Generative AI often produces results misaligned with user intentions, for example, resolving ambiguous prompts in unexpected ways. Despite existing approaches to clarify intent, a major challenge remains: understanding and influencing AI’s interpretation of user intent through simple, direct inputs requiring no expertise or rigid procedures. We present ToMigo, representing intent as design concept graphs: nodes represent choices of purpose, content, or style, while edges link them with interpretable explanations. Applied to graphic design, ToMigo infers intent from reference images and text. We derived a schema of node types and edges from pre-study data, informing a multimodal large language model to generate graphs aligning nodes externally with user intent and internally toward a unified design goal. This structure enables users to explore AI reasoning and directly manipulate the design concept. In our user studies, ToMigo’s design concept graphs received high alignment ratings and captured most user intentions well. Users reported greater control and found interactive features—editable graphs, reflective chats, concept-design realignment—useful for evolving and realizing their design ideas.

Generative AI enables educators to create interactive learning content by describing goals in natural language. However, without programming affordances such as traceability, refinement, and debugging, teachers struggle to align simulations with learners’ needs, refine them step by step, or verify that they reflect intended learning concepts. We propose a task-level abstraction approach that structures authoring as a sequence of representations, mirroring how teachers plan lessons and providing checkpoints for specification, inspection, and refinement. We instantiate this approach in SimStep, an authoring environment that scaffolds simulation design with four abstractions, including Concept Graph, Scenario Graph, Learning Goal Graph, and UI Graph, and introduces an inverse correction process to revise hidden model assumptions without requiring code manipulation. A technical evaluation shows that these abstractions preserve fidelity across transformations, while a user study with educators demonstrates their effectiveness in authoring simulations. Our work reframes AI-assisted programming as human–AI co-authoring through structured, domain-aligned abstractions.

As Artificial Intelligence (AI) conversational agents become widespread, people are increasingly using them for health information seeking. The use of off-the-shelf conversational agents for health information seeking could place high metacognitive demands (the need for extensive monitoring and control of one’s own thought process) on individuals, which could compromise their experience of seeking health information. However, currently, the specific demands that arise while using conversational agents for health information seeking, and the strategies people use to cope with those demands, remain unknown. To address these gaps, we conducted a think-aloud study with 15 participants as they sought health information using our off-the-shelf AI conversational agent. We identified the metacognitive demands such systems impose, the strategies people adopt in response, and propose considerations for designing beyond off-the-shelf interfaces to reduce these demands and support better user experiences and affordances in health information seeking.

When exploring data, analysts construct narratives about what the data means by asking questions, generating visualizations, reflecting on patterns, and revising their interpretations as new insights emerge. Yet existing analysis tools treat narrative as an afterthought, breaking the link between reasoning, reflection, and the evolving story from exploration. Consequently, analysts lose the ability to see how their reasoning evolves, making it harder to reflect systematically or build coherent explanations. To address this gap, we propose Narrative Scaffolding (NS), a framework for narrative-driven exploration that positions narrative construction as the primary interface for exploration and reasoning. We implemented this framework in a system that externalizes iterative reasoning through narrative-first entry, semantically aligned view generation, and reflection support via insight provenance and inquiry tracking. In a within-subject study (N = 20), we demonstrated that narrative scaffolding facilitates broader exploration, deeper reflection, and more defensible narratives. An evaluation with visualization literacy experts (N = 6) confirmed that the system produced outputs aligned with narrative intent and facilitated intentional exploration.

Peer Reviewed

Clients often partner with AI experts to develop AI applications tailored to their needs. In these partnerships, careful planning and clear communication are critical, as inaccurate or incomplete specifications can result in misaligned model characteristics, expensive reworks, and potential friction between collaborators. Unfortunately, given the complexity of requirements ranging from functionality, data, and governance, effective guidelines for collaborative specification of requirements in client-AI expert collaborations are missing. In this work, we introduce AINeedsPlanner, a workbook that AI experts and clients can use to facilitate effective interchange of clear specifications. The workbook is based on (1) an interview of 10 completed AI application project teams, which identifies and characterizes steps in AI application planning and (2) a study with 12 AI experts, which defines a taxonomy of AI experts’ information needs and dimensions that affect the information needs. Finally, we demonstrate the workbook’s utility with two case studies in real-world settings.

