Reimagining Government, Part 4: Transforming Capacity

Artificial intelligence is already making its presence felt in the American government and other governments around the world. Looking globally, OECD’s “Principles on Artificial Intelligence” in 2024 documented policy actions related to more than 1,000 AI initiatives in over 70 countries, almost half of which were for AI use in public sector organizations. In the U.S., an OMB inventory in 2024 showed nearly 1,800 AI use cases across 37 Federal agencies. Looking at Generative AI alone, analysis by GAO in 2025 showed a ninefold rise in use cases across eleven selected agencies, from 32 in 2023 to 282 in 2024.

As in private industry and in people’s daily lives, much of the thinking about what AI can do in government centers on greater efficiencies and effectiveness in what government already does today. Easier faster processing of permits, licenses, benefits applications. Automation of data entry into extant IT systems. Instant answers to citizen queries by AI chatbots. Data analytics on large datasets to enhance medical research and track disease outbreaks, detect financial misconduct and fraud, identify suspicious patterns in border crossings. In some use cases, the vision (and the emerging reality) is to supplant the human public servants who performed these functions. In others, it’s to free up time for public servants to focus on those tasks that require judgement, creativity, discretion, and decision-making. There is immense value in both visions, and realizing them will transform how government operates.

What if AI also transformed what government does, what it is capable of doing, opening the doors to things that no government has ever been able to do before?

One of the things AI is good at is synthesizing information to generate tailored content. When we think about that in a government context, it’s often for personalized service delivery, like chatbots telling users about benefits or programs specifically relevant and responsive to their individual needs. Imagine it providing personalized civics education to promote public service – making information about what government does succinct and approachable, meeting each person where they are to empower and inspire them to be part of civic dialogue and action. Imagine a level beyond that, where AI powers participatory deliberation platforms that synthesize millions of citizen inputs into policy proposals.

Regulations and laws often are reactive, developed in response to events, problems, or conditions that have already occurred or been identified. Some are proactive, designed to anticipate and prevent future issues – think of some public health measures or land use regulations. What if they could be both at once – what if AI enabled adaptive laws and regulations that monitored themselves and evolved without executive, legislative, or judicial processes? Environmental rules that tighten when pollution rises. Social safety net protections that calibrate dynamically to changes in local or national economic conditions. We may well choose to never have such rules automatically implement themselves. But imagine putting AI to work to design and dynamically redesign them, with all their implications exhaustively analyzed and documented, for human leaders in the loop to then authorize for implementation (or not)? AI reasoning models will soon be able to do it.

AI may also have the potential to help realize the elimination of cross-agency and cross-level frictions like we imagined above. At any one level of government, and on any given issue government addresses, there is no collection of human minds capable of detecting, assessing, and depicting all the interrelationships and bottlenecks between agencies. Add multiple levels, or the interrelationships among multiple issues, and the complexity is even less navigable by human intellect. Future AI models can have the capability, perhaps not to “understand” public policy systems of systems of such scale and complexity, but to render them understandable and actionable for public servants in unprecedented ways.

AI will replace people and eliminate jobs in government, just as it will in the private sector, perhaps in large numbers. And: the potential of AI in government will be unrealizable without what public servants are uniquely capable of bringing to its use. AI already today is more than just a technology imperative. It’s a leadership imperative. An organizational design imperative. A cultural imperative. It is a responsibility not just for Chief Information Officers or Chief Technology Officers but for every executive in every government enterprise – and not just an executive responsibility. The multi-dimensionality of AI is critical to build into professional development for current public servants at every level, and into university-level public administration education and earlier civics education. There may be no greater key to reimagining government than putting AI to work to change not just how government operates but what it can be, creating the futures described here, or other futures.

Reimagining government as an array of alternative futures isn’t an exercise in prediction. It’s a way to surface and consider some of the choices we can make today that can shape the future in different ways. Deliberate actions today are what determine our prospects for bringing about the futures we want to see for government, and for avoiding the ones we don’t. The kinds of futures imagined at the NAPA conference are some among many. What are some of the actions that could contribute to decision-making about futures like these?

• Inventory and evaluate examples of innovations in each theme area

A number of the individual ideas within the themes discussed above are already being seen around the country as well as outside the United States. Some are referenced above – New Mexico’s use of social media influencers, the UK’s data trust, and others. Research can find more such examples and compile them so they can be examined and evaluated for how well they’re working, what impacts they’re having, what obstacles they’re facing. Doing so could inform decisions that government entities, citizen groups, and others may want to make about emulating or scaling them. Academics in public policy programs could join in new collaborations with public interest groups and others in this research.

• Survey public servants and citizens on futures that look like these – or look different

The participants at the NAPA conference represent a valuable starting point for exploring themes like partnering in governing with social media influencers, easing cross-level collaboration, new forms of citizen engagement, and transformative use of AI. Widening the pool of people considering such futures could shed important light on their benefits and potential drawbacks, and surface other visions for government that may be even more compelling. America’s semiquincentennial celebration year in 2026 is an opportune time for leaders at Federal, state, county, and municipal levels to learn through surveys or other means what kinds of futures their peers and their constituents want to see and experience.

• Develop pilot designs that could be provided to interested communities

The ;ideas explored above involve government entities, citizen groups and individual citizens, media figures and enterprises, tech companies, and others working together in new ways. Before making decisions to try to bring them into being, communities interested in them would benefit from detailed attention to questions like: how would this actually work? What resources would be required? Who should have what kinds of roles and responsibilities? How would impacts be measured? Designing pilots that carefully think through these kinds of “how-to” considerations could be another fruitful area for researchers at public policy schools to work with diverse stakeholders including elected and appointed government officials, issue advocacy groups, and citizen volunteers.