stakeholder engagement

This paper identifies a core dilemma in organizational responsible AI governance: ambiguous responsibility boundaries across AI lifecycle stages and a lack of role- and stage-appropriate operational tools. Methodologically, the study systematically reviews over 220 responsible AI tools and proposes a novel two-dimensional (Actor, Stage) classification framework, integrating systematic review, meta-analysis, and qualitative coding. It identifies three critical governance gaps: (1) unclear accountability attribution, (2) absence of empirical validation for most tools, and (3) severe coverage imbalance across actors and stages. Results show that >80% of tools target developers during data and modeling phases; tools for leadership, deployers, end users, and stages such as value proposition definition and deployment are virtually absent. Moreover, >90% of tools lack empirical evidence. The study establishes a theoretically grounded, empirically benchmarked framework to advance actor–stage–aligned AI governance tool ecosystems.

This study identifies “moral stress”—an unrecognized affective discomfort and vulnerability arising from blurred role boundaries and contested decision authority—as a critical barrier to the operationalization of AI ethics tools within urban technology teams. Employing ethnographic fieldwork, qualitative interviews, and organizational behavior analysis, it introduces the concept of moral stress systematically into AI ethics practice research for the first time. Findings reveal that even under optimal conditions—adequate resources and standardized ethical workflows—existing organizational structures and tool designs fail to mitigate this stress; instead, affective experience emerges as a key mechanism underlying ethical intervention failure. The study thus extends AI ethics theory by foregrounding emotion as a constitutive dimension of ethical practice, and proposes a concrete pathway: redesigning ethics tools and organizational support systems with affective sensitivity. This reframing offers both a critical caution and empirically grounded design principles for enhancing the real-world efficacy of AI ethics implementation.

Market fundamentalism and demographic imbalances among software practitioners jointly impede the integration of ethics in software development. Method: A mixed-methods study—comprising a survey of 217 practitioners across roles, industries, and countries, supplemented by qualitative analysis—provides the first empirical evidence that marginalized groups (women, BIPOC, and persons with disabilities) exhibit significantly higher ethical sensitivity, frequency of ethical issue reporting, and willingness to intervene than their majority-group counterparts—challenging the “neutral developer” assumption and revealing structural demographic bias in ethical advocacy. The study identifies two primary barriers: market-driven organizational cultures that suppress ethical deliberation, and widespread institutional deficits—including absent ethical processes, insufficient authority delegation, and inadequate ethics training. Contribution/Results: It establishes demographic background as a critical analytical dimension for understanding variation in ethical practice and provides empirically grounded foundations for designing inclusive, equity-oriented ethics governance mechanisms in software engineering.

This study investigates cross-role (e.g., engineers, product managers, ethics specialists) and cross-national (43 countries, N=414) variations in AI ethics awareness, policy comprehension, and risk mitigation practices within AI development teams. Employing a mixed-methods design, it integrates large-scale surveys with in-depth interviews, combining quantitative statistical analysis and qualitative thematic coding. Results reveal a pronounced role-based ethical responsibility gap and a non-uniform global distribution of regulatory sensitivity and implementation capacity. Building on these findings, the study proposes a “collaborative, role-sensitive ethics governance framework” that mandates multi-stakeholder engagement across the AI lifecycle and incorporates localization mechanisms for contextual adaptation. This framework advances AI ethics practice from prescriptive, one-size-fits-all guidelines toward inclusive, situated governance—offering an actionable, differentiated pathway for global AI policy implementation and responsible innovation.

This study addresses key challenges in ESG reporting—namely unstructured data, inconsistent terminology, and complex regulatory standards—compounded by the absence of automation and dynamic feedback in current workflows. To overcome these limitations, this work proposes the first AI-driven, multi-agent framework for end-to-end ESG lifecycle management, systematically integrating five phases: identification, measurement, reporting, stakeholder engagement, and continuous improvement. The framework enables a shift from static disclosure to adaptive, accountable governance by leveraging large language models within three agent configurations: single-model, single-agent, and multi-agent. It supports automated report generation, cross-validation, multi-version comparison, and knowledge base maintenance. A prototype implementation demonstrates significant improvements in report consistency, accuracy, and adaptability, with code and data publicly released.

