Over-reliance on generative AI risks eroding managers’ practical wisdom, study finds

Generative artificial intelligence has quietly become part of the daily routine of working life, drafting emails, summarising reports and generating ideas on demand. But a new study led by the University of Bath warns that the convenience of these tools may carry a hidden cost: the gradual erosion of the very judgement that makes managers effective. Published in the Academy of Management Review, the research explores how tools such as ChatGPT may affect what academics call managerial phronesis, the practical wisdom that managers develop through real-world experience, reflection and human interaction, and concludes that over-reliance on generative AI could undermine this capacity unless the technology is deployed with care.

The research team, comprising Professor Dirk Lindebaum of the University of Bath’s School of Management, Professor Natarajan Balasubramanian of Ohio State University, Dr Mehreen Ashraf of Cardiff University and Dr Patrick Haack of the University of Lausanne, set out to model how the growing presence of generative AI in the workplace interacts with the processes through which managers acquire and refine practical wisdom. Their paper, titled A Process Model of Managerial Phronesis in the Age of Generative AI, argues that the appeal of these systems lies in their speed, but that speed is precisely what makes them risky when they are treated as substitutes for thinking rather than as aids to it.

Professor Lindebaum’s central point is that generative AI, however fluent its outputs may appear, cannot replicate the lessons that come from first-hand experience. Unlike humans, AI does not experience the world, understand the consequences of decisions, or grasp the social and emotional complexities that shape behaviour in workplaces. Instead, it produces responses based on patterns found in existing data. That distinction matters because managerial phronesis is not a body of information that can be retrieved; it is a capability that accumulates through acting, observing outcomes and reflecting on them, often in situations where the right answer depends on context, relationships and moral considerations that no dataset fully captures.

The danger, according to the researchers, is that as managers increasingly outsource their thinking to generative AI, whether for idea generation or when a practical problem arises at work, they may rely less on their own judgement. Over time, this could reduce their ability to learn from experience, think critically, and anticipate what kinds of actions are needed now to meet future goals. The concern is not that AI produces wrong answers in every case, but that the habit of deferring to it weakens the mental muscles that managers need precisely in the situations where AI is least reliable: ambiguous, novel and morally charged decisions.

To describe this dynamic, the team coined the concept of epistemic de-skilling, a process in which people gradually lose knowledge-related capabilities because they outsource too much of their thinking to generative AI. The researchers suggest this is most likely to occur when managers are under intense time pressure and use the technology as a shortcut rather than engaging deeply with a problem themselves. In such situations, managers may stop asking important questions, seeking different perspectives or learning from real-world interactions. Instead of developing a nuanced understanding of employees, customers or organisational challenges, they may come to depend on AI-generated answers that lack the context and moral judgement needed for complex decisions.

The mechanism matters as much as the outcome. Practical wisdom, in the Aristotelian tradition from which the term phronesis derives, is built through the repeated cycle of confronting a concrete situation, judging what is at stake, acting, and reflecting on the results. If the first and most demanding step of that cycle is handed to a machine, the subsequent steps lose their raw material. A manager who no longer wrestles with a problem before consulting an AI system has nothing of their own to test the system’s output against, and the reflective part of the loop collapses into passive acceptance. The study frames this as a process model precisely because the damage is cumulative: each shortcut makes the next one easier, and the capability fades through disuse rather than through any single dramatic failure.

Crucially, the researchers do not conclude that generative AI is inherently corrosive to managerial judgement. They also developed the idea of how the technology may be used to develop the judgement of managers, although they note this is likely to prove more challenging. Generative AI can strengthen managerial judgement when it is used as a tool for reflection rather than a replacement for thinking, a process the researchers dubbed epistemic up-skilling. Rather than accepting AI outputs at face value, managers can use them to challenge assumptions, explore alternative scenarios and test the reasoning behind their own decisions. Because AI systems often struggle to explain why they produce particular answers, the gaps in those explanations can encourage people to think more deeply about their choices and the consequences of their actions.

That up-skilling path, however, demands deliberate effort. Professor Lindebaum emphasised that it requires persistent effort on the part of managers to fill the explanatory gaps that AI leaves behind. The study suggests this beneficial outcome is most likely when managers know they will be held accountable for their decisions. In workplaces where individuals must justify their actions and explain their reasoning, generative AI can become a trigger for deeper reflection instead of a substitute for judgement. As Lindebaum put it, it is that which generative AI cannot satisfactorily explain that managers must explain to themselves and others. Accountability, in this framing, functions as a safeguard: the obligation to give reasons forces the manager back into the reflective cycle that passive reliance on AI would otherwise short-circuit.

The implications for organisations are direct. It is becoming increasingly clear, the researchers argue, that simply introducing AI tools will not automatically improve decision-making or organisational performance. Instead, organisations need to carefully design roles, responsibilities and workflows to ensure employees continue developing the human skills that AI cannot replicate. That design challenge extends to how performance is evaluated and how time pressure is managed, since the study identifies intense time pressure as a key condition under which de-skilling takes hold. Workplaces that reward speed above deliberation may inadvertently push managers toward the shortcut that erodes their judgement, while those that build in reflection and accountability can turn the same tools into instruments of learning.

The study, published in the Academy of Management Review with the DOI 10.5465/amr.2024.0582, arrives at a moment when generative AI is being embedded into professional work faster than the norms governing its use are being established. Its message is neither technophobic nor celebratory. The same system that can hollow out a manager’s capacity for judgement under time pressure and passive reliance can, under conditions of accountability and active reflection, sharpen that capacity by exposing the limits of machine-generated answers. The difference lies not in the technology itself but in how it is woven into the daily practice of management, and in whether organisations treat AI as a replacement for experienced human thinking or as a sparring partner for it.

Subject of Research: The effects of generative AI on managers' practical wisdom and judgement in the workplace

Article Title: AI could undermine managers' judgement unless used carefully, University of Bath-led study warns

Article References: AI could undermine managers' judgement unless used carefully, University of Bath-led study warns. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: generative AI, managerial phronesis, epistemic de-skilling, epistemic up-skilling, managerial judgement, University of Bath, Academy of Management Review, workplace decision-making, accountability, ChatGPT, organisational behaviour, practical wisdom