The Auditing Officer: More tools. Less judgment. (Part I)

The views expressed are those of the author and do not necessarily reflect the views of ASPA as an organization.

By Mauricio Covarrubias
June 26, 2026

Imagine a senior program officer at a federal agency. She has been in public service for fifteen years. She knows the communities her agency serves. She can read a budget line and feel, before the spreadsheet confirms it, where the pressure points are. Her instincts have been sharpened over years of fieldwork, budget cycles and the kind of institutional memory that no onboarding manual can replicate. She has made difficult decisions in ambiguous situations and lived with the consequences. She knows what it means to be accountable for a decision.

Now imagine her desk in 2026. She has an AI assistant that drafts her briefing notes, flags anomalies in program data, summarizes public consultations, suggests policy options ranked by projected impact and even anticipates the questions her director will likely ask. The system is remarkable. It saves hours. Her director calls her one of the most productive officers in the agency.

But something is changing. She notices it first in small ways: she used to arrive at conclusions through a slow, sometimes uncomfortable process of sitting with complexity. Now she arrives at the system’s conclusion and asks herself whether she agrees. The direction of reasoning has reversed. She is no longer generating judgment. She is auditing it. The system is right most of the time. That is precisely what makes the shift so difficult to notice.

This is the paradox of the well-equipped public servant: as the tools multiply, the exercise of judgment atrophies. Not through negligence or incompetence, but through the perfectly rational delegation of cognitive work to systems that perform it faster, more consistently and at greater scale. The paradox is that she is doing exactly what a modern professional should do. And in doing so, she may be gradually losing the capacity that makes her irreplaceable.

What Is Being Eroded

The capacity at risk is not technical skill. It is something older and harder to name: the ability to act with genuine intention under conditions of uncertainty, to deliberate from one’s own values, to sustain the complexity of a situation long enough to understand it and to own responsibility for a conclusion that no algorithm has prevalidated.

Philosophy of action has a term for this: agency. Not the bureaucratic meaning, the government agency down the hall, but the deeper sense: the capacity to be the genuine author of one’s own judgments and decisions. It is what separates a decision from a rubber stamp, a recommendation from a reflection.

What AI systems do to this capacity is not dramatic. It happens gradually through three specific mechanisms that public sector leaders should understand.

The first is the delegation of judgment. Every time a public servant accepts an AI recommendation without fully reconstructing the reasoning behind it, the muscle of independent judgment goes unexercised. It is not a failure of critical thinking. It is a rational response to cognitive overload. But repeated over months and years, it produces a professional who can evaluate options but struggles to generate them, who can question a conclusion but finds it hard to reach one independently.

The second is the fragmentation of attention. AI tools are designed to surface what is urgent, anomalous and actionable. They are optimized for immediate response. But the quality of judgment in complex public problems depends on a different cognitive mode: sustained, unhurried attention on a problem that resists easy formulation. The capacity to stay with a difficult question, to let it develop rather than resolve it, is precisely what AI-assisted work architecture trains us to abandon.

The third is the erosion of narrative accountability. Experienced public servants carry with them a coherent story of why they do what they do: a professional identity shaped by past decisions, lessons learned and a sense of what the public interest actually demands. This narrative is not sentiment. It is the mechanism through which accountability becomes personal rather than procedural. An officer who can say, “I made this call because I understood the situation this way,” is an officer who can be held responsible and who can learn from being wrong. AI systems are, by design, accountable to data and objectives. They have no stake in the narrative. When the system becomes the primary author of recommendations, the officer’s professional story gradually loses its generative force. She becomes a character in someone else’s account rather than the author of her own.

In Part II: Why this matters for public administration and what public sector training needs to become.

---

Author: Mauricio Covarrubias is Professor at the National Institute of Public Administration in Mexico. He holds a Ph.D. from the National Autonomous University of Mexico (UNAM) and completed a postdoctoral fellowship in Government and Public Policy at the University of New Mexico in Albuquerque. He is co-founder of the International Academy of Political-Administrative Sciences (IAPAS). He can be reached at [email protected] and followed on X (formerly Twitter) @OMCovarrubias.

(2 votes, average: 5.00 out of 5)
Loading...