AI Governance and the Decentralization of Technology Production: An Investigation of AI-Based IPA Bots

• 成果类型: Article; Early Access
• 署名作者: Kathuria, Abhishek; Karhade, Prasanna P.; Malik, Ojaswi; Baruri, V. K. Pani; Konsynski, Benn R.
• 署名单位: Deakin University; Chinese University of Hong Kong; University of Washington Seattle; University of Washington; University of Washington Seattle; Emory University
• 刊物名称: INFORMATION SYSTEMS RESEARCH
• ISSN/ISSBN: 1047-7047; 1526-5536
• DOI: 10.1287/isre.2023.0588
• 发表日期: 2026-02-20
• 关键词: AI governance; Artificial intelligence; IS project management; IT governance; business value of IT; information-systems; artificial-intelligence; THEORETICAL FRAMEWORK; performance; management; MODEL; PARTICIPATION; architecture; principles; modularity

摘要

The enduring tension between centralization and decentralization in IT governance takes renewed significance in the age of artificial intelligence (AI). Intelligent Process Automation (IPA) bots combine robotic process automation with AI technologies and process mining based on deep, mindful domain expertise. IPA bots challenge foundational assumptions of the IT governance literature as they represent the decentralization of technology production, their AI components learn and adapt, and their functional scope continues to expand after they are deployed. We accordingly problematize the literature and revisit the centralization-decentralization dichotomy in the context of decentralized technology production at the AI frontier. We collaborate with a Fortune 200 multinational to study how IPA projects yield successful governance outcomes of utilization and repeatability. Using a multiphase research design integrating hermeneutics, induction, and abduction, we formulate a theoretical framework that categorizes characteristics of IPA projects along the dimensions of Governance, Process, and Technology. Our research yields three distinct governance mechanisms formulated as three propositions that showcase the limits to centralization, pathways to democratization, and limits to democratization of AI deployed in the form of IPA bots. First, centralization, in the form of centrally mandated IPA projects that especially require user involvement after deployment, underperforms despite the advantages traditionally associated with centralization. Second, decentralization, in the form of democratization through business user-led initiation and implementation, yields successful outcomes. Third, boundary conditions where democratization of AI has low performance lie in the combination of business user-led initiation and development, paired with human-AI hybrid bots having a low extent of AI. We thereby reframe the centralization-decentralization tension to include not only who makes technology decisions, but also who produces, implements, and deploys AI systems, how extensive the underlying AI is, and how governance activities interact with postdeployment dynamics, thereby making contributions to the burgeoning literature on AI governance.