As LLMs gain adoption in high-stakes domains like mental health, domain experts are increasingly consulted to provide input into policies governing their behavior. From an observation of 19 policymaking workshops with 9 experts over 15 weeks, we identified opportunities to better support rapid experimentation, feedback, and iteration for collaborative policy design processes. We present PolicyPad, an interactive system that facilitates the emerging practice of LLM policy prototyping by drawing from established UX prototyping practices, including heuristic evaluation and storyboarding. Using PolicyPad, policy designers can collaborate on drafting a policy in real time while independently testing policy-informed model behavior with usage scenarios. We evaluate PolicyPad through workshops with 8 groups of 22 domain experts in mental health and law, finding that PolicyPad enhanced collaborative dynamics during policy design, enabled tight feedback loops, and led to novel policy contributions. Overall, our work paves expert-informed paths for advancing AI alignment and safety.

AI agents, or bots, serve important roles in online communities. However, they are often designed by outsiders or a few tech-savvy members, leading to bots that may not align with the broader community’s needs. How might communities collectively shape the behavior of community bots? We present Botender, a system that enables communities to collaboratively design LLM-powered bots without coding. With Botender, community members can directly propose, iterate on, and deploy custom bot behaviors tailored to community needs. Botender facilitates testing and iteration on bot behavior through case-based provocations: interaction scenarios generated to spark user reflection and discussion around desirable bot behavior. A validation study found these provocations more useful than standard test cases for revealing improvement opportunities and surfacing disagreements. During a five-day deployment across six Discord servers, Botender supported communities in tailoring bot behavior to their specific needs, showcasing the usefulness of case-based provocations in facilitating collaborative bot design.

The wide adoption of platformized work has generated remarkable advancements in the labor patterns and mobility of modern society. Underpinning such progress, gig workers are exposed to unprecedented challenges and accountabilities: lack of data transparency, social and physical isolation, as well as insufficient infrastructural safeguards. Gig2Gether presents a space designed for workers to engage in an initial experience of voluntarily contributing anecdotal and statistical data to affect policy and build solidarity across platforms by exchanging unifying and diverse experiences. Our 7-day field study with 16 active workers from three distinct platforms and work domains showed existing affordances of data-sharing: facilitating mutual support across platforms, as well as enabling financial reflection and planning. Additionally, workers envisioned future use cases of data-sharing for collectivism (e.g., collaborative examinations of algorithmic speculations) and informing policy (e.g., around safety and pay), which motivated (latent) worker desiderata of additional capabilities and data metrics. Based on these findings, we discuss remaining challenges to address and how data-sharing tools can complement existing structures to maximize worker empowerment and policy impact.

AI tools are increasingly deployed in community contexts. However, datasets used to evaluate AI are typically created by developers and annotators outside a given community, which can yield misleading conclusions about AI performance. How might we empower communities to drive the intentional design and curation of evaluation datasets for AI that impacts them? We investigate this question on Wikipedia, an online community with multiple AI-based content moderation tools deployed. We introduce Wikibench, a system that enables communities to collaboratively curate AI evaluation datasets, while navigating ambiguities and differences in perspective through discussion. A field study on Wikipedia shows that datasets curated using Wikibench can effectively capture community consensus, disagreement, and uncertainty. Furthermore, study participants used Wikibench to shape the overall data curation process, including refining label definitions, determining data inclusion criteria, and authoring data statements. Based on our findings, we propose future directions for systems that support community-driven data curation.

Recent years have seen growing adoption of AI-based decision-support systems (ADS) in homeless services, yet we know little about stakeholder desires and concerns surrounding their use. In this work, we aim to understand impacted stakeholders’ perspectives on a deployed ADS that prioritizes scarce housing resources. We employed AI lifecycle comicboarding, an adapted version of the comicboarding method, to elicit stakeholder feedback and design ideas across various components of an AI system’s design. We elicited feedback from county workers who operate the ADS daily, service providers whose work is directly impacted by the ADS, and unhoused individuals in the region. Our participants shared concerns and design suggestions around the AI system’s overall objective, specific model design choices, dataset selection, and use in deployment. Our findings demonstrate that stakeholders, even without AI knowledge, can provide specific and critical feedback on an AI system’s design and deployment, if empowered to do so.

