While multimodal generative AI can support creative activities, users often struggle to prompt models to achieve desired aesthetic, acoustic, or stylistic characteristics. Besides, existing generative models are predominantly driven by text-based prompts regardless of their output modality, i.e., using text-based prompts for creating images, videos, and sounds, which often leads to high prompt uncertainty. In response, recent multimodal AI systems introduce interaction strategies to better align model interpretations with users’ creative intent, but the growing variety of strategies makes it hard to judge what works for a given use case. We address this gap with a systematic literature review (n=71) that categorizes prompt-uncertainty-reduction strategies into six types: Guiding Prompt Construction, System Refining of the Prompt, Direct Manipulations of Output Elements, Explaining Reasoning about Prompt Interpretation, Displaying Multiple Outputs, and Controlling Modifier Contribution. For each type, we summarize mechanisms, benefits, and challenges, enabling more efficient navigation of the prompt-support design space.

LLM-powered conversational recommender systems (CRSs) empower users to personalize recommendation services, giving them control over how recommendations are represented and explained. Explanations of why particular options are recommended are shown to be especially valuable when users explore unfamiliar items. While prior work on personalized explanations in recommender systems has focused predominantly on explanation style, there is still little understanding of what types of information explanations should contain to meaningfully support users’ exploration. To allow CRSs to better align the explanations with users’ informational needs, in this paper, we present the information composition for recommendation explanations. Informed by an exploratory interview-based user study, we propose four key informational dimensions: Essence, Experience, Exchange, and Entwinement. We then report a comparative evaluation showing that explanations structured along these dimensions are perceived as more supportive of engagement-related outcomes than baseline LLM-generated explanations. We conclude by outlining design implications for LLM-powered CRSs.

In subjective decision-making, where decisions are based on contextual interpretation, Large Language Models (LLMs) can be integrated to present users with additional rationales to consider. The diversity of these rationales is mediated by the ability to consider the perspectives of different social actors. However, it remains unclear whether and how models differ in the distribution of perspectives they provide. We compare the perspectives taken by humans and different LLMs when assessing subtle sexism scenarios. We show that these perspectives can be classified within a finite set (perpetrator, victim, decision-maker), consistently present in argumentations produced by humans and LLMs, but in different distributions and combinations, demonstrating differences and similarities with human responses, and between models. We argue for the need to systematically evaluate LLMs' perspective-taking to identify the most suitable models for a given decision-making task. We discuss the implications for model evaluation.

Agentic AI is being heralded as the next step in the development of AI systems. Agentics, complex ensembles of different machine learning, data processing, and generative AI models, can provide new autonomous and proactive decision-making capabilities to organizations, participate in complex workflows, and, when needed, seek guidance from and provide insights to human users in natural languages. Collectively, we wish to explore how and why to design agentic systems to be situated within specific social and organizational contexts, the value of social theory and perspectives to this work, and the potential of this move to address critical issues with AI. Given the focus on social and organization context as essential to agentic design, we see this work as directly related to the ACM COMPASS 2025 theme “computing in place”. We seek to bring together scholars from the diversity of disciplines within ACM to develop research agendas, projects, and joint teaching initiatives that support the development of social agentic design and analysis.

To provide long-term personalized assistance to users, AI agents must have effective long-term memory (LTM). However, there is little understanding of users’ perceptions, practices, and challenges with LTM in agents. We interviewed 21 users of agents such as ChatGPT and Claude to understand people’s everyday experiences with agent LTM. Our findings shed light on the flow of memory in agents as a three-stage process consisting of (1) information intake, (2) storage and management, and (3) retrieval and interpretation. Users’ perceptions of agent LTM are mainly influenced by Stage 3, and thus users’ interactions with agent LTM are mainly attempts at influencing and understanding how the agent retrieves and interprets information from memory. Therefore, we recommend that technological approaches to user interaction with agent LTM focus at least as much on memory retrieval and interpretation as they do on memory intake, storage, and management.

Although engagement in self-directed wellness exercises typically declines over time, integrating social support such as coaching can sustain it. However, traditional forms of support are often inaccessible due to the high costs and complex coordination. Large Language Models (LLMs) show promise in providing human-like dialogues that could emulate social support. Yet, in-depth, in situ investigations of LLMs to support behavior change remain underexplored. We conducted two randomized experiments to assess the impact of LLM agents on user engagement with mindfulness exercises. First, a single-session study, involved 502 crowdworkers; second, a three-week study, included 54 participants. We explored two types of LLM agents: one providing information and another facilitating self-reflection. Both agents enhanced users' intentions to practice mindfulness. However, only the information-providing LLM agent, featuring a friendly persona, significantly improved engagement with the exercises. Our findings suggest that specific LLM agents may bridge the social support gap in digital health interventions.

