Haiyi Zhu’s research while affiliated with Carnegie Mellon University and other places

The proliferating adoption of platform-based gig work increasingly raises concerns for worker conditions. Past studies documented how platforms leveraged design to exploit labor, withheld information to generate power asymmetries, and left workers alone to manage logistical overheads as well as social isolation. However, researchers also called attention to the potential of helping workers overcome such costs via worker-led datasharing, which can enable collective actions and mutual aid among workers, while offering advocates, lawmakers and regulatory bodies insights for improving work conditions. To understand stakeholders' desiderata for a data-sharing system (i.e. functionality and policy initiatives that it can serve), we interviewed 11 policy domain experts in the U.S. and conducted co-design workshops with 14 active gig workers across four domains. Our results outline policymakers' prioritized initiatives, information needs, and (mis)alignments with workers' concerns and desires around data collectives. We offer design recommendations for data-sharing systems that support worker needs while bringing us closer to legislation that promote more thriving and equitable gig work futures.

As public sector agencies rapidly introduce new AI tools in high-stakes domains like social services, it becomes critical to understand how decisions to adopt these tools are made in practice. We borrow from the anthropological practice to "study up" those in positions of power, and reorient our study of public sector AI around those who have the power and responsibility to make decisions about the role that AI tools will play in their agency. Through semi-structured interviews and design activities with 16 agency decision-makers, we examine how decisions about AI design and adoption are influenced by their interactions with and assumptions about other actors within these agencies (e.g., frontline workers and agency leaders), as well as those above (legal systems and contracted companies), and below (impacted communities). By centering these networks of power relations, our findings shed light on how infrastructural, legal, and social factors create barriers and disincentives to the involvement of a broader range of stakeholders in decisions about AI design and adoption. Agency decision-makers desired more practical support for stakeholder involvement around public sector AI to help overcome the knowledge and power differentials they perceived between them and other stakeholders (e.g., frontline workers and impacted community members). Building on these findings, we discuss implications for future research and policy around actualizing participatory AI approaches in public sector contexts.

Public sector agencies aim to innovate not just for efficiency but also to enhance equity. Despite the growing adoption of data-driven decision-making systems in the public sector, efforts to integrate equity as a primary goal often fall short. This typically arises from inadequate early-stage involvement of underserved stakeholders and prevalent misunderstandings concerning the authentic meaning of equity from these stakeholders' perspectives. Our research seeks to address this gap by actively involving undersevered stakeholders in the process of envisioning the integration of equity within public sector data-driven decisions, particularly in the context of a building department in a Northeastern mid-sized U.S. city. Applying a speed dating method with storyboards, we explore diverse equity-centric futures within the realm of local business development, a domain where small businesses, particularly women-and minority-owned businesses, historically confront inequitable distribution of public services. We explored three essential aspects of equity: monitoring equity, resource allocation prioritization, as well as information and equity. Our findings illuminate the complexities of integrating equity into data-driven decisions, offering nuanced insights about the needs of stakeholders. We found that attempts to monitor and incorporate equity goals into public sector decision-making can unexpectedly backfire, inadvertently sparking community apprehension and potentially exacerbating existing inequities. Small business owners, including those identifying as women-and minority-owned, advocated against the use of demographic-based data in equity-focused data-driven decision-making in the public sector, instead emphasizing factors such as community needs, application complexity, and uncertainties inherent in small businesses. Drawing from these insights, we propose design implications to assist designers of public sector data-driven decision-making systems to better accommodate equity considerations.

A growing body of literature has focused on understanding and addressing workplace AI design failures. However, past work has largely overlooked the role of occupational devaluation in shaping the dynamics of AI development and deployment. In this paper, we examine the case of feminized labor: a class of devalued occupations historically misnomered as ``women's work,'' such as social work, K-12 teaching, and home healthcare. Drawing on literature on AI deployments in feminized labor contexts, we conceptualize AI Failure Loops: a set of interwoven, socio-technical failures that help explain how the systemic devaluation of workers' expertise negatively impacts, and is impacted by, AI design, evaluation, and governance practices. These failures demonstrate how misjudgments on the automatability of workers' skills can lead to AI deployments that fail to bring value and, instead, further diminish the visibility of workers' expertise. We discuss research and design implications for workplace AI, especially for devalued occupations.

