Generative AI has rapidly become a central concern for organizations seeking to develop their ways of working, streamline processes, and create new forms of organizational value. At the same time, research on digital innovation and affordances shows that the potential of technology is not automatically realized through implementation, but is shaped through the interplay between technical properties, organizational conditions, and the actions of actors. The purpose of this study is therefore to examine how organizations go about creating organizational conditions for the implementation and orchestration of generative AI in existing work systems.
The study is conducted as a qualitative interview study with six participants in managerial and supervisory roles with experience of digitalization, organizational development, and generative AI, across different industries. The theoretical framework combines Kohli and Melville’s model of digital innovation, Strauss, Klein, and Scornavacca’s affordance process for emergent IT, Berente, Gu, Recker, and Santhanam’s perspective on the properties of AI, namely autonomy, learning, and inscrutability, as well as resource orchestration. These perspectives are used to analyze how the participants consider that generative AI is perceived, prepared, orchestrated, and actualized, and what effects this produces in organizations, through a combination of deductive and inductive thematic coding.
The deductive contribution of the study consists of deepened, emic explanations of how existing conditions, affordance preparation, and affordance actualization manifest in practice during the implementation of generative AI. The inductive contribution is an extended conceptual framework in which five patterns emerge as central mechanisms for affordance actualization, patterns not explicitly anticipated by the deductive framework. Data emerges as a foundational primary condition upon which all other conditions depend, without which affordances cannot be actualized. Individual willingness to change is identified as a decisive personal condition whose complexity is not fully captured by existing theory. Preparation is characterized by an experiment-driven logic in which iteration and risk minimization occur in parallel. Motivation behind implementation and prioritization proves to shape the entire process in ways theory does not anticipate. Finally, human-in-the-loop emerges as a shared developed governance principle that runs through all organizations, from preparation through actualization, and becomes institutionalized as a permanent part of how work is organized even after affordances have been actualized.
Taken together, the study shows that the management and orchestration involved in actualizing affordances with generative AI is a dynamic and non-linear process that requires more than technical implementation, and that organizational conditions interact in specific and difficult-to-predict ways that vary across organizations.