As generative AI tools like ChatGPT become integral to everyday writing, critical questions arise about how to preserve writers' sense of agency and ownership when using these tools. Yet, a systematic understanding of how AI assistance affects different aspects of the writing process-and how this shapes writers' agency-remains underexplored. To address this gap, we conducted a systematic review of 109 HCI papers using the PRISMA approach. From this literature, we identify four overarching design strategies for AI writing support- structured guidance, guided exploration, active co-writing , and critical feedback -mapped across the four key cognitive processes in writing: planning, translating, reviewing , and monitoring . We complement this analysis with interviews of 15 writers across diverse domains. Our findings reveal that writers' desired levels of AI intervention vary across the writing process: content-focused writers (e.g., academics) prioritize ownership during planning, while form-focused writers (e.g., creatives) value control over translating and reviewing. Writers' preferences are also shaped by contextual goals, values, and notions of originality and authorship. By examining when ownership matters, what writers want to own, and how AI interactions shape agency, we surface both alignment and gaps between research and user needs. Our findings offer actionable design guidance for developing human-centered writing tools for co-writing with AI, on human terms.

Traditional interventions for academic procrastination often fail to capture the nuanced, individual-specific factors that underlie them. Large language models (LLMs) hold immense potential for addressing this gap by permitting open-ended inputs, including the ability to customize interventions to individuals' unique needs. However, user expectations and potential limitations of LLMs in this context remain underexplored. To address this, we conducted interviews and focus group discussions with 15 university students and 6 experts, during which a technology probe for generating personalized advice for managing procrastination was presented. Our results highlight the necessity for LLMs to provide structured, deadline-oriented steps and enhanced user support mechanisms. Additionally, our results surface the need for an adaptive approach to questioning based on factors like busyness. These findings offer crucial design implications for the development of LLM-based tools for managing procrastination while cautioning the use of LLMs for therapeutic guidance.

Exploring alternative ideas by rewriting text is integral to the writing process. State-of-the-art Large Language Models (LLMs) can simplify writing variation generation. However, current interfaces pose challenges for simultaneous consideration of multiple variations: creating new variations without overwriting text can be difficult, and pasting them sequentially can clutter documents, increasing workload and disrupting writers’ flow. To tackle this, we present ABScribe, an interface that supports rapid, yet visually structured, exploration and organization of writing variations in human-AI co-writing tasks. With ABScribe, users can swiftly modify variations using LLM prompts, which are auto-converted into reusable buttons. Variations are stored adjacently within text fields for rapid in-place comparisons using mouse-over interactions on a popup toolbar. Our user study with 12 writers shows that ABScribe significantly reduces task workload (d = 1.20, p < 0.001), enhances user perceptions of the revision process (d = 2.41, p < 0.001) compared to a popular baseline workflow, and provides insights into how writers explore variations using LLMs.

As generative AI (GenAI) rapidly evolves, human-AI collaboration emerges as a prevalent new working style. However, within this collaborative pipeline, multiple stakeholders are involved besides the user and the system itself, raising controversy around ownership over co-creations. In this paper, we explored everyday users’ sense of ownership toward human-AI co-creation, aiming to provide insights for practitioners on future GenAI design to enhance user experience. We identify three primary factors associated with people’s perception of psychological ownership towards human-AI co-creation and systematically analyze individuals’ approaches to assessing these factors. The findings serve to inform strategies for facilitating an appropriate sense of ownership for productive and safe usage of GenAI tools.

Research continuously shows that, despite the wide range of skills developed for Intelligent Personal Assistants (IPAs), users tend to engage with only a small number of them. One reason for this discrepancy is the issue of skill discoverability, which is commonly addressed through conversational recommendations. Current recommendation strategies, however, are limited due to information asymmetry, lack of interactivity, and an underdeveloped understanding of appropriate grouping of available skills. In this paper, we explore opportunities for interactive faceted skill recommendations using voice interfaces. Through an open card sort user study and semi-structured interviews, we identify and describe five facets driving users’ natural grouping of IPA skills (Thematic, Procedural, Cross-system, Environmental, and Recipient), and demonstrate the need for simultaneous support of these facets. We then discuss the implications of these findings for advancing the discoverability of IPA skills through the design of interactive conversational recommendations.

