Paper Number
ECIS2026-2543
Paper Type
CRP
Abstract
Generative AI systems, also referred to as generative AI agents, exhibit different levels of autonomy, defined as a function of their generative autonomy and workflow autonomy. We conceptualize generative autonomy as the extent to which a system is autonomous in creating content, and workflow autonomy as the extent to which a system is autonomous in how its components interact with each other and with other systems. Both dimensions of autonomy create challenges for providers of generative AI systems, and it remains unclear how they manage the autonomy. We conduct a multiple-case study of seven generative AI systems that differ in their autonomy to identify 20 practices for harnessing value-adding autonomy or reducing autonomy that is unproductive or destructive. Our findings contribute to the literature on the autonomy of generative AI systems, managing this autonomy, and the design of generative AI systems. We also inform providers of generative AI systems.
Recommended Citation
Heimburg, Vincent and Wiesche, Manuel, "Harnessing and Reducing The Generative and Workflow Autonomy Of Generative AI Systems: A Multiple-Case Study" (2026). ECIS 2026 Proceedings. 23.
https://aisel.aisnet.org/ecis2026/is_adopt/is_adopt/23
Harnessing and Reducing The Generative and Workflow Autonomy Of Generative AI Systems: A Multiple-Case Study
Generative AI systems, also referred to as generative AI agents, exhibit different levels of autonomy, defined as a function of their generative autonomy and workflow autonomy. We conceptualize generative autonomy as the extent to which a system is autonomous in creating content, and workflow autonomy as the extent to which a system is autonomous in how its components interact with each other and with other systems. Both dimensions of autonomy create challenges for providers of generative AI systems, and it remains unclear how they manage the autonomy. We conduct a multiple-case study of seven generative AI systems that differ in their autonomy to identify 20 practices for harnessing value-adding autonomy or reducing autonomy that is unproductive or destructive. Our findings contribute to the literature on the autonomy of generative AI systems, managing this autonomy, and the design of generative AI systems. We also inform providers of generative AI systems.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.