PhD defence by Jon Eklöf

Title

AI Transformation in the Manufacturing Industry

Abstract

The manufacturing industry is becoming increasingly complex and connected. The rise of artificial intelligence (AI) presents a transformative opportunity through advanced analytics to process vast amounts of manufacturing data. While many studies explore AI's potential benefits in manufacturing, a small percentage of firms have truly integrated AI, while most have only been implementing AI in isolated business processes. Some predict AI will revolutionize manufacturing. This thesis investigates the prerequisites for manufacturing leaders to effectively support the broad introduction of AI. It combines quantitative and qualitative methods with real-life AI integration instances from an aerospace manufacturing company.

Current research provides is limited guidance for leaders on how to best support an AI transformation. In a first study, we developed a capability framework for leaders working with AI implementation. A standout capability was a genuine willingness to learn about AI, despite the evolving nature of the technology and its sometimes vague definition. To be able to support an AI transformation, leaders require some understanding of the nature of AI. Recent technological developments have made AI development accessible beyond experts. In a second study, we therefore explored abstraction in AI, especially in deep learning, noting a significant decrease in the lines of code, suggesting increased abstraction. While this facilitates AI's democratization and collaboration, it also introduces challenges like quality issues and talent scarcity that many firms are not equipped to address.

Existing AI research in manufacturing often exists in isolation from day-to-day operations. To address this gap, a third action research case study was performed on a global manufacturer employing AI to augment capabilities and decisions. This study provides a comprehensive examination of the challenges and considerations in AI application in manufacturing. It underscores the pivotal role of leaders in this adoption. Although AI can be potent, it is not a universal fix. Its deployment should be carefully considered. The findings of the thesis emphasize the significance of a multidisciplinary approach and collaboration for a successful AI transformation.

Supervisors

Principal Supervisor Thomas Wim Hamelryck

Assessment Committee

Associate prof. Cosmin E. Oancea, DIKU
Professor Lars Walter, University West, Sweden
Data and analytics expert Salla Franzén, IKEA

For an electronic copy of the thesis, please visit the PhD Programme page.