use of generative Artificial Intelligence (AI) has been notoriously mixed: some findings suggest AI use can enhance learning and development while others suggest it can be harmful. Adding to this lack of clarity is the fact that there is little available literature regarding the prevalence of AI use among students, the student factors that contribute to AI use, and the downstream effects of students’ AI usage – factors likely important for understanding the roles AI use might play in educational outcomes. Institutions report observing rapid increases in AI use among students, however associated research has not kept pace with rapid advances in AI capabilities, accessibility, and potential uses. In addition to clarifying the impacts of AI on students, institutions need this information to inform policies and effective practices regarding use of AI in education. This study summarizes the available data on generative AI usage from the 2023–2024 Healthy Minds Survey (HMS), a cross-sectional study of United States of America (USA) college students, to address this knowledge gap.

As in-vitro fertilization (IVF) becomes increasingly accessible, patients are turning to social media to supplement clinical information and emotional support. This study investigates the fertility-related content shared on Facebook by IVF healthcare providers, focusing on Traditional Chinese-language posts from both individual physicians and organizational accounts. Using keyword-based searches, we identified 40 public accounts and analyzed 4,285 posts published in 2024. Posts were categorized by month, content theme, media format, and interactivity. Findings revealed clear differences in communication strategies. Personal accounts emphasized outpatient information, educational explanations, and service promotion, with peak posting activity in the fourth quarter. Organizational accounts focused on case sharing and internal work documentation, maintaining a steadier posting rhythm throughout the year. Personal accounts predominantly used image-and-text formats, while organizational accounts more frequently employed short videos to enhance engagement. This study highlights how IVF-related health communication varies across account types, offering insight into how content is framed and delivered in Taiwan’s digital health landscape. The findings inform future strategies for more audience-specific and platform-appropriate reproductive health outreach.

The goal of eradicating friction has driven the development of information spaces. Yet, the problems associated with a loss of friction have led many to suggest ways to reintroduce friction. Unfortunately, libraries are often overlooked in these discussions. This poster outlines an initial framework for the role of librarianship in introducing necessary friction in ways that improve the quality of society’s navigation of information spaces.

To address the need to study the role of social media during political crisis in non-Western context, this study investigates Facebook digital activism during the peak of the July 2024 student-led revolution in Bangladesh. Through thematic and lexical analyses of 4,000 comments from 60 protest-related posts, we found that most comments (57.8%) in our sample were ideological, unprovocative, and hashtag-driven. Provocative comments (24.6%) exhibited higher vocabulary density and rhetorical variation than unprovocative comments. While some comments supported the protest, others attempted to delegitimize it. By focusing on this political non-Western crisis, this study challenges the binary view that provocative trolling and digital activism are mutually exclusive and instead it underscores their overlapping roles in online political discourse.

This study investigates data risks in human-AI interaction within personal health management apps and systems. Drawing on literature, cases, and policies, it analyses data types, governance practices, and potential risks. Using PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines and thematic analysis, 116 articles, 9 cases, and 4 key policies were examined. Identified data types include physiological indicators, lifestyle habits, personal identity, and social relations. Governance practices and risks are categorised into six dimensions: governance subjects, policies and standards, technologies and facilities, lifecycle measures, data rights protection, and risk management. The study offers governance strategies for human-centred AI and lays the groundwork for further assessment of risk types and severity.

We introduce AnnoLoom, a tool designed to assist researchers with codebook development, annotation tasks, and evaluation of human vs. AI’s annotation results. AnnoLoom contributes to human expert-AI collaboration and its efficacy in the context of using Large Language Models (LLMs) for research involving text-based data. We conducted a cognitive walkthrough to iteratively improve the design of AnnoLoom. We discuss the design of the tool and future work in evaluation.

Social facilitation—formal and informal training and assistance in using information systems—is crucial for enabling people to access, use, and benefit from ICTs. To reduce digital inequalities, it is important to understand how marginalized communities overcome barriers and sustain effective ICT use. Based on ethnographic fieldwork with a campesino organization on the border of the Colombian Amazon, this research examines the forms of social facilitation that support meaningful ICT engagement in a community served by remote infrastructures. Findings show that formal social facilitation in this setting mirrors the unreliability of physical infrastructure, offering support through intermittent interventions by state agencies, NGOs, and researchers. Community members employ four key practices to derive meaningful knowledge from these sporadic interventions: repeated training as habituation, note-taking, tinkering, and extending incursions online. Between these external efforts, the community collectively addresses ICT challenges, forming emergent local spaces of ICT support that enable situated and ongoing transitions to the information age. This research sheds light on the everyday digital practices of communities often excluded from dominant narratives of the digital society. It also expands the concept of social support beyond formal venues such as libraries, community technology centers, and cybercafés.