Human-AI collaborative decision-making can achieve better outcomes than either party individually. The success of this collaboration can depend on whether the human decision-maker perceives the AI contribution as beneficial to the decision-making process. Beneficial AI explanations are often described as relevant, convincing, and trustworthy. Yet, we know little about the characteristics of explanations that result in these perceptions. Focusing on collaborative subjective decision-making, using the context of subtle sexism, where explanations can surface new interpretations, we conducted a user study (N=20) to explore the structural and content characteristics that affect perceptions of human and AI-generated verbal (text and audio) explanations. We find four groups of characteristics ( Tone, Grammatical Elements, Argumentative Sophistication and Relation to User ), and that the effect of these characteristics on the perception of explanations for subtle sexism depends on the perceived author. Thus, we also identify which explanation characteristics participants use to identify the author of an explanation. Demonstrating the relationship between these characteristics and explanation perceptions, we present a categorized set of characteristics that system builders can leverage to produce the appropriate perception of an explanation for various sensitive contexts. We also highlight human perception biases and associated issues resulting from these perceptions.

Smartphone overuse around family and friends has been shown to be increasing over the past years and often leads to limited one-to-one interaction between co-located individuals. Smartphone-based virtual agents have been shown to be effective for behavior intervention and mediation, such as promoting physical activity. Little is known about leveraging smartphone-based agents to play a role in communication and facilitate conversation between co-located individuals. In this paper, we explore strengthening conversations between co-located couples by introducing a smartphone-based agent that acts as a conversation facilitator between them. We contrast the results with a text-based alternative. Our findings suggest that virtual agents serve as a valuable social entity mediating support in couples' communication and relationship dynamics. Through this, we suggest design considerations for this context that leverage the unique qualities of virtual agents.

Personalized chatbot-based teaching assistants can be crucial in addressing increasing classroom sizes, especially where direct teacher presence is limited. Large language models (LLMs) offer a promising avenue, with increasing research exploring their educational utility. However, the challenge lies not only in establishing the efficacy of LLMs but also in discerning the nuances of interaction between learners and these models, which impact learners' engagement and results. We conducted a formative study in an undergraduate computer science classroom (N=145) and a controlled experiment on Prolific (N=356) to explore the impact of four pedagogically informed guidance strategies on the learners' performance, confidence and trust in LLMs. Direct LLM answers marginally improved performance, while refining student solutions fostered trust. Structured guidance reduced random queries as well as instances of students copy-pasting assignment questions to the LLM. Our work highlights the role that teachers can play in shaping LLM-supported learning environments.

Assigning lifelike qualities to robotic agents (Anthropomorphism) is associated with complex affective interpretations of their behavior. These anthropomorphized perceptions are traditionally elicited through robots' designs. Yet, aerial robots (or drones) present a special case due to their – traditionally – non-anthropomorphic design, and prior research shows conflicting evidence on their perception as either person-like, animal-like, or machine-like. In this work, we explore how people perceive drones in a cross-dimensional space between these three dimensions by varying the affective state presented on the drone. To capture these perceptions, we developed a novel measurement instrumentAnZoMa. We describe the design, use, and deployment of the instrument in an online study (N=98). The study results suggest that different drone emotions triggered people to attribute various characteristics to the drone (e.g., interaction metaphors, traits, and features) and variations in acceptability of drone affective states. These results demonstrate the interdependencies between affective perceptions and anthropomorphism of drones. We conclude by discussing the necessity to integrate cross-dimensional perception of anthropomorphism in human-drone interaction and affective computing. This work contributes a novel tool to measure the dimensions and gravity of anthropomorphism and insights into interdependencies between different affective states displayed on drones and their anthropomorphized perception.

The use of algorithms for decision-making in higher education is steadily growing, promising cost-savings to institutions and personalized service for students but also raising ethical challenges around surveillance, fairness, and interpretation of data. To address the lack of systematic understanding of how these algorithms are currently designed, we reviewed an extensive corpus of papers proposing algorithms for decision-making in higher education. We categorized them based on input data, computational method, and target outcome, and then investigated the interrelations of these factors with the application of human-centered lenses: theoretical, participatory, or speculative design. We found that the models are trending towards deep learning, and increased use of student personal data and protected attributes, with the target scope expanding towards automated decisions. However, despite the associated decrease in interpretability and explainability, current development predominantly fails to incorporate human-centered lenses. We discuss the challenges with these trends and advocate for a human-centered approach.

