Generative AI Overreliance Risks 'Epistemic Deskilling' of Managers

Overreliance on generative AI in the workplace may be eroding the practical wisdom and moral insight essential for effective leadership. Researchers warn that when managers outsource critical thinking to AI, they risk a systemic decline in their ability to navigate complex organizational challenges.

In a study published in the Academy of Management Review (2026), a team from the University of Bath, Ohio State University, University of Lausanne, and Cardiff University identified a phenomenon termed "epistemic deskilling." This process occurs when managers lose the ability to learn from firsthand experience and anticipate future goals because they rely too heavily on AI-generated outputs. The researchers describe a state of "cognitive atrophy," characterized by a "prompt-and-accept" reflex and a shrinking threshold for the mental effort required to solve difficult problems.

The Erosion of Managerial Phronesis

This decline centers on the loss of "managerial phronesis," or the practical wisdom developed through lived experience. While AI can process vast data patterns, it lacks the capacity to understand the social and emotional complexities of a workplace. Professor Dirk Lindebaum of the University of Bath notes that Gen-AI cannot replace lessons learned through experience because it does not experience the world or understand the consequences of the decisions it suggests.

According to the research, the tension between AI efficiency and human judgment is most acute under intense time pressure. In these environments, decision-makers risk becoming passive "decision-approvers" who can no longer justify their actions without the assistance of an AI tool. The real-world stakes of this atrophy were highlighted by the retirement of West Midlands police chief Craig Guildford, a case cited as a wake-up call following the use of incorrect AI-generated evidence.

Implications for Organizational Oversight

If leaders stop engaging in the "struggle" of thinking, organizations face a significant risk to their ethical oversight and critical decision-making capabilities. The research suggests that the danger lies not in the tool itself, but in the replacement of human cognition. When managers stop explaining the "why" behind a decision to themselves, they lose the very skills that make them leaders.

However, the study offers a path toward "epistemic upskilling." The researchers suggest that AI can be used as a tool for reflection rather than a replacement for thought. By using AI to challenge assumptions or provide alternative perspectives—rather than simply generating the final answer—managers can actually enhance their judgment.

The Path Forward

As generative AI becomes a routine part of corporate infrastructure, the focus must shift toward maintaining human agency. Professor Lindebaum emphasizes that managers must prioritize explaining to themselves and others the elements that Gen-AI cannot satisfactorily explain.

Industry observers will now be watching how organizations implement AI governance to prevent cognitive atrophy. The primary challenge remains whether firms can balance the speed of AI with the slow, often difficult process of developing human wisdom.