Query reformulation is a critical aspect of the exploratory search model. During exploratory browsing, searchers issue broad queries to learn about the breadth of the topic; during focused searching, they issue narrow queries to learn about a specific aspect of the topic. Typically, little support is provided by the search interface to help searchers refine and reformulate their queries. This research addresses this gap, demonstrating a generative AI approach to suggesting both broad and narrow queries. Using a timeline of past search activities (queries issued and search results saved), the approach allows searchers to reformulate their queries based on specific saved search results, previous queries issued, or the current query. This gives the searcher interactive control not only over whether to request broad or narrow query suggestions, but also upon what information to base those suggestions.

The relationship between data governance and data organization is cyclical and mutually reinforcing. Data governance defines the frameworks, standards, and roles that shape how data should be organized, accessed, and used. These guidelines ensure that data is secure, consistent, and aligned with institutional goals. At the same time, the structure and management of data – the core of data organization – significantly influences governance. As institutions evolve, changes in how data is stored, shared, and analyzed often necessitate updates to governance policies. Well-aligned data governance not only standardizes data definitions and security measures but also breaks down silos and encourages cross-unit collaboration for the common institutional strategic goals. This dynamic interplay is particularly important in complex, decentralized environments like higher education. Adopting Anthony Giddens’s Structuration Theory (Gidden, 1984) and Orlikowski’s Structurational Model of Technology (Orlikowski, 1992, 2007), this study proposes a new theoretical framework – Structuration Data Governance Model – to analyze the interaction between data governance and data organization. It also suggests practical applications of the theory.

Online discourse about contentious and ideological issues like undocumented immigration, trigger massive user reactions. Online hostility towards marginalized groups and its effect on them have attracted scholarly attention. However, reactions to posts on the same issue may take different sides; they may vary not only because of the specific views expressed, but also because of the perceived social roles of those who post about the issue. We examined how users’ provocative comments vary among posts by the public, politicians, and immigrants. Given this controversial topic, we found that almost 85% of the 4,800 comments were provocative. We found that the prevalence and nature of these comments varied significantly between posts by the public and those by either by politicians or by immigrants; We also found significant variations between posts on Facebook and on X.

Artificial intelligence (AI) now appears on the search page and this may influence how users interact with search results. This study examined how this development in the information seeking process affects people’s Health Information Literacy (HIL). Using a HIL measurement for literacy indicators, young adults were presented with a simulated Google search page. Findings show that participants disliked Google’s design for this task. Frequency coding revealed that only 17 of the 41 participants used AI to improve their health insurance literacy. Participants revealed that they felt AI should provide a simplified explanation of health insurance information.

Qualitative data analysis (QDA) is a process of knowledge construction through human interpretation. Designs for computational tools supporting the QDA have abounded for decades, and most recently, have focused on how to infuse Artificial Intelligence (AI) supportive features. Yet, despite communities of qualitative researchers’ routine engagement in sensemaking, frameworks portraying their moment-by-moment engagement with data remain largely underexplored in the qualitative methodological literature. This gap constrains designs of human-centered, epistemologically aligned AI-infused Computer-Assisted Qualitative Data Analysis Software (AI-CAQDAS). Unlike traditional CAQDAS, which focused more on supporting data organization and retrieval, AI-CAQDAS is designed to assist with qualitative sensemaking and moment-by-moment interpretation of data. This proposed study investigates how qualitative researchers construct and shift their epistemic relationships with data during analysis, foregrounding the role of researcher positionality, including disciplinary background, social identity, and methodological commitments, in these interactions. Using a think-aloud protocol on QDA tasks with participants who are experienced qualitative researchers from diverse disciplines, we apply positioning theory to make visible the often-invisible processes of qualitative data sensemaking to challenge the prevailing code-and-retrieve logic embedded in current AI-CAQDAS designs. We argue for an urgent need to design AI-CAQDAS that supports nuanced human conversations at the heart of qualitative sensemaking.

AI bots, deepfakes, and the misinformation they generate on social media prompt users’ behavioral and affective reactions. While the relationships between social media use and emotion are frequently studied, less is known about how users' responses (e.g., fact-check, report posts/posters) are related to affective outcomes. This survey of undergraduates using X (known for AI bots and misinformation) found that Disappointed, Disgusted, and Angry were the top emotions after users responded to issues on social media. A latent class analysis identified four user profiles. The most prevalent class shows positive emotional outcomes, but mixed with anxious feelings. The latter was notable across multiple profiles, suggesting an area for intervention.

