The Conceptual Ambiguity of AI Adoption: A Construct Clarification and Layered Taxonomy

Abstract

Artificial intelligence (AI) adoption has become a central topic in organisational research, particularly as firms increasingly invest in AI-enabled systems to improve decision-making, automate processes, enhance productivity, and support new forms of value creation. However, the construct of “AI adoption” remains inconsistently defined and operationalised across the literature. This lack of construct clarity creates difficulty in comparing findings, developing cumulative theory, and identifying the specific organisational effects of learning-based AI capability. This study aims to clarify how AI adoption is used in organisational research and to distinguish learning-based AI capability from related distinct technological layers. Using a targeted qualitative literature review, the study examines scholarly articles published between 2019 and 2026 that address organisational, managerial, or firm-level AI adoption, readiness, implementation, or utilisation. Each paper was analysed according to the technological artefact examined, the extent to which AI was explicitly defined, and whether the study distinguished AI capability from broader digital technologies. The findings show that 45% of the reviewed corpus examined bounded AI applications such as machine learning, natural language processing, computer vision, or predictive algorithms. The remaining studies focused on AI-adjacent digital infrastructure, automation without learning, or standard enterprise digitalisation. In response, the study develops a layered taxonomy of organisational AI adoption and proposes boundary criteria for identifying AI capability. The study concludes that clearer operationalisation of AI adoption is necessary to improve measurement precision, reduce conceptual ambiguity, and support more consistent future research on AI adoption in organisational settings.