Managing AI Risk in Public Agencies: What Public Servants Need to Know Now

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

By Chiamaka Nwede
April 10, 2026

Several years ago, while working with a local education program in Nigeria, I sat in a meeting where administrators introduced a new digital assessment platform. The vendor described it as “data-driven” and “AI-enhanced.” The promise was compelling: faster grading, early detection of learning gaps, and better resource allocation.

But when teachers began using the system, something felt off. Students who actively participated in class were flagged as “at risk.” English language learners were scored inconsistently. When educators asked how the system reached its conclusions, no one in the room could explain the logic. The answer was simply, “That’s how the model works.”

At that moment, I realized something important: public servants remain accountable for decisions even when those decisions are shaped by tools they did not build and do not fully understand. Artificial intelligence (AI) is no longer a future issue for public agencies. It is embedded in hiring systems, benefits platforms, fraud detection tools, and service chatbots across federal, state, and local government. Many managers do not think they are implementing AI. Yet they are already supervising it. Managing AI risk is now a core responsibility of public administration.

AI Is Quietly Embedded in Government Work

Today, agencies use AI-enabled systems to:

• Screen job applications
• Flag potential fraud in assistance programs
• Predict infrastructure maintenance needs
• Route 311 service requests
• Analyze large datasets for budgeting decisions

Often, these tools arrive through procurement contracts bundled inside larger enterprise software systems. A department may purchase a vendor platform without realizing it includes automated scoring or predictive components. This creates a governance challenge. The technology may be complex, but accountability remains simple: the agency is responsible.

AI Risk Is a Management Issue

AI risk is not just technical. It is managerial.

Three risks are especially relevant to public administrators:

Transparency gaps. If staff cannot explain how a decision was made, public trust suffers.
Equity concerns. Automated systems can replicate historical disparities embedded in data.
Automation bias. Employees may defer to algorithmic outputs even when professional judgment suggests otherwise.

Public agencies operate under due process requirements, open records laws, and heightened public scrutiny. That makes proactive oversight essential.

Procurement Sets the Tone

In my experience working across education and public programs, the most critical moment is often the procurement phase. Most AI tools enter agencies through procurement contracts. However, procurement processes often focus on cost, speed, and functionality rather than algorithmic accountability.

Public managers should ask vendors clear questions:

• Does this system use automated decision-making?
• What data was used to train the model?
• How frequently is the system audited?
• Can human staff override automated decisions?
• What documentation is available if outcomes are challenged?

If these questions are not addressed during procurement, agencies may struggle to enforce accountability later.

Build Internal Capacity Before Problems Arise

Many agencies lack in-house expertise to evaluate AI tools. That is understandable. But waiting for a public controversy is not a strategy.

Leaders can take practical steps now:

Designate a technology oversight lead. This may simply involve assigning responsibility to an existing senior staff member rather than creating a new department.

Create an AI inventory. Identify which systems use automated decision-making and document their purpose, vendor, and oversight structure.

Train managers, not just IT staff. Supervisors who manage frontline employees need to understand how automated tools influence workflow and service delivery.

Establish review protocols. Regularly assess whether outcomes align with agency mission and equity goals.

These steps strengthen governance without slowing innovation, and they reduce risk before it escalates.

Communicate with the Public

When agencies adopt new technology, they often communicate internally but not externally. That can create information gaps. Constituents deserve to know when automated systems affect their applications, permits, or services.

Clear public communication should explain:

• What the system does
• What it does not do
• How decisions are reviewed
• How individuals can appeal

Silence creates space for misinformation.

A Practical Framework for Public Managers

Public agencies do not need to halt innovation. They need to manage it responsibly.

A simple framework can guide action:

1. Identify where AI is used.
2. Assess risk and equity implications.
3. Document decision-making processes.
4. Train supervisors and staff.
5. Communicate clearly with the public.

These steps align with core public administration values: accountability, transparency, and service quality.

Leadership in the Age of Algorithms

Public administration has always adapted to technological change. But AI is different because it influences judgment itself. If a benefits application is denied due to automated scoring, the public will hold the agency accountable. If a predictive tool produces inequitable outcomes, elected officials will demand answers.

The lesson from that meeting years ago still holds: technology does not replace responsibility.

AI adoption will continue. Budget pressures, workforce shortages, and service demands will accelerate it. The question is not whether agencies will use AI. It is whether they will manage it wisely.

Public servants are stewards of public trust. That trust depends not only on innovation but on oversight. Technology evolves quickly. Public accountability does not.

---

(5 votes, average: 4.80 out of 5)
Loading...