AI reliance may trigger irreversible knowledge loss in companies

A new study warns that the growing reliance on AI could lead to a dangerous cycle of knowledge loss within organisations. While artificial intelligence offers major benefits, its long-term impact on human expertise raises concerns. Researchers suggest that without careful management, both AI systems and human skills may degrade over time. The study combines ideas from organisational learning and machine learning to explore how human and AI-driven knowledge interact. When AI systems experience concept drift—where changes in technical, societal, or business conditions reduce their accuracy—they can lose predictive power. If no one remains to retrain or correct them, this erosion becomes permanent.

One real-world example involved an automotive company where a highly skilled production worker’s expertise was lost after an AI system drifted. With no one left to adjust the model, the knowledge gap widened. The study also highlights that mass layoffs, often justified by AI-driven efficiency gains, worsen the problem. Firms like Amazon and SAP have cut jobs while increasing automation, risking long-term expertise loss. There is no fixed timeline for when knowledge loss begins—it depends on how quickly the environment changes. The researchers stress that organisations must actively preserve human expertise to prevent irreversible damage. Regularly updating AI models and critically reviewing their outputs can help, though these steps slow down short-term efficiency gains. The findings remain conceptual, with further empirical research needed to confirm the risks. However, the authors argue that without intervention, AI adoption could ultimately weaken both human and machine-driven knowledge.

The study calls for organisations to balance AI efficiency with long-term knowledge retention. Investing in model updates and maintaining human oversight are key to avoiding permanent losses. Without these measures, the combined decline of human expertise and AI reliability could harm future performance.