This paper analyzes 25 health information behavior studies published in the period of 15 years (between 2009 and 2023) using eye-tracking as the research methodology. Eye-tracking technology has become a valuable tool to study information behavior. This study examines the ways of adopting eye trackers in health information behavior research and how the research is designed with an extended method the technology provides. This paper contributes to understanding methodological trends for health information behavior research.

This pilot study applies intersectionality and insider/outsider dynamics to examine mental health information-seeking practices among Chinese LGBTQIA+ immigrants in the US. Through semi-structured interviews and thematic analysis, the findings reveal that intersecting identities influence participants’ perceptions of insiders and outsiders when seeking mental health information. Participants further categorized insiders into social types based on behaviors, authority, and intimacy. Information and communication technologies (ICTs) facilitate interaction among trusted insiders across geographic distances. This study advocates for the design of culturally competent mental health information resources that reflect intersecting identities and insider trust dynamics.

Artificial intelligence has undergone an iterative evolution from ‘reactive’ to ‘discriminative’ to ‘generative,’ with its penetration into society becoming increasingly widespread. As a result, the cultivation of children's artificial intelligence literacy has become increasingly important. China's unique technological ecosystem, cultural context, and government priorities provide a favorable backdrop for research. However, we know little about the mechanisms and content of AI literacy development among Chinese children. In this exploratory study, we conducted interviews with children from different regions of China through both online and offline methods. Through coding analysis of the interview records, we found that Chinese children develop a series of skills through a bidirectional interaction between ‘cognition and behaviour.’ After independently mastering these skills, children internalize AI literacy. Based on this, Chinese children's AI literacy can be categorized into cognitive-associated literacy, behavioral-associated literacy, and skill-associated literacy. These findings provide insights for establishing a child-centered framework for AI literacy.

This work-in-progress study examines how researchers in medical and health sciences engage in data sharing, focusing on Data Availability Statements (DAS) from articles published in three mega journals indexed in PubMed Central. Preliminary results reveal a variety of DAS categories, each reflecting a distinct approach to sharing research data. The distribution of these types was broadly similar between articles with and without Korean-affiliated authors. This study could contribute to understanding how data sharing practices are evolving in medical and health research and highlights opportunities to better align global norms with national-level practices.

Over the course of 2020, as the COVID-19 pandemic was unfolding across the world, numerous information institutions implemented COVID-19 rapid response collections. Rapid response collections are an emerging collecting modality that focus on documenting an event as it happens or in the immediate aftermath. Growing literature reflecting on these collections typically tells the story of individual collections and focuses on the process of starting and building them. Five years after these collections initially began, there is now the opportunity to reflect on the scope and impact of rapid response collecting during the COVID-19 pandemic, as well as the status of these collections today. This pilot study, which aims to provide a foundation for a broader understanding of rapid response collecting across US GLAM institutions, assesses the presence, scope, and status of rapid response COVID-19 projects started at universities in Pennsylvania.

Palliative care is described as health service to patients and caregivers to improve quality of life when facing a life-threatening illness. We analyzed preliminary 46 TikToks tagged with #palliativecare (mostly uploaded between 2024 and 2025) to enable first insights into who as creator and what content stand behind the #palliativecare (RQ1). To this end, we developed a codebook (inductive and deductive approach). Almost half of the TikToks were posted by healthcare professionals, while 30% were created by patients. The content of the TikTok varies, while most of them are categorized as a Personal Experience (61%). Furthermore, we aimed to gain first insights by analyzing 10 comments on five randomly selected TikToks shared by different types of creators. We wanted to explore research potential surrounding affective information behavior and the emergence of parasocial relationships in response to highly vulnerable and/or emotionally charged content (RQ2). TikTok users provide emotional support by showing sympathy, prayer and are also asking questions. This work-in-progress study enables first insights into relatable and personal content considering palliative care. These insights could be used to improve health communication and deepen our understanding of belonging within these vulnerable, emotionally charged communities.

Data governance in practice is shaped by a complex interplay of socio-legal frameworks, institutional arrangements, cultural attitudes, and technological capabilities. Understanding these interlinked dimensions is essential for effective management of information systems, particularly in sectors that are highly data-dependent for forecasting and delivering social services, like transportation. This study explores these dynamics through the specific case of vehicle speed data governance. Employing a mixed-methods approach, the research combines computational topic modeling using Latent Dirichlet Allocation (LDA) on a corpus of 237 policy and institutional documents with qualitative analysis to identify key governance themes. Findings reveal the prominence of national reporting mandates and technological adoption at the local level, alongside notable gaps in data lifecycle management and stewardship. To complement these insights, forthcoming interviews with data managers aim to clarify governance processes and challenges not evident in policy texts. This work advances our understanding of how socio-technical and legal factors collectively influence the governance and effectiveness of urban data infrastructures.

