Outlet Title

Association for Computing Machinery

Document Type

Conference Proceeding

Publication Date

2026

Abstract

As artificial intelligence (AI) systems become increasingly embedded in organizational decision-making and work practices, Information Systems (IS) research has devoted growing attention to questions of agency: who or what acts, decides, and bears responsibility in AI-enabled contexts. However, existing scholarship examines agency across multiple parallel research streams, raising questions about how coherently the construct is theorized as AI technologies evolve. This study conducts a bibliometric analysis of 741 AI- and agency-related articles published between 1977 and 2025 in the AIS Senior Scholars' Basket of Journals to examine how IS research conceptualizes agency and how these conceptualizations have evolved over time. Using keyword co-occurrence and temporal overlay analyses, we identify five dominant conceptual clusters through which agency is framed: algorithmic capability, trust and intentional delegation, information structuring and sensemaking, organizational transformation and governance, and evaluation, learning, and organizational value. Our findings show that while IS research increasingly conceptualizes agency as relational and negotiated, these dimensions remain weakly integrated across research streams. Interpreting these patterns through the historical evolution of AI, we argue that this fragmentation reflects a growing misalignment between how agency is theorized in IS research and how it is enacted as AI systems move from decision-support tools toward judgment-oriented generative roles. By clarifying both the structure and limitations of existing scholarship, this study contributes to IS theory by identifying a key obstacle to cumulative theorizing on human–AI agency and provides a foundation for more integrative research on the design, governance, and consequences of AI-enabled systems.

Download

Share

COinS