Existing policy design approaches are predominantly top-down and disconnected from community contexts, resulting in low legitimacy and misalignment between policy provisions and actual stakeholder needs. Method: This paper introduces PolicyCraft—a human-computer interaction (HCI)-informed system that embeds real-world cases as negotiation anchors within the policy design workflow. It supports participatory governance through case annotation, structured deliberation, multi-round voting, and versioned collaborative editing, enabling stakeholders to co-author, critique, and iteratively refine policy proposals while dynamically aligning abstract rules with concrete situational constraints. Contribution/Results: Evaluated in two university courses focused on educational policy, groups using PolicyCraft achieved significantly higher consensus and produced policies substantiated by richer, more relevant case evidence—outperforming a baseline system lacking case-based scaffolding. The work advances participatory policy design by grounding abstraction in lived experience and demonstrates scalable HCI methods for democratic, context-sensitive governance.

This study examines how the federally mandated Homeless Management Information System (HMIS) data infrastructure, implemented within a context of compulsory interorganizational collaboration in U.S. homeless services, simultaneously enhances cross-agency coordination and reinforces power asymmetries. Drawing on in-depth interviews with six domain experts, the analysis reveals that while standardization improves collaborative efficiency and knowledge sharing, disparities in resources, analytical capacity, and institutional authority render less powerful participants prone to passive compliance, thereby diminishing their influence in decision-making. The findings demonstrate that mandatory data-sharing regimes can entrench structural inequities, offering critical theoretical insights and practical guidance for designing more equitable public data infrastructures.

This study addresses the persistent challenges in implementing effective accountability mechanisms in software systems, which often stem from ambiguous definitions, unclear responsibilities, and insufficient cross-domain collaboration, hindering alignment with legal, business, and societal requirements. Through interdisciplinary workshops, expert panels, and focused group discussions, the research systematically integrates legal, technical, and social perspectives to develop a multidimensional conceptual model of software accountability. It proposes a structured approach for translating legal mandates into design specifications and introduces an evidence preservation mechanism. The work establishes, for the first time, an embedded accountability implementation framework that clarifies principles for responsibility allocation, thereby pioneering a new direction in which accountability is inherently integrated into software design. Additionally, it identifies critical challenges and capacity-building needs inherent in interdisciplinary collaboration.

Current AI governance is highly centralized among a few actors, risking technological bias and insufficient public participation, while lacking transparent, auditable, and multi-stakeholder evaluation tools. Method: We propose the AI Plurality Index (AIPI)—the first verifiable, auditable quantitative framework for pluralistic governance—systematically measuring stakeholders’ substantive involvement in goal-setting, data practices, safety assurance, and deployment decisions. It innovatively incorporates evidence coverage metrics and lower-bound scoring to enable public arbitration and versioned updates. Our reproducible pipeline integrates structured analysis of networks/codebases, third-party assessments, and expert interviews, with robustness ensured via evidence coding, cross-validation, and sensitivity analysis. Contribution/Results: We open-source the protocol, coding manual, scoring scripts, and evidence graph; conduct pilot evaluations across major AI providers; and establish comparable benchmarks against existing governance frameworks.

This study addresses the challenges of communicating climate mitigation data—characterized by diverse audiences, complex datasets, and variable contexts—through a ten-month co-design process involving eight workshops, fifteen rounds of stakeholder engagement, and iterative prototyping. The effort culminated in the UK Co-benefits Atlas, a comprehensive resource spanning over 400 pages. The project introduces a five-dimensional framework centered on data, people, stories, context, and the atlas itself, integrating interactive visualization, explanatory narrative, and participatory design to support both guided exploration and interpretive presentation. Beyond delivering a functional atlas, the research uncovers key user behaviors, including interest-driven navigation, friction caused by data overload, and interaction patterns shaped by real-world usage scenarios, thereby offering a structured design paradigm for visualizing complex policy-related data.

This study addresses the challenge of accountability in open-source software ecosystems, where the often-conflicting demands of diverse stakeholders—including volunteers, corporations, and end users—hinder communities’ ability to effectively identify and fulfill their responsibilities. For the first time, this work systematically focuses on the issue of accountability in open source, convening a cross-disciplinary workshop at Carnegie Mellon University with 24 domain experts to foster qualitative dialogue between researchers and practitioners. The project proposes a stakeholder-centered accountability agenda, articulates key research questions, and outlines an initial roadmap to guide future scholarly inquiry and community practice in this critical area.