Game-based learning represents a promising avenue for educators to increase student engagement and academic results. However, research has found mixed results in the effectiveness of games to improve various learning outcomes. Researchers view artificial intelligence (AI), particularly large language models (LLMs), as augmenting technology to increase the interactivity of games, yet it’s unclear how effective integrating generative AI within games is for improving student learning. This study seeks to identify how researchers in various disciplines have incorporated AI into educational games through a preliminary systematic review using the PRISMA framework. After analyzing 12 empirical studies, we found that researchers largely compared generative AI games against traditional teaching methods through experimental studies. Moreover, these studies largely reported positive results for the LLM-integrated games in increasing engagement, particularly in STEM and AI literacy domains. However, further research should be conducted to understand the effectiveness of LLM-infused games in improving long-term learning outcomes.

This poster details the findings from a study investigating the efficacy of community based and organized information sessions around New York Police Department smart city surveillance practices in New York City. The purpose of this research is to better understand localized perspectives on surveillance infrastructures within New York City and promote community dialogue through local public libraries and community groups as a mode of information activism.

Doctoral students face a lack of information early in their programs, creating an information need that may be addressed through peer interaction. While prior studies have looked at this interaction in structured environments, much less is known about unstructured ones. This paper reports on a study that explored peer interaction and information seeking and sharing of doctoral students in unstructured environments. Based on twenty semi-structured interviews with social sciences and humanities students, it was found that information on academic, administrative, personal, social, and tips and tricks is exchanged among peers. The study underscores the importance of having access to unstructured environments (e.g., shared workspace) to satisfy the information needs of doctoral students.

When chatbots lack physical bodies, how relatable should their body advice be? Concerningly, when an AI chatbot generates health information, it is unaware of its audience and contextual sensitivities. This study explores whether AI’s neutral tone compensates for its lack of framing in this context. It examines which contextual sensitivities young women preferred from Google’s AI chatbot (Gemini) about diet advice and why it matters. Findings suggest that young women feel AI lacks contextual awareness when framing sensitive diet issues. Specifically, participants feel the AI prioritizes a “neutral” or “technical” tone when they prefer a reassuring framing of the information.

Consumer health informatics is becoming increasingly important in a world where more than 5 billion people rely on the internet and social media for health information. eHealthyinfo is a physician-initiated and reviewed artificial intelligence (AI) applied digital health information website, aiming to be essential in advancing health literacy and promoting preventive health behaviors. To assess the effectiveness of its digital content, we conducted a mixed-methods evaluation of 13 wellness and nutrition videos randomly selected from eHealthyinfo, representing 30% of its relevant content. Two independent reviewers used the Patient Education Materials Assessment Tool for Audiovisual Materials (PEMAT-A/V) and the modified version of the CDC Clear Communication Index (CCI) to evaluate each video. Website log analytics were also collected to assess user engagement. The average PEMAT scores were 89.5% for understandability and 100% for actionability, while the average CCI score was 94.4%, indicating understandable, actionable, and clear content. eHealthyinfo demonstrates a strong potential as a digital health education tool, with high-quality, user-friendly content.

The widespread use of generative AI (GenAI) is seen as beneficial for team collaboration, yet full access for all members is often impractical in real-world settings. This study explores how different GenAI integration structures—no access, unequal access, and full access—affect team performance, focusing on the mediating roles of team communication style differences and communication effectiveness. Findings from a lab experiment involving 60 two-person teams reveal that teams with unequal GenAI access demonstrate higher communication style differences and superior performance compared to teams with no AI or full access to AI. Two mediators—team communication style differences and communication effectiveness—sequentially mediate the impact of unequal AI access on enhanced team performance. This research provides theoretical insights into the effects of varied GenAI integration structures on human-AI collaboration and offers practical guidance for optimizing team design in human-agent systems.

This study investigates data literacy and its impact on information anxiety through the mediation of cognitive overload. We surveyed students, faculty/staff, and community library users at University of Rochester (UR) Libraries during the Data Bloom 2024 and Love Data Month 2025 events. Over 100 participants self-reported their information anxiety pre- and post-training using Likert scales. Partial Least Squares Equation Modeling showed significant association between data literacy, cognitive overload, and information anxiety, with cognitive overload serving as a strong mediator. Multigroup Analysis further revealed that the effect of data literacy on post-training anxiety varied across participant roles and disciplines, with students and health/social professionals benefiting more than STEM participants and faculty/staff. The findings highlight the value of academic libraries in promoting data literacy and reducing anxiety around data use.