There is a growing interest in extending crowdwork beyond traditional desktop-centric design to include mobile devices (e.g., smartphones). However, mobilizing crowdwork remains significantly tedious due to a lack of understanding about the mobile usability requirements of human intelligence tasks (HITs). We present a taxonomy of characteristics that defines the mobile usability of HITs for smartphone devices. The taxonomy is developed based on findings from a study of three consecutive steps. In Step 1, we establish an initial design of our taxonomy through a targeted literature analysis. In Step 2, we verify and extend the taxonomy through an online survey with Amazon Mechanical Turk crowdworkers. Finally, in Step 3 we demonstrate the taxonomy’s utility by applying it to analyze the mobile usability of a dataset of scraped HITs. In this paper, we present the iterative development of the taxonomy, highlighting the observed practices and preferences around mobile crowdwork. We conclude with the implications of our taxonomy for accessibly and ethically mobilizing crowdwork not only within the context of smartphone devices, but beyond them.

Drones are rapidly populating human spaces, yet little is known about how these flying robots are perceived and understood by humans. Recent works suggested that their acceptance is predicated upon their sociability. This paper explores the use of facial expressions to represent emotions on social drones. We leveraged design practices from ground robotics and created a set of rendered robotic faces that convey basic emotions. We evaluated individuals’ response to these emotional facial expressions on drones in two empirical studies (N = 98, N = 98). Our results demonstrate that individuals accurately recognize five drone emotional expressions, as well as make sense of intensities within emotion categories. We describe how participants were emotionally affected by the drone, showed empathy towards it, and created narratives to interpret its emotions. As a consequence, we formulate design recommendations for social drones and discuss methodological insights on the use of static versus dynamic stimuli in affective robotics studies.

New technology is moving towards intuitive and natural interaction techniques that are increasingly embedded in human space (e.g., home and office environment) and aims to support multiple users, yet their interfaces do not cover it to the full. Imagine that you have a multi-user device, should it act differently in different situations, people, and group settings? Current Multi-User Interfaces address each of the users as an individual that works independently from others, and there is a lack of understanding of the mechanisms that impact shared usage of these products. Thus we have linked environmental (external) and user-centered (internal) factors to the way users interact with multi-user devices. We analyzed 124 papers that involve multi-user interfaces and created a classification model out of 8 factors. Both the model and factors were validated by a large-scale online study. Our model defines the factors affecting multi-user usage with a single device and leads to a decision on the most important ones in different situations. This paper is the first to identify these factors and to create a set of practical guidelines for designing Multi-User Interfaces.

Participants in video meetings have long struggled with asymmetrical attention levels, especially when participants are distributed unevenly. While technological advances offer exciting opportunities to augment remote users' attention, the phenomenological complexity of attention means that to design attention-fostering features we must first understand what aspects of it are functionally meaningful to support. In this paper, we present a functional classification of observable attention for video meetings. The classification was informed by two studies on sense-making and selectiveness of attention in work meetings. It includes categories of attention accessible for technological support, their functions in a meeting process, and meeting-related activities that correspond to these functions. This classification serves as a multi-level representation of attention and informs the design of features aiming to support remote participants' attention in video meetings.

This paper presents a qualitative multi-phase study seeking to identify patterns in users' anthropomorphized perceptions of conversational agents. Through a comparative analysis of behavioral perceptions and visual conceptions of three agents - Alexa, Google Assistant, and Siri - we first show that the perceptions of an agent's character are structured according to five categories: approachability, sentiment toward a user, professionalism, intelligence, and individuality. We then explore visualizations of the agents' appearance and discuss the specifics assigned to each agent. Finally, we analyze associative explanations for these perceptions. We demonstrate that the anthropomorphized behavioral and visual perceptions of agents yield structural consistency and discuss how these perceptions are linked with each other and system features.

As multi-touch displays grow in size and shrink in price, they are more commonly used as gaming devices. When co-located users play games on a single, large display, establishing and maintaining their physical and digital territories poses a social challenge to their interaction. To gain insight into the mechanisms of establishing and maintaining users' physical and digital territories, we analyze territorial interactions in cooperative and competitive multiplayer gameplay. Participants reported weighing each game interaction based on perceived intent to determine how socially acceptable they deemed each behaviour. In light of our observations, we contribute and discuss implications for the design of multi-user, large display, co-located, touchscreen games that consider display properties, digital and physical space, permeability of boundaries, and asymmetry of play to create interactions between players.

In the age of the internet, the ability to fully control online information about oneself is no longer possible; once information is online, it is easily disseminated and quickly becomes part of the internet archive. In this paper, we explore data from a set of in-person questionnaires and contextual interviews with both searchers and searched individuals and an additional data set from an online survey of internet users to identify strategies for vetting, managing, and interpreting content online. We see a desire for agency, a contextualization of information based on social context and temporal context, and a desire to constrain the exploration of online content to a relevant subset of information when searching others, i.e. a desire for selectiveness. We offer an evocative design sketch to highlight these issues and foster a discussion of emerging socio-cultural norms around online search behavior.