Research in Responsible AI has developed a range of principles and practices to ensure that machine learning systems are used in a manner that is ethical and aligned with human values. However, a critical yet often neglected aspect of ethical ML is the ethical implications that appear when designing evaluations of ML systems. For instance, teams may have to balance a trade-off between highly informative tests to ensure downstream product safety, with potential fairness harms inherent to the implemented testing procedures. We conceptualize ethics-related concerns in standard ML evaluation techniques. Specifically, we present a utility framework, characterizing the key trade-off in ethical evaluation as balancing information gain against potential ethical harms. The framework is then a tool for characterizing challenges teams face, and systematically disentangling competing considerations that teams seek to balance. Differentiating between different types of issues encountered in evaluation allows us to highlight best practices from analogous domains, such as clinical trials and automotive crash testing, which navigate these issues in ways that can offer inspiration to improve evaluation processes in ML. Our analysis underscores the critical need for development teams to deliberately assess and manage ethical complexities that arise during the evaluation of ML systems, and for the industry to move towards designing institutional policies to support ethical evaluations.

Large language models (LLMs) exhibit dynamic capabilities and appear to comprehend complex and ambiguous natural language prompts. However, calibrating LLM interactions is challenging for interface designers and end-users alike. A central issue is our limited grasp of how human cognitive processes begin with a goal and form intentions for executing actions, a blindspot even in established interaction models such as Norman’s gulfs of execution and evaluation. To address this gap, we theorize how end-users ‘envision’ translating their goals into clear intentions and craft prompts to obtain the desired LLM response. We define a process of Envisioning by highlighting three misalignments on not knowing: (1) what the task should be, (2) how to instruct the LLM to do the task, and (3) what to expect for the LLM’s output in meeting the goal. Finally, we make recommendations to narrow the gulf of envisioning in human-LLM interactions.

ChatGPT has entered classrooms, circumventing typical training and vetting procedures. Unlike other educational technologies, it placed teachers in direct contact with the versatility of generative AI. Consequently, teachers are urgently tasked to assess its capabilities to inform their use of ChatGPT. However, it is unclear what support teachers have and need and whether existing documentation, such as model cards, provides adequate direction for educators in this new paradigm. By interviewing 22 middle- and high-school ELA and Social Studies teachers, we connect the discourse on AI transparency and documentation with educational technology integration, highlighting the information needs of teachers. Our findings reveal that teachers confront significant information gaps, lacking clarity on exploring ChatGPT’s capabilities for bespoke learning tasks and ensuring its fit with the needs of diverse learners. As a solution, we propose a framework for interactive model documentation that empowers teachers to navigate the interplay between pedagogical and technical knowledge.

Large Language Model (LLM) assistants, such as ChatGPT, have emerged as potential alternatives to search methods for helping users navigate complex, feature-rich software. LLMs use vast training data from domain-specific texts, software manuals, and code repositories to mimic human-like interactions, offering tailored assistance, including step-by-step instructions. In this work, we investigated LLM-generated software guidance through a within-subject experiment with 16 participants and follow-up interviews. We compared a baseline LLM assistant with an LLM optimized for particular software contexts, SoftAIBot, which also offered guidelines for constructing appropriate prompts. We assessed task completion, perceived accuracy, relevance, and trust. Surprisingly, although SoftAIBot outperformed the baseline LLM, our results revealed no significant difference in LLM usage and user perceptions with or without prompt guidelines and the integration of domain context. Most users struggled to understand how the prompt’s text related to the LLM’s responses and often followed the LLM’s suggestions verbatim, even if they were incorrect. This resulted in difficulties when using the LLM’s advice for software tasks, leading to low task completion rates. Our detailed analysis also revealed that users remained unaware of inaccuracies in the LLM’s responses, indicating a gap between their lack of software expertise and their ability to evaluate the LLM’s assistance. With the growing push for designing domain-specific LLM assistants, we emphasize the importance of incorporating explainable, context-aware cues into LLMs to help users understand prompt-based interactions, identify biases, and maximize the utility of LLM assistants.