This poster reports preliminary analysis of an online survey of 191 college students in South Korea examining their use of Generative Artificial Intelligence (Gen AI) tools, perceived benefits, and institutional supports. Participants mainly used Gen AI for various purposes in academic/work-related contexts, such as Assignments/Writing Reports, where most participants perceived the benefits. However, participants also used Gen AI in everyday life/non-work contexts, notably for areas such as travel planning, entertainment recommendation information, and mental health improvement methods. Only 33.5% of participants received support for using Gen AI from universities/university libraries, which suggests opportunities to help students learn how to use Gen AI effectively and responsibly. This research contributes an empirical study of user behaviors around Gen AI use in rapidly changing sociotechnical environments, for academic/work-related and everyday life/non-work contexts, calling for more institutional support and human-centered research for equitable access to and effective use of Gen AI for college students.

This analysis uses Derridean deconstruction and reconstruction to examine the relationship between privacy, the at-home experience, and hospitality from the perspective of information ethics. It is argued that informational privacy is foundational in ensuring the informational at-home, establishing one as host rather than hostage in the informational spaces required for the development of being and knowing. The informational at-home, in turn, informs the possibility of hospitality. Hospitality is deconstructed beyond industry or transaction, instead revealing an etymological and conceptual depth that grounds the possibility of information ethics.

Unhoused patrons are uniquely vulnerable to marginalizing public library policies. To first capture the scope of the problem and then to offer a lens for productive critique, I perform a content analysis on the behavior policies of all thirty-five library systems in Los Angeles County, California, USA, describing the methods by which libraries deny unhoused patrons access to library spaces. Then, by applying Iris Marion Young’s work on “enabling justice” to these policies, I give a name to these processes: oppression. I argue that if public libraries are to live up to their name, policies that affect public access need to be viewed through a justice lens. Public libraries that truly are “public” need to design policies that enable justice for all members of their public, and the methods I apply to the Los Angeles County case study offer tools to aid that transformation.

Social movements have been using social media to share their narrative which supports their goals, values, and grievances, and it usually spreads across platforms in visual form. This research creates a framework to specifically look at how social movement images are structured into a narrative based on their visual semantic relationships. In the framework, the images go through feature embedding, and zero-shot captioning to create a network of semantic relationships. These relationships are represented in an ontological network where the labeled images are nodes and their descriptions are the edges semantically connecting them based on similarity scores. This network is analyzed using community detection models to understand the latent visual semantics supporting social movement images. Comparing the feature labels and visual semantic image captions graph to a graph of semantically similar images proved to be somewhat difficult for domain-specific images like social movements. The combined framework analysis shows the various nuanced themes that arrive even within grouped classes.

The rapid evolution of artificial intelligence (AI) continues to reshape how information is produced, shared, and accessed, raising urgent questions about equity, accountability, and fairness. This poster, inspired by the upcoming volume Assessing AI and Social Justice Intersections: Challenges and Opportunities, explores the intersection of AI and social justice, highlighting select forthcoming chapter content from real-world experiences and challenges, community-driven solutions, and varied theoretical and computational frameworks. The poster brings together work of librarians, educators, researchers, and community advocates to examine the ethical dimensions of AI and its tangible consequences for marginalized groups. It draws on diverse philosophical, methodological, ontological, and practice-based perspectives to critically evaluate AI’s potential both as a mechanism for empowerment and force capable of exacerbating existing inequalities. Issues of algorithmic bias, inclusive data governance, and culturally responsive AI implementation provide an insightful, timely, and comprehensive examination of AI through the lens of social justice imperatives. The poster provides multidisciplinary and international coverage of ethical frameworks, resistance strategies, educational applications, data autonomy, cultural competence, and equity in information systems design. Algorithmic oppressions in information retrieval, equitable implementation of AI in libraries and educational institutions, culturally responsive AI practices, and community-driven AI initiatives are included.

This paper takes some first steps to trying to define metadata that may curb some issues related to digital ephemera, such as context collapse and information decay. We attempt to show aspects of cultural context as related to how memes (each of which are collections of visual artifacts that carry a cultural idea) are indexed by Google Cloud Vision (GCV). Considering three meme collections that differentiate in their visual content’s expression of an “idea,” we contribute some exploratory metrics. These metrics attempt to account for locations of cultural ideas across web domains using Gini coefficients as a measure of distribution across web domains.

