Logo: to the web site of the Swedish Defence University

This thesis explores how Artificial Intelligence and machine learning can create military utility within Integrated Logistics Support processes to increase the availability of ground vehicles in the Swedish Armed Forces. As modern military logistics face demands for higher efficiency and resilience, AI offers potential solutions, but also introduces socio-technical risks. Through a qualitative, descriptive case study based on seven expert interviews, this study analyzes the phenomenon using the Military Utility framework (Effectiveness, Suitability, Affordability).

The findings indicate that AI has high potential effectiveness. Reducing Mean Time to Repair via Large Language Models used for searching technical manuals, and enabling predictive maintenance through sensor analysis. Affordability is enhanced by AI's ability to run robust simulations for material demand, challenging vulnerable "just-in-time" supply chains. However, Military Suitability is currently constrained by poor data standardization for ground vehicles, strict information security demands, and organizational skepticism toward "black box" technology. Furthermore, the study highlights critical risks, such as "peacetime bias" in training data and "automation bias", where personnel risk losing fundamental mechanical competence. Conclusively, to safely leverage its military utility, AI must be implemented strictly as an advisory decision-support tool, keeping the human in the loop.