Visualization for machine learning (VIS4ML) research aims to help experts apply their prior knowledge to develop, understand, and improve the performance of machine learning models. In conceiving VIS4ML systems, researchers characterize the nature of human knowledge to support human-in-the-loop tasks, design interactive visualizations to make ML components interpretable and elicit knowledge, and evaluate the effectiveness of human-model interchange. We survey recent VIS4ML papers to assess the generalizability of research contributions and claims in enabling human-in-the-loop ML. Our results show potential gaps between the current scope of VIS4ML research and aspirations for its use in practice. We find that while papers motivate that VIS4ML systems are applicable beyond the specific conditions studied, conclusions are often overfitted to non-representative scenarios, are based on interactions with a small set of ML experts and well-understood datasets, fail to acknowledge crucial dependencies, and hinge on decisions that lack justification. We discuss approaches to close the gap between aspirations and research claims and suggest documentation practices to report generality constraints that better acknowledge the exploratory nature of VIS4ML research.

Despite the widespread use of artificial intelligence (AI), designing user experiences (UX) for AI-powered systems remains challenging. UX designers face hurdles understanding AI technologies, such as pre-trained language models, as design materials. This limits their ability to ideate and make decisions about whether, where, and how to use AI. To address this problem, we bridge the literature on AI design and AI transparency to explore whether and how frameworks for transparent model reporting can support design ideation with pre-trained models. By interviewing 23 UX practitioners, we find that practitioners frequently work with pre-trained models, but lack support for UX-led ideation. Through a scenario-based design task, we identify common goals that designers seek model understanding for and pinpoint their model transparency information needs. Our study highlights the pivotal role that UX designers can play in Responsible AI and calls for supporting their understanding of AI limitations through model transparency and interrogation.

To design with AI models, user experience (UX) designers must assess the fit between the model and user needs. Based on user research, they need to contextualize the model’s behavior and potential failures within their product-specific data instances and user scenarios. However, our formative interviews with ten UX professionals revealed that such a proactive discovery of model limitations is challenging and time-intensive. Furthermore, designers often lack technical knowledge of AI and accessible exploration tools, which challenges their understanding of model capabilities and limitations. In this work, we introduced a failure-driven design approach to AI, a workflow that encourages designers to explore model behavior and failure patterns early in the design process. The implementation of fAIlureNotes, a designer-centered failure exploration and analysis tool, supports designers in evaluating models and identifying failures across diverse user groups and scenarios. Our evaluation with UX practitioners shows that fAIlureNotes outperforms today’s interactive model cards in assessing context-specific model performance.

Creative coding tasks are often exploratory in nature. When producing digital artwork, artists usually begin with a high-level semantic construct such as a “stained glass filter” and programmatically implement it by varying code parameters such as shape, color, lines, and opacity to produce visually appealing results. Based on interviews with artists, it can be effortful to translate semantic constructs to program syntax, and current programming tools don’t lend well to rapid creative exploration. To address these challenges, we introduce Spellburst, a large language model (LLM) powered creative-coding environment. Spellburst provides (1) a node-based interface that allows artists to create generative art and explore variations through branching and merging operations, (2) expressive prompt-based interactions to engage in semantic programming, and (3) dynamic prompt-driven interfaces and direct code editing to seamlessly switch between semantic and syntactic exploration. Our evaluation with artists demonstrates Spellburst’s potential to enhance creative coding practices and inform the design of computational creativity tools that bridge semantic and syntactic spaces.