In elementary through higher education, data-driven technologies have long been proselytized as forward-looking answers to the financial problems that stem from neoliberal gutting of public resources (Besser & Bonn, 1996; Eaton, 2022; Newfield, 2016). This project consists of a two-part study about AI in education based on a survey of 500 members of the American Association of University Professors (AAUP) concerning workers’ perspectives on technology deployment in education and 15 interviews with union members, which include graduate students, contingent faculty, full-time faculty, librarians, and staff, to understand how those on the ground conceptualize and prioritize technology issues related to transparency, surveillance, deskilling, and work intensification, and how they see claims of AI’s capacity to increase accessibility.

The intersection of technology and the legal profession has evolved significantly, with legal practitioners using tools like Westlaw, LexisNexis, and Google Search for legal research. More recently, artificial intelligence (AI) tools, such as ChatGPT, have been integrated into legal practice, offering both promise and challenges. In particular, incidents like the Mata v. Avianca case, where attorneys faced sanctions for submitting fictitious citations generated by ChatGPT, highlight the risks of relying on AI in legal work. While courts have issued guidelines for the responsible use of AI, there remains a clear need for due diligence and a thorough understanding of its limitations. This study evaluates the effectiveness of AI tools in legal research by presenting two complex legal questions, one involving U.S. asylum law and the other concerning Colombian legal principles. The responses from ChatGPT and Google’s AI were analyzed for accuracy and consistency, revealing significant gaps in both tools’ performance, particularly in providing complete and current legal information. While they can assist in research, they are not yet reliable enough to replace expert legal analysis. The findings suggest that legal professionals should approach AI-generated legal information with caution and verify results with human expertise.

This study presents preliminary findings on health information professionals’ perceptions of the importance of reliable healthcare information. This study analyzes 336 responses from health information professionals across 61 countries. The results reveal statistically significant differences in perceived needs for health literacy promotion and the enforcement of professional standards and ethics across country income levels. These findings highlight the importance of tailoring equitable healthcare information policies to address global disparities.

Cloud storage is widely used by researchers for collaborative projects, offering clear benefits for data sharing and collaborative work. However, managing shared files in cloud storage can be challenging, especially when collaborators have different personal information management practices. By conducting a grand-tour interview and semi-structured interviews with 30 researchers across disciplines in the United States, this study explored different strategies researchers use for managing shared files and identified three successful cases. These cases revealed effective strategies, including developing a set of rules for managing files, creating a guide file, and providing a training session. In particular, the results showed that even small efforts can significantly enhance the organization and accessibility of research files. This study deepens our understanding of shared file management and provides practical strategies that can support more effective shared file management for collaborative research projects.

This study analyzes two avian metaphors — “Canary in the Coal Mine” and “Pigeon packet”—in cybersecurity discourse. Using Kueffer and Larson’s (2014) metaphor criteria, alongside Ritchie’s (1999) incongruity theory and Koestler’s (1964) concept of bisociation, we examine their factual grounding, rhetorical function, and humor effects. While both rely on incongruity for impact, their effectiveness varies across social and cultural contexts. Findings highlight how metaphor and humor can enhance engagement and risk awareness in cybersecurity communication.

Explaining cybersecurity concepts to the public is often challenging due to their complexity, leading to the wide use of metaphors, but their effectiveness remains under-explored. Building on prior work identifying eight metaphor categories, this study engaged six cybersecurity experts to evaluate their appropriateness and communicative value. Preliminary findings suggest replacing dominant fear-based metaphors like war and crime with health and risk management frames that better promote public engagement and responsible behavior. Experts also offered guidance on crafting effective metaphors, emphasizing contextual clarity, precision, and relatability. By foregrounding key considerations in metaphor design, this study lays groundwork for building more responsible and accessible cybersecurity knowledge infrastructures, contributing to broader efforts in human-centered science communication

This preliminary study explores citizens' perceptions regarding the adoption of artificial intelligence (AI) in healthcare, grounded in Cultural Cognitive Theory. A total of 10,523 responses from the Pew Research Center's American Trends Panel Survey Wave 119 were analyzed using Pearson’s Chi-square test of independence. The findings indicate that citizens’ view, specifically their optimism or skepticism toward AI technology and their ideological stance, are significantly associated with their perceptions for AI use in healthcare.

