Artificial Intelligence in the Utility Sector: A Pragmatic Study on How Utilities Can Acquire Capabilities Needed for AI Maturity

Abstract

This thesis examines how organizations within the utility sector can acquire both technical and organizational capabilities to achieve artificial intelligence (AI) maturity. Adopting a qualitative, pragmatic approach, it employs a comparative case study of seven Danish utility organizations supported by 18 semi-structured interviews. The analysis reveals that organizations acquire technical capabilities through enhancing IT infrastructure and data governance, and organizational capabilities are acquired through cultivating an AI-friendly culture, initiating small scale pilots to test potential value and build knowledge at a reasonable financial commitment, and supplementing internal skills with external expertise. However, the development of capabilities in the ethical domain remains underdeveloped. Findings indicate that AI maturity levels vary among these utilities, but they are incrementally building foundational AI capabilities without formal Responsible AI frameworks guiding their efforts. This highlights a common focus on technical and organizational capabilities while ethical considerations lag. The study contributes to existing literature by contextualizing established AI maturity frameworks, such as Akbarighatar et al. (2023), within the unique institutional, regulatory, and operational realities of the utility sector. Sector-specific nuances, such as regulatory constraints, risk-averse cultures, and public accountability, significantly shape how capabilities are prioritized and developed. Notably, organizational size plays a dual role, as smaller utilities demonstrate agility in adopting vendor-based AI solutions but face structural limitations in developing internal capabilities.These insights not only extend current theoretical frameworks but also offer practical implications, showing that utilities progress toward AI maturity through gradual capability-building tailored to their unique industry context and constraints.By offering both theoretical insights and actionable recommendations, the thesis supports utility organizations, policymakers, and practitioners in understanding the sociotechnical alignment required for AI maturity. It argues that AI maturity is not a linear progression, but a dynamic process influenced by both technological and organizational maturity.