Scientific knowledge develops through cumulative discoveries that build on, contradict, contextualize, or correct prior findings. Scientists and journalists often communicate these incremental findings to lay people through visualizations and text (e.g., the positive and negative effects of caffeine intake). Consequently, readers need to integrate diverse and contrasting evidence from multiple sources to form opinions or make decisions. However, the underlying mechanism for synthesizing information from multiple visualizations remains under-explored. To address this knowledge gap, we conducted a series of four experiments ( N=1166) in which participants synthesized empirical evidence from a pair of line charts presented sequentially. In Experiment 1, we administered a baseline condition with charts depicting no specific context where participants held no strong belief. To test for the generalizability, we introduced real-world scenarios to our visualizations in Experiment 2 and added accompanying text descriptions similar to online news articles or blog posts in Experiment 3. In all three experiments, we varied the relative direction and magnitude of line slopes within the chart pairs. We found that participants tended to weigh the positive slope more when the two charts depicted relationships in the opposite direction (e.g., one positive slope and one negative slope). Participants tended to weigh the less steep slope more when the two charts depicted relationships in the same direction (e.g., both positive). Through these experiments, we characterize participants' synthesis behaviors depending on the relationship between the information they viewed, contribute to theories describing underlying cognitive mechanisms in information synthesis, and describe design implications for data storytelling.

Conventional tools for visual analytics emphasize a linear production workflow and lack organic “work surfaces.” A better surface would simultaneously support collaborative visualization construction, data and design exploration, and reasoning. To facilitate data-driven design within existing design tools such as card sorting, we introduce Composites, a tangible, augmented reality interface for constructing visualizations on large surfaces. In response to the placement of physical sticky-notes, Composites projects visualizations and data onto large surfaces. Our spatial grammar allows the designer to flexibly construct visualizations through the use of the notes. Similar to affinity-diagramming, the designer can “connect” the physical notes to data, operations, and visualizations which can then be re-arranged based on creative needs. We develop mechanisms (sticky interactions, visual hinting, etc.) to provide guiding feedback to the end-user. By leveraging low-cost technology, Composites extends a working surface to support a broad range of workflows without limiting creative design thinking.

In conventional software development, user experience (UX) designers and engineers collaborate through separation of concerns (SoC): designers create human interface specifications, and engineers build to those specifications. However, we argue that Human-AI systems thwart SoC because human needs must shape the design of the AI interface, the underlying AI sub-components, and training data. How do designers and engineers currently collaborate on AI and UX design? To find out, we interviewed 21 industry professionals (UX researchers, AI engineers, data scientists, and managers) across 14 organizations about their collaborative work practices and associated challenges. We find that hidden information encapsulated by SoC challenges collaboration across design and engineering concerns. Practitioners describe inventing ad-hoc representations exposing low-level design and implementation details (which we characterize as leaky abstractions) to “puncture” SoC and share information across expertise boundaries. We identify how leaky abstractions are employed to collaborate at the AI-UX boundary and formalize a process of creating and using leaky abstractions.

When prototyping AI experiences (AIX), interface designers seek effective ways to support end-user tasks through AI capabilities. However, AI poses challenges to design due to its dynamic behavior in response to training data, end-user data, and feedback. Designers must consider AI's uncertainties and offer adaptations such as explainability, error recovery, and automation vs. human task control. Unfortunately, current prototyping tools assume a black-box view of AI, forcing designers to work with separate tools to explore machine learning models, understand model performance, and align interface choices with model behavior. This introduces friction to rapid and iterative prototyping. We propose Model-Informed Prototyping (MIP), a workflow for AIX design that combines model exploration with UI prototyping tasks. Our system, ProtoAI, allows designers to directly incorporate model outputs into interface designs, evaluate design choices across different inputs, and iteratively revise designs by analyzing model breakdowns.

