The Hidden Cost of AI: How Losing Human Expertise Undermines Artificial Intelligence

Tech companies globally are implementing artificial intelligence (AI) and, often, reducing their workforce. However, a new study from researchers at the University of Passau in Germany and Arizona State University in the US suggests these decisions may be shortsighted. The research indicates that relying too heavily on AI without preserving human expertise can lead to a decline in the quality of AI systems over time.

The Cycle of Knowledge Depreciation

While AI can effectively preserve the knowledge of experienced employees when they leave a company, this knowledge is not static. Machine-learned knowledge ages and requires regular updates. This creates a cycle where AI systems accept over tasks previously performed by humans, leading to a loss of human expertise. As fewer humans possess the necessary skills, updating and maintaining the AI models becomes increasingly tricky, potentially leading to inaccurate predictions and diminished performance.

How the Cycle Works:

• AI Takes Over: AI systems automate tasks, reducing the need for employees to utilize specialized knowledge.
• Loss of Knowledge: Human expertise diminishes as employees utilize their skills less frequently or leave the company. New employees may not acquire the same level of knowledge.
• AI Models Become Outdated: Aging AI models require updates using new training data and expert assessment, but the necessary human expertise is increasingly scarce.

AI as a Knowledge Trap

The study, titled “Fading Memories: The Role of Machine Learning in Organizational Knowledge Depreciation,” highlights the risk of AI becoming a “knowledge trap.” According to Prof. Dr. Jin Gerlach of the University of Passau, “Lost human specialist knowledge can affect the quality of AI models over time – in the worst case, gradually and unnoticed.” Uncritically accepting predictions from outdated AI can further erode human judgment and accelerate knowledge loss.

The Importance of Preserving Human Expertise

The researchers emphasize that long-term success with AI requires a parallel commitment to cultivating and preserving human expertise. Prof. Dr. Gerlach notes that the findings suggest a need to reconsider recent company decisions that prioritize AI implementation at the expense of human know-how. Neglecting human expertise ultimately weakens the AI systems themselves.

Study Details

The study, published in the Academy of Management Review, is a conceptual theoretical study. It combines existing research from organizational studies and computer science to create a new model explaining how organizational knowledge can be lost due to AI implementation. It does not rely on new empirical data.

About the Researchers

Prof. Dr. Jin Gerlach holds the Chair of Business Informatics with a focus on data and information management at the University of Passau. His research focuses on the impact of technology on the economy and society. Email: Jin.Gerlach@uni-passau.de

Prof. Dr. Donald Lange researches and teaches ethical issues in management at the WP Carey School of Business at Arizona State University.

Original Publication

Gerlach, J. P. Und Lange, D.: Fading Memories: The Role of Machine Learning in Organizational Knowledge Depreciation, Academy of Management Review.