top of page

The Future of Government

Charlie Hamer
Nov 11
12 min read

What a re-imagined government looks like in the age of AI

Maya loses her job on a Friday. By Monday, she has learned that government is not one organisation but many. She registers as unemployed with one office, applies for income support at another and searches for training on a third website, each asking for the same documents and none aware of the others. The childcare support she needs to attend a course depends on an employment status the benefits office has not yet updated. Weeks pass before anything joins up, and every delay costs her household money and agony. 


Some governments are already tackling this. When New South Wales’s customer service ministry looked at what families had to do after a death, it found they could need to notify up to 40 government, private and non-government organisations. It responded by building what became the Australian Death Notification Service, which lets families notify many of them through a single online form[1]. In Estonia, registering a child’s birth automatically triggers an offer of family benefits, so parents confirm an offer rather than file an application[2]. These steps matter because AI builds on them. AI cannot make a fragmented government intelligent; it can only amplify the system it finds.


Now imagine the same weekend for Maya in a “reimagined government”, which the Future Government Institute calls Government 4.0. Maya tells a government AI assistant what happened, in her own language. With her consent, it accesses her employment and education records, identifies the financial support she qualifies for, matches her skills to available jobs and relevant training, and helps arrange childcare and transport. She can track progress, make changes or speak to a person at any point. Rather than Maya navigating government departments, government organises itself around her needs.


The test is not simply whether a citizen like Maya receives a faster service, but whether she has to repeat her story, chase agencies or carry the burden of coordinating government. A person should be able to tell government once, see what is happening next and get help without losing control.


The technology to make much of this possible already exists. Governments are beginning to use AI to deliver services, inform decisions, understand and predict citizens’ needs and support public servants. By connecting existing systems, data and services, and applying predictive models, governments can move from fragmented, department-led processes to integrated, citizen-centred and citizen-curated delivery. The real opportunity is not simply to introduce AI into existing government structures, but to rethink how government itself works and delivers. This requires changes in how priorities are set, institutions are organised, decisions are made and results are delivered. We see six areas where this transformation matters most.


Leader’s Office & Strategy

Every head of government faces the same fundamental challenge: setting clear priorities, aligning government behind them and delivering measurable results for citizens. AI does not change this responsibility. It changes how effectively, and how quickly, governments can fulfil it.


This is easier said than done. Presidents, Prime Ministers and Chief Ministers must align dozens of ministries, hundreds of programmes and thousands of officials around a small number of national priorities, while responding to daily crises and competing demands. Too often, urgent issues crowd out long-term priorities, decisions are made with incomplete information and leaders lack a clear picture of what is working and what is not.


Much of the problem is institutional rather than technical. Proposals to let departments share data responsibly often meet strong resistance, and the cost is human: when police, health and education cannot pool what they know about a child at risk, the information that should prompt action can take weeks to arrive. Connecting government is not just a technical task: it means changing institutional incentives, power and habits.


AI is beginning to change this. In September 2026, the UAE Cabinet approved the rollout of 32 specialised AI advisers to analyse policies and support ministerial decisions[3]. In Europe, the Destination Earth initiative is developing digital twins and AI capabilities to simulate extreme weather and climate risks[4][5]. These examples show how AI can help governments anticipate risks, test policy choices and make better-informed decisions, provided the information underneath is connected.


Imagine a Cabinet using AI to test its national employment strategy against different scenarios: industries disrupted by automation, emerging skills shortages or jobs being created in regions without adequate housing and transport. Rather than responding once these challenges emerge, governments could anticipate them, assess trade-offs across ministries and adjust their strategies earlier.


This does not just change how strategy is created; it offers an opportunity to fundamentally rethink the role of the leader’s office. Rather than primarily coordinating meetings, correspondence and reporting, it could become the centre of government strategy and delivery, giving the leader a real-time view of national priorities, implementation progress, emerging risks and citizen sentiment.


Better information alone does not deliver results. Leaders need an operating model that connects priorities to decisions, responsibilities and follow-through, and that protects their time for the choices only leaders can make.


Making this work would require a small, empowered team bringing together strategy, policy, delivery, strategic communications and government effectiveness, led by a highly capable Chief of Staff working directly with the leader. As Figure 1 illustrates, AI could strengthen each function, from tracking delivery and testing policies to understanding citizen concerns, identifying institutional capability gaps and responding to crises. Underpinning these functions would be secure government data, reliable AI systems and clear accountability for decisions.



Figure 1: The leader’s office in the age of AI


AI would not replace political judgement or leadership. It would give leaders better information, greater foresight and stronger tools to coordinate government around what matters most. AI can help the leader’s