Video is an effective medium for knowledge communication and learning. Yet active viewing and note-taking from videos remain a challenge. Specifically, during note-taking, viewers find it difficult to extract essential information such as representation, composition, motion, and interactions of graphical objects and narration. Current approaches rely on creating static screenshots, manual clipping, manual annotation and transcription. This is often done by repeatedly pausing and rewinding the video, thus disrupting the viewing experience. We propose VideoSticker, a tool designed to support visual note-taking by extracting expressive content and narratives from videos as ‘object stickers.’ VideoSticker implements automated object detection and tracking, linking objects to the transcript, and supporting rapid extraction of stickers across space, time, and events of interest. VideoSticker’s two-pass approach allows viewers to capture high-level information uninterrupted and later extract specific details. We demonstrate the usability of VideoSticker for a variety of videos and note-taking needs.

Various tools and practices have been developed to support practitioners in identifying, assessing, and mitigating fairness-related harms caused by AI systems. However, prior research has highlighted gaps between the intended design of these tools and practices and their use within particular contexts, including gaps caused by the role that organizational factors play in shaping fairness work. In this paper, we investigate these gaps for one such practice: disaggregated evaluations of AI systems, intended to uncover performance disparities between demographic groups. By conducting semi-structured interviews and structured workshops with thirty-three AI practitioners from ten teams at three technology companies, we identify practitioners' processes, challenges, and needs for support when designing disaggregated evaluations. We find that practitioners face challenges when choosing performance metrics, identifying the most relevant direct stakeholders and demographic groups on which to focus, and collecting datasets with which to conduct disaggregated evaluations. More generally, we identify impacts on fairness work stemming from a lack of engagement with direct stakeholders or domain experts, business imperatives that prioritize customers over marginalized groups, and the drive to deploy AI systems at scale.

Thinking of technology as a design material is appealing. It encourages designers to explore the material’s properties to understand its capabilities and limitations—a prerequisite to generative design thinking. However, as a material, AI resists this approach because its properties only emerge as part of the user experience design. Therefore, designers and AI engineers must collaborate in new ways to create both the material and its application experience. We investigate the co-creation process through a design study with 10 pairs of designers and engineers. We find that design ‘probes’ with user data are a useful tool in defining AI materials. Through data probes, designers construct designerly representations of the envisioned AI experience (AIX) to identify desirable AI characteristics. Data probes facilitate divergent design thinking, material testing, and design validation. Based on our findings, we propose a process model for co-creating AIX and offer design considerations for incorporating data probes in AIX design tools.

When prototyping AI experiences (AIX), interface designers seek useful and usable ways to support end-user tasks through AI capabilities. However, AI poses challenges to design due to its dynamic behavior in response to training data, end-user data, and feedback. Designers must consider AI’s uncertainties and offer adaptations such as explainability, error recovery, and automation vs. human task control. Unfortunately, current prototyping tools assume a black-box view of AI, forcing designers to work with separate tools to explore machine learning models, understand model performance, and align interface choices with model behavior. This introduces friction to rapid and iterative prototyping. We propose Model-Informed Prototyping (MIP), a workflow for AIX design that combines model exploration with UI prototyping tasks. Our system, ProtoAI, allows designers to directly incorporate model outputs into interface designs, evaluate design choices across different inputs, and iteratively revise designs by analyzing model breakdowns. We demonstrate how ProtoAI can readily operationalize human-AI design guidelines. Our user study finds that designers can effectively engage in MIP to create and evaluate AI-powered interfaces during AIX design.

People often use text in their drawings to communicate their ideas. For visually impaired people, adding textual information to tactile graphics is challenging. Labeling in braille is a laborious process and clutters the drawings. Audio labels provide an alternative way to add text. However, digital drawing tools for visually impaired people have not examined the use of audio for creating labels. We conducted a study comprising three tasks with 11 visually impaired adults. Our goal was to understand how participants explored and created labeled tactile graphics (both braille and audio), and their interaction preferences. We find that audio labels were quicker to use and easier to create. However, braille labels enabled flexible exploration strategies. We also find that participants preferred multimodal interaction commands, and report hand postures and movements observed during the drawing process for designing recognizable interactions. Based on our findings, we derive design implications for digital drawing tools.