KEYWORDS

Although generative AI is widely recognized for its potential to enhance team collaboration, unequal access among team members is often overlooked. To explore how this seemingly detrimental uneven AI integration impacts team composition and, consequently, team effectiveness, this study extends the classic Input-Mediator-Output model to an Input-Process-State-Output (IPSO) framework. A lab experiment with 60 two-person teams reveals that unequal AI access yields the highest productivity, improving both task quality and completion time compared to no or full AI access. This effect is driven by two key mechanisms. First, negative socio-emotional interactions and increased cognitive diversity serve as a positive serial mediation pathway linking unequal AI access to enhanced task quality. Second, unequal AI access leads to more concentrated and imbalanced questioning behaviors, accelerating task completion. This study provides a theoretical explanation of how AI integration structures operate in teams and offers guidance for the design and management of human-AI systems.

This paper revisits Kling's (1994) framework for analyzing technology discourse by applying it to expert commentary on Generative AI (GenAI) posted on X (formerly Twitter) from November 2022 to June 2023. Using thematic content analysis on 1,000 top-engaged posts, we assessed how Kling's five genres i.e., utopianism, anti-utopianism, social realism, social theory, and analytical reduction map onto current GenAI discussions. We found that 30% of the discourse fell outside these categories, prompting us to introduce a new genre: Enhanced Technological Realism, which includes Collaborative Tech Assessment (CTA) and Technological Futurism. This addition accounts for the evaluative and speculative tones present in today’s GenAI discourse. Our work updates social informatics theory for the era of decentralized, user-driven online discussions.

As Artificial Intelligence tools become increasingly integrated into library systems, concerns about privacy, security, and ethical implementation have grown alongside their benefits. This study presents a systematic literature review examining AI-powered recommender systems within library contexts, with a focus on their applications, associated risks, and governance. The literature reviewed highlights key security and privacy risks, including data poisoning, inference attacks, and algorithmic manipulation, which threaten core library values such as intellectual freedom and patron confidentiality. Existing policy frameworks, such as the GDPR and the AI Act, offer guidance on fairness and transparency, but a lack of library-specific governance standards persists. A balanced security approach is proposed, consisting of technological safeguards, AI awareness training among staff and patrons, and the creation of library-specific standards.

Technology-supported emotional wellness interventions have gained significant traction in recent years, yet many fail to address the unique needs of marginalized populations whose identities and experiences often go unrecognized in standardized approaches. Drawing inspiration from art therapy’s established success in utilizing creative expression for emotional processing, this research investigates how AI-generated content can serve as a bridge between traditional therapeutic approaches and culturally responsive care. Through a pilot study involving three participants, we identified key challenges and opportunities in leveraging AI-assisted creative tools for therapeutic and expressive purposes. Our findings reveal both technical limitations and promising pathways for developing more inclusive and culturally sensitive AI-powered visual expression platforms.

Diffraction is a well-known term in the context of quantum mechanics, where the nature of light and its behaviors can be observed. Diffraction as a critical research methodology was coined by Haraway (1990, 2013, 2018) and Barad (2003, 2007, 2010) in the context of critical science and technology studies (STS). Unlike reflection, where discourse and abstraction are leveraged to fill in the gaps between knowledge and representation, diffraction acknowledges the ontological inseparability of the material and discursive. While diffractive methodologies are growing in popularity in other critical research areas, there appears to be almost no explicit mention of diffraction in library and information research. One exception is Østerlund et al. (2020), who propose diffraction as a methodology to study trace data. In this poster, I describe diffraction as a methodology and situate my efforts to build on the existing literature and develop a diffractive methodology for critical LIDS research.

In vitro fertilization (IVF) is a common and effective treatment for infertility, requiring patients to access diverse information for informed decision-making. While healthcare professionals provide specialized reproductive advice, many patients seek additional emotional support and shared experiences from peers through online support forums. This study examines how patients undergoing IVF engage in self-disclosure within these forums, focusing on the topics, purposes, and effects of sharing personal information. Through content analysis of 600 posts and 1800 responses, the study reveals that patients frequently disclose information such as age, location, and medication history. The primary motivations for sharing include emotional venting and self-clarification. Patients mainly seek empathy and clarification, especially during critical IVF stages, such as starting treatment or confirming pregnancy. Findings suggest that timely emotional support from healthcare professionals, alongside clinical advice, could enhance patient well-being. Additionally, fostering open information-sharing within patient support groups can offer significant emotional benefits. Platform designers are encouraged to develop user-friendly, privacy-conscious environments that balance information sharing with security. This research highlights the importance of emotional disclosure in IVF communities and suggests ways to improve support structures for patients navigating infertility treatments.

