Participation and Procedural Justice in AI Governance

Session Submission Type: Panel

Abstract

This panel examines how artificial intelligence shapes governance, participation, and decision-making across public and organizational contexts. As AI systems are increasingly used in workforce programs, regulatory processes, and employment decisions, they raise questions about transparency, participation, and how human impacts are recognized. Bringing together four studies, the panel focuses on how AI systems and policy processes structure participation, shape decision-making, and influence how individuals experience and respond to these systems.

The first paper introduces the concept of “civic algorithms,” defined as algorithmic systems that mediate access to government opportunities, benefits, and services. It develops the Civic Algorithm Governance Standard (CAGS), a framework that specifies requirements for transparency, human oversight, data governance, equity monitoring, and public accountability. Using the case of Michigan’s MiDAS system, the paper shows how gaps in procedural safeguards can scale into systemic harm.

The second paper examines how policy framing shapes participation in AI-related federal rulemaking. Using computational analysis of policy texts and public comments, it analyzes how features such as complexity, readability, and technical orientation are associated with participation intensity, semantic breadth, and inclusiveness across stakeholder groups, including industry, civil society, academia, government, and individuals.
The third paper evaluates whether AI literacy programs can expand participation in AI policy agenda-setting. Using a survey experiment, it distinguishes between technical and socio-ethical forms of AI literacy and examines their effects on macro-political and micro-political participation. The findings show that while literacy increases competencies, it does not meaningfully increase participation, with political self-efficacy remaining a key constraint.

The fourth paper studies whether large language models recognize human impacts in employment decisions. Using experimental vignettes comparing human managers and AI systems, it measures labor harm sensitivity across domains such as dignity, agency, fairness, and safety. The results show systematic differences across models, with variation in how different types of harm are recognized.
Together, these papers show how AI systems and policy processes shape participation, structure decision-making, and influence how impacts are recognized across different domains. By examining governance frameworks, policy design, capacity-building interventions, and model behavior, the panel provides empirical and conceptual insights into the challenges of governing AI in ways that are transparent, accountable, and responsive to human impacts.

Policy Area

• Politics, Media, and the Policy Process

Chair

• Jae Yeon Kim, University of North Carolina at Chapel Hill

Discussants

• Nicholas Croce, University of Pittsburgh

• Luisa Nazareno, Virginia Commonwealth University

Organizer

• Daniel S. Schiff, Purdue University

Individual Presentations

• Policy Framing and Participation in Complex Technology Rulemaking: Evidence from AI Regulation - Presenting Author: Jin Ai, Rutgers University - Newark

• Governing Civic Algorithms: A Procedural Justice Framework for AI-Mediated Workforce Systems - Presenting Author: DORCAS U ATTAH, Workforce Intelligence Institute

• Is AI Pro-Human - Presenting Author: Yong Suk Lee, University of Notre Dame

• Civic Capacity or Technical Competence: Can AI Literacy Democratize AI Agenda-Setting? - Presenting Author: Indira Dhananjaya Patil, Purdue University