Learning from text is a constructive activity in which sentence-level information is combined by the reader to build coherent mental models. With increasingly complex texts, forming a mental model becomes challenging due to a lack of background knowledge, and limits in working memory and attention. To address this, we are taught knowledge externalization strategies such as active reading and diagramming. Unfortunately, paper-and-pencil approaches may not always be appropriate, and software solutions create friction through difficult input modalities, limited workflow support, and barriers between reading and diagramming. For all but the simplest text, building coherent diagrams can be tedious and difficult. We propose Active Diagramming, an approach extending familiar active reading strategies to the task of diagram construction. Our prototype, texSketch, combines pen-and-ink interactions with natural language processing to reduce the cost of producing diagrams while maintaining the cognitive effort necessary for comprehension. Our user study finds that readers can effectively create diagrams without disrupting reading.

Despite the availability of software to support Affinity Diagramming (AD), practitioners still largely favor physical sticky-notes. Physical notes are easy to set-up, can be moved around in space and offer flexibility when clustering un-structured data. However, when working with mixed data sources such as surveys, designers often trade off the physicality of notes for analytical power. We propose AffinityLens, a mobile-based augmented reality (AR) application for Data-Assisted Affinity Diagramming (DAAD). Our application provides just-in-time quantitative insights overlaid on physical notes. Affinity Lens uses several different types of AR overlays (called lenses) to help users find specific notes, cluster information, and summarize insights from clusters. Through a formative study of AD users, we developed design principles for data-assisted AD and an initial collection of lenses. Based on our prototype, we find that Affinity Lens supports easy switching between qualitative and quantitative 'views' of data, without surrendering the lightweight benefits of existing AD practice.

Details-on-demand is a crucial feature in the visual information-seeking process but is often only implemented in highly constrained settings. The most common solution, hover queries (i.e., tooltips), are fast and expressive but are usually limited to single mark (e.g., a bar in a bar chart). 'Queries' to retrieve details for more complex sets of objects (e.g., comparisons between pairs of elements, averages across multiple items, trend lines, etc.) are difficult for end-users to invoke explicitly. Further, the output of these queries require complex annotations and overlays which need to be displayed and dismissed on demand to avoid clutter. In this work we introduce SmartCues, a library to support details-on-demand through dynamically computed overlays. For end-users, SmartCues provides multitouch interactions to construct complex queries for a variety of details. For designers, SmartCues offers an interaction library that can be used out-of-the-box, and can be extended for new charts and detail types. We demonstrate how SmartCues can be implemented across a wide array of visualization types and, through a lab study, show that end users can effectively use SmartCues.

Performance animation is an expressive method for animating characters through human performance. However, character motion is only one part of creating animated stories. The typical workflow also involves writing a script, coordinating actors, and editing recorded performances. In most cases, these steps are done in isolation with separate tools, which introduces friction and hinders iteration. We propose TakeToons, a script-driven approach that allows authors to annotate standard scripts with relevant animation events like character actions, camera positions, and scene backgrounds. We compile this script into a story model that persists throughout the production process and provides a consistent structure for organizing and assembling recorded performances and propagating script or timing edits to existing recordings. TakeToons enables writing, performing and editing to happen in an integrated and interleaved manner that streamlines production and facilitates iteration. Informal feedback from professional animators suggests that our approach can benefit many existing workflows supporting individual authors and production teams with many different contributors.

Studies on technology adoption typically assume that a user's perception of usability and usefulness of technology are central to its adoption. Specifically, in the case of accessibility and assistive technology, research has traditionally focused on the artifact rather than the individual, arguing that individual technologies fail or succeed based on their usability and fit for their users. Using a mixed-methods field study of smartphone adoption by 81 people with visual impairments in Bangalore, India, we argue that these positions are dated in the case of accessibility where a non-homogeneous population must adapt to technologies built for sighted people. We found that many users switch to smartphones despite their awareness of significant usability challenges with smartphones. We propose a nuanced understanding of perceived usefulness and actual usage based on need-related social and economic functions, which is an important step toward rethinking technology adoption for people with disabilities.