Rethinking AI in organizations : assumptions, affordances and architectures

Abstract

This dissertation examines how the adoption of Generative AI, particularly large language model tools such as ChatGPT, reshapes human–AI relations in knowledge-intensive work and the implications for organizational practices and business models. While earlier AI systems focused on automation, prediction, and structured decision support, Generative AI introduces distinct capabilities in content creation, analysis, and interactive collaboration. Despite its rapid diffusion across industries, existing research remains fragmented and often relies on implicit assumptions that portray AI as either perfectly rational or human-like, obscuring how its organizational role is shaped by context, social relations, and the ways it is enacted in practice. To address these gaps, the study combines three methodologically complementary approaches: a problematizing literature review, semi-structured interviews with 29 professionals across knowledge-intensive industries, and a multiple-case study of eight organizations. Drawing on agency and relational perspectives, affordance theory, and business model innovation, the dissertation connects conceptual assumptions about AI with the practical capabilities Generative AI provides in knowledge work and the organizational outcomes that follow.

The findings yield three interrelated contributions. First, prevailing assumptions of rationality and anthropomorphism structure how AI is theorized in organization and management research in ways that risk oversimplifying its organizational role and obscuring its relational and context-dependent nature. Second, Generative AI introduces creational and conversational affordances that evolve through use, reinforcing one another in ways that progressively expand the scope and value of human–AI interaction over time. Third, although Generative AI supports business model innovation, its impact is asymmetric: organizations can more readily enhance internal processes than develop new offerings, and they find it hardest to establish viable revenue models. Collectively, these findings reconceptualize Generative AI as a relational, embedded technology whose organizational consequences emerge through ongoing human–AI interaction, thereby advancing organization and management theory with an integrated framework spanning theoretical assumptions, practical capability development, and business model transformation.