Machine learning research has long focused on models rather than datasets, and prominent datasets are used for common ML tasks without regard to the breadth, difficulty, and faithfulness of the underlying problems. Neglecting the fundamental importance of data has given rise to inaccuracy, bias, and fragility in real-world applications, and research is hindered by saturation across existing dataset benchmarks. In response, we present DataPerf, a community-led benchmark suite for evaluating ML datasets and data-centric algorithms. We aim to foster innovation in data-centric AI through competition, comparability, and reproducibility. We enable the ML community to iterate on datasets, instead of just architectures, and we provide an open, online platform with multiple rounds of challenges to support this iterative development. The first iteration of DataPerf contains five benchmarks covering a wide spectrum of data-centric techniques, tasks, and modalities in vision, speech, acquisition, debugging, and diffusion prompting, and we support hosting new contributed benchmarks from the community. The benchmarks, online evaluation platform, and baseline implementations are open source, and the MLCommons Association will maintain DataPerf to ensure long-term benefits to academia and industry.

We propose autocomplete for the design and development of virtual breadboard circuits using software prototyping tools. With our system, a user inserts a component into the virtual breadboard, and it automatically provides a user with a list of suggested components. These suggestions complete or ex- tend the electronic functionality of the inserted component to save the user's time and reduce circuit error. To demon- strate the effectiveness of autocomplete, we implemented our system on Fritzing, a popular open source breadboard circuit prototyping software, used by novice makers. Our autocomplete suggestions were implemented based upon schematics from datasheets for standard components, as well as how components are used together from over 4000 circuit projects from the Fritzing community. We report the results of a controlled study with 16 participants, evaluating the effectiveness of autocomplete in the creation of virtual breadboard circuits, and conclude by sharing insights and directions for future research.

We present TilePoP, a new type of pneumatically-actuated interface deployed as floor tiles which dynamically pop up by inflating into large shapes constructing proxy objects for whole-body interactions in Virtual Reality. TilePoP consists of a 2D array of stacked cube-shaped airbags designed with specific folding structures, enabling each airbag to be inflated into a physical proxy and then deflated down back to its original tile shape when not in use. TilePoP is capable of providing haptic feedback for the whole body and can even support human body weight. Thus, it allows new interaction possibilities in VR. Herein, the design and implementation of TilePoP are described in detail along with demonstrations of its applications and the results of a preliminary user evaluation conducted to understand the users' experience with TilePoP.

Eye trackers are found on various electronic devices. In this paper, we propose to exploit the gaze information acquired by an eye tracker for depth estimation. The data collected from the eye tracker in a fixation interval are used to estimate the depth of a gazed object. The proposed method can be used to construct a sparse depth map of an augmented reality space. The resulting depth map can be applied to, for example, controlling the visual information displayed to the viewer. A mathematical model for determining whether two depths in the augmented reality space are statistically distinguishable is also developed. Experimental results show that the proposed method can estimate and distinguish different object depths effectively.

The sensation of being able to feel the shape of an object when grasping it in Virtual Reality (VR) enhances a sense of presence and the ease of object manipulation. Though most prior works focus on force feedback on fingers, the haptic emulation of grasping a 3D shape requires the sensation of touch using the entire hand. Hence, we present Pop-up Prop on Palm (PuPoP), a light-weight pneumatic shape-proxy interface worn on the palm that pops several airbags up with predefined primitive shapes for grasping. When a user's hand encounters a virtual object, an airbag of appropriate shape, ready for grasping, is inflated by way of the use of air pumps; the airbag then deflates when the object is no longer in play. Since PuPoP is a physical prop, it can provide the full sensation of touch to enhance the sense of realism for VR object manipulation. For this paper, we first explored the design and implementation of PuPoP with multiple shape structures. We then conducted two user studies to further understand its applicability. The first study shows that, when in conflict, visual sensation tends to dominate over touch sensation, allowing a prop with a fixed size to represent multiple virtual objects with similar sizes. The second study compares PuPoP with controllers and free-hand manipulation in two VR applications. The results suggest that utilization of dynamically-changing PuPoP, when grasped by users in line with the shapes of virtual objects, enhances enjoyment and realism. We believe that PuPoP is a simple yet effective way to convey haptic shapes in VR.