Dr. Ryosuke Nakajima Management Consultant at PwC | Adjunct Faculty at PSM & SSBM | Adjunct Professor at TBL(Dean) & GLOBIS | Guest Lecturer at VU | Postdoc Research Fellow at SWU | Editorial Board Member | Doctorate at SSBM Geneva

The Generative AI Development Paradox: Democratizing Enterprise System Development While Deepening Corporate Governance Risks                                                                       Generative AI is rapidly changing how digital solutions are created inside organizations. With the support of low-code platforms and AI-assisted development tools, non-engineering employees can now build internal applications, dashboards, workflow automations, and other digital assets with limited formal programming knowledge. This development offers clear advantages for firms, especially in environments where IT resources are constrained and business units require faster, more flexible problem-solving. However, this shift also introduces a significant governance challenge. While business users may understand the operational problem they seek to address, they often lack sufficient knowledge of system architecture, cybersecurity, data governance, integration dependencies, and long-term maintenance requirements.This paper conceptualizes this tension as the Generative AI Development Paradox: the democratization of system development through generative AI can improve local productivity while simultaneously weakening enterprise-level visibility, accountability, and control. Drawing on prior research on shadow IT, business-managed IT, low-code development, and AI governance, the paper argues that this is not simply a technical issue, but a corporate governance issue involving oversight, risk ownership, compliance, and internal control.Using a governance-oriented conceptual and qualitative approach, the study examines cases in which non-engineers use generative AI tools to develop business-relevant systems. It identifies a dual-sided pattern. On one hand, generative AI accelerates experimentation, reduces dependence on technical teams, and enables closer alignment with local business needs. On the other hand, it often produces weak documentation, vague ownership, limited testing, and insufficient understanding of data exposure and system risk.The paper concludes that firms should not suppress AI-enabled citizen development, but govern it through a model of controlled democratization. Generative AI can create substantial business value, but only if innovation is matched by stronger frameworks for documentation, review, accountability, and risk control.