As artificial intelligence moves rapidly from concept to implementation, governments are under increasing pressure to demonstrate that adoption will be both effective and accountable.
In the United States, the ongoing overhaul of the Federal Acquisition Regulation, combined with agency workforce reductions and heightened congressional scrutiny, underscores the urgency of this moment. Lawmakers want answers on AI. Agencies want direction. And across government, the pressure to modernize is building.
Recent hearings highlighted the fact that while enthusiasm for AI is strong, fundamental questions of oversight, governance, and measurable outcomes remain unresolved. Although these challenges are most visible in Washington, the principles for addressing them are relevant across governments worldwide.
Accountability Beyond Compliance
True accountability requires more than compliance checks or procedural reviews. Leaders must ask which programs are delivering meaningful outcomes, how progress is being measured, and why certain initiatives succeed where others falter. Effectiveness should be demonstrated not by organizational reputation or scale, but by proximity to mission needs, clarity of responsibility, and a record of execution that withstands scrutiny. When outcomes are placed at the center of decision-making, public institutions are better equipped to justify investment and sustain trust.
That is what effective modernization looks like. It is a matter of proximity, urgency, and ownership.
Clear Leadership and Proximity to the Mission
Strong programs begin with ownership that is clearly established and publicly understood. Each major AI initiative should have a leader with explicit authority over funding, personnel, and timelines, ensuring that disputes are resolved decisively and accountability cannot be diffused. The strength of this leadership is amplified when technical teams are placed directly alongside the operators and analysts who will ultimately rely on the systems. Proximity shortens the distance between design and feedback, exposes operational realities that are invisible on briefing slides, and enables leaders to make informed decisions about which projects warrant expansion and which should be discontinued before resources are wasted.
Measuring Progress with Purpose
Funding and oversight mechanisms must reinforce this focus on results. Open-ended commitments that lack measurable benchmarks risk encouraging activity without guaranteeing value. Instead, agencies should identify a limited number of operationally relevant outcomes at the outset, such as faster processing times for critical workflows, improved detection or accuracy in high-value tasks, or costs avoided through efficiency, and require regular reviews against those benchmarks. When these results are tracked through simple, transparent scorecards, leaders, oversight bodies, and the public can judge progress for themselves, ensuring that momentum reflects reality rather than aspiration.
Laying the Groundwork for Scale
The foundation of accountability also depends on addressing data rights and evaluation frameworks at the beginning of a program rather than as an afterthought. Establishing agreements on access, privacy, and performance standards early allows engineering teams to focus on solving mission problems rather than negotiating policy conflicts. With test data, performance thresholds, and independent checks in place, governments can scale initiatives with confidence.
Procurement practices should support this foundation by combining speed with control: competitive, time-limited awards encourage innovation and learning, while contract terms must guarantee that the data, models, and artifacts produced with public resources remain under public ownership. Flexible contracting vehicles can reduce cycle times, but awards should be structured to require regular demonstration of progress before further expansion.
Preparing the Government Workforce to Carry AI Forward
None of these measures will succeed without an empowered workforce that understands both the mission and the technology. Program managers must be equipped to evaluate vendor claims, operators need to articulate mission requirements and constraints, and analysts must be trained to interpret model behavior in ways that drive improvement.
Training should be practical and embedded in mission tasks, beginning early enough that adoption takes place during development rather than being bolted on after deployment. When the workforce is prepared to use AI tools with confidence, governments are able to adapt more quickly, identify weaknesses before they undermine performance, and ensure that technology enhances rather than disrupts their core responsibilities.
Trust as the Ultimate Measure
The ultimate test of modernization is not whether governments can launch pilots or announce bold initiatives, but whether they can deliver measurable improvements that strengthen public confidence. Programs that establish clear ownership, embed technical expertise where it is most needed, reward demonstrated progress, and prepare their people to carry these systems forward are best positioned to succeed. By retaining control of the knowledge and data generated with public funds, governments also preserve the ability to refine and adapt their systems over time, creating durable capacity rather than fleeting experiments.
Artificial intelligence offers the potential to improve efficiency and precision in government, but without strong measures of responsibility and transparency, it risks becoming an expensive drain on resources and public trust. Institutions earn legitimacy by demonstrating who is accountable, what is improving, and how those improvements are known. These challenges are not unique to Washington, as governments around the world face similar questions of procurement, workforce readiness, and public trust as they scale AI.
AI does not lower that threshold for accountability; it raises it. Leaders who embrace ownership, visibility, and measurable outcomes will define whether AI becomes a trusted tool for public service or another costly promise that failed to deliver for taxpayers.