Stress is an inevitable part of day-to-day life yet many find themselves unable to manage it themselves, particularly when professional or peer support are not always readily available. As self-care becomes increasingly vital for mental well-being, this paper explores the potential of social simulation as a safe, virtual environment for practicing stress relief for everyday situations. Leveraging the immersive capabilities of VR, AR, and LLMs, we developed eight interactive prototypes for various everyday stressful scenarios (e.g. public speaking) then conducted prototype-driven semi-structured interviews with 19 participants. We reveal that people currently lack effective means to support themselves through everyday stress and found that social simulation fills a gap for simulating real environments for training mental health practices. We outline key considerations for future development of simulation for self-care, including risks of trauma from hyper-realism, distrust of LLM-recommended timing for mental health recommendations, and the value of accessibility for self-care interventions.

Background Online mental health communities (OMHCs) are an effective and accessible channel to give and receive social support for individuals with mental and emotional issues. However, a key challenge on these platforms is finding suitable partners to interact with given that mechanisms to match users are currently underdeveloped or highly naive. Objective In this study, we collaborated with one of the world’s largest OMHCs; our contribution is to show the application of agent-based modeling for the design of online community matching algorithms. We developed an agent-based simulation framework and showcased how it can uncover trade-offs in different matching algorithms between people seeking support and volunteer counselors. Methods We used a comprehensive data set spanning January 2020 to April 2022 to create a simulation framework based on agent-based modeling that replicates the current matching mechanisms of our research site. After validating the accuracy of this simulated replication, we used this simulation framework as a “sandbox” to test different matching algorithms based on the deferred acceptance algorithm. We compared trade-offs among these different matching algorithms based on various metrics of interest, such as chat ratings and matching success rates. Results Our study suggests that various tensions emerge through different algorithmic choices for these communities. For example, our simulation uncovered that increased waiting time for support seekers was an inherent consequence on these sites when intelligent matching was used to find more suitable matches. Our simulation also verified some intuitive effects, such as that the greatest number of support seeker–counselor matches occurred using a “first come, first served” protocol, whereas relatively fewer matches occurred using a “last come, first served” protocol. We also discuss practical findings regarding matching for vulnerable versus overall populations. Results by demographic group revealed disparities—underaged and gender minority groups had lower average chat ratings and higher blocking rates on the site when compared to their majority counterparts, indicating the potential benefits of algorithmically matching them. We found that some protocols, such as a “filter”-based approach that matched vulnerable support seekers only with a counselor of their same demographic, led to improvements for these groups but resulted in lower satisfaction (–12%) among the overall population. However, this trade-off between minority and majority groups was not observed when using “topic” as a matching criterion. Topic-based matching actually outperformed the filter-based protocol among underaged people and led to significant improvements over the status quo among all minority and majority groups—specifically, a 6% average chat rating improvement and a decrease in blocking incidents from 5.86% to 4.26%. Conclusions Agent-based modeling can reveal significant design considerations in the OMHC context, including trade-offs in various outcome metrics and the potential benefits of algorithmic matching for marginalized communities.

Community and organizational policies are typically designed in a top-down, centralized fashion, with limited input from impacted stakeholders. This can result in policies that are misaligned with community needs or perceived as illegitimate. How can we support more collaborative, participatory approaches to policy design? In this paper, we present PolicyCraft, a system that structures collaborative policy design through case-grounded deliberation. Building on past research that highlights the value of concrete cases in establishing common ground, PolicyCraft supports users in collaboratively proposing, critiquing, and revising policies through discussion and voting on cases. A field study across two university courses showed that students using PolicyCraft reached greater consensus and developed better-supported course policies, compared with those using a baseline system that did not scaffold their use of concrete cases. Reflecting on our findings, we discuss opportunities for future HCI systems to help groups more effectively bridge between abstract policies and concrete cases.

Platform-based laborers face unprecedented challenges and working conditions that result from algorithmic opacity, insufficient data transparency, and unclear policies and regulations. The CSCW and HCI communities increasingly turn to worker data collectives as a means to advance related policy and regulation, hold platforms accountable for data transparency and disclosure, and empower the collective worker voice. However, fundamental questions remain for designing, governing and sustaining such data infrastructures. In this workshop, we leverage frameworks such as data feminism to design sustainable and power-aware data collectives that tackle challenges present in various types of online labor platforms (e.g., ridesharing, freelancing, crowdwork, carework). While data collectives aim to support worker collectives and complement relevant policy initiatives, the goal of this workshop is to encourage their designers to consider topics of governance, privacy, trust, and transparency. In this one-day session, we convene research and advocacy community members to reflect on critical platform work issues (e.g., worker surveillance, discrimination, wage theft, insufficient platform accountability) as well as to collaborate on codesigning data collectives that ethically and equitably address these concerns by supporting working collectivism and informing policy development.