While generative AI (GAI) is widely used in cultural heritage promotion, how humans engage with this technology in relevant artistic creation remains underexplored. This study conducted in-depth interviews to investigate the approaches, challenges and mitigations of human-GAI collaboration in such contexts. We found that human-GAI collaboration includes four stages: (1) creative context setting, (2) GAI tools evaluation, (3) creative concepts formation and (4) human-GAI collaborative production. In this process, two main challenges were identified: (1) cultural misrepresentation and (2) intellectual property. Relatedly, corresponding mitigations were proposed.

This study proposes the Tri-RealitY framework(TRY), an innovative framework for examining how individuals cognitively interpret and respond to misinformation when navigating inconsistent or ambiguous information. Grounded in incongruity theory, the framework identifies three dimensions influencing how individuals cognitively interpret information: General Reality (grounded in common knowledge), Contextual Reality (guided by situational cues), and Ideological Reality (shaped by personal beliefs and social identities). We argued that when individuals encounter inconsistent or ambiguous information, they engage in compensatory cognitive processes across these dimensions to restore interpretative coherence. Successful compensation can lead to the acceptance of contradictory messages, whereas unsuccessful compensation results in disbelief or rejection. Rather than defining misinformation solely by intent or factual accuracy, this framework reconceptualizes misinformation as a case of misaligned acceptance—a receiver’s failure to correctly interpret or compensate for informational inconsistencies. Thus, misinformation can misinformation can arise even when message types are correctly identified, reflecting deeper interpretive misalignments rather than mere factual errors.

This study presents an exploratory mixed-methods study investigating student perceptions of political ideological bias in university classrooms, directly addressing claims made by Turning Point USA's Professor Watchlist. Researchers surveyed undergraduate students at a large southeastern university, defining discrimination as negative impacts due to group membership and propaganda as flawed argumentation used to sway perceptions. While the majority of students reported no bias, those who did often confused normal classroom debates and differences of opinion with discrimination. The study also highlights that Republican students were more likely to perceive course content as discriminatory, whereas liberal students reported experiencing xenophobic language and stigmatization.

Living in a rapidly evolving digital environment, various forms of literacy are crucial for students to navigate the digital world. This study investigates the presence of literacy-related courses within undergraduate programs at iSchools. Using topic modeling, we analyzed 1,507 undergraduate course descriptions from 26 iCaucus schools in the North American region, identifying a total of 22 topics. Two noteworthy findings emerged: 1) only 28 literacy-related courses were identified across 12 institutions, and 2) these courses were scattered across eight different topics rather than forming a cohesive, standalone topic. These results reveal that literacy, in its various forms, is not currently a central focus in the design of undergraduate curricula within iCaucus schools. As a next step, we plan to expand our sample to include courses from iSchools across all membership levels and geo-regions and to further investigate how literacy-related content is integrated into broader curricular structures.

The ability of the public to access and use an Open Government Data Portal (OGDP) is essential for unlocking the potential of Open Government Data (OGD). Through OGDP, the government facilitates public access to its data and, at the same time, supports broader open government initiatives. The accessibility and usability of OGDPs remain critical, as they enable the public to engage with and derive value from OGD. To deepen understanding of these aspects, this study assesses the effectiveness of the OGDP user interface (UI) and user experience (UX) design through an evaluation of the Satu Data Indonesia (SDI) portal, the Indonesian national OGDP. The study involves ten participants in a 30-minute experiment using the SDI portal to complete three tasks, as well as a post-experiment survey. The findings indicate that most participants experienced difficulties in searching for datasets, which negatively affected their willingness to use the portal in the future.

As algorithm-driven content curation becomes an increasingly common feature of short video platforms, it becomes critical to reveal the role of short video algorithms in cultural sustainability for Ethnic Minorities (EMs). This study conducted one-on-one semi-structured interviews with 15 young Tibetan users, and employ the thematic analysis for exploring the findings. Furthermore, the research results indicate the importance of leveraging short video platforms to enhance the visibility and sustainability of ethnic minority cultures, thereby providing a foundation for developing future algorithmic systems that support global cultural communication.

This study presents the results of an action research project exploring racial justice in public library communities and discusses the implications of its findings for AI bias. Two research questions were addressed: “How is racial justice conceptualized by the LIS practitioner community in the American South?” and “What are potential implications of racial justice discourse in race-related AI bias?” Data was gathered from a strategic action plan created from a survey of southern U.S. public library staff that was analyzed using thematic analysis and interpretative description (ID). Public library staff have defined “racial justice” in terms of the constructs of access, equity, participation/inclusion, representation, accountability, and promotion/support. The results of this characterization of racial justice have implications for remediating race-based bias in artificial intelligence systems by promoting the inclusion of more training and operational data gathered from an emic perspective.