Industrial businesses can pursue high automation rates without removing people from decision-making altogether, according to research by Kai Hoberg, Christina Imdahl and William Schmidt. Their proposed model assigns routine cases to algorithms and reserves human attention for decisions where expert intervention is likely to add measurable value.
The framework first identifies when employees improve algorithmic recommendations. A model then predicts the likely benefit of human input, allowing companies to automate low-value interventions while keeping higher-value cases under review. Human corrections are subsequently fed back into the system to support ongoing algorithm improvement.
The researchers tested the approach using more than 475,000 order-related decisions collected over four years at a major commercial vehicle manufacturer with annual revenue above €4 billion. Rather than comparing human and AI performance in general terms, they examined the conditions in which employees delivered better decisions than the algorithm.
About 84% of decisions could be automated in the empirical application. The result points to a targeted allocation of expertise rather than a universal replacement of human judgement.
For industrial companies, the practical implication is to measure the value of intervention before setting automation targets. This can reduce repetitive work while preserving feedback that helps systems respond to changing operating conditions. The research involves Kühne Logistics University.
A multilingual professional experienced in Europe, Canada, and China, Herbert has developed invaluable networks in the automotive and energy industries. He has led high-profile projects involving ENBW, Mercedes-Benz Group, Siemens Group, and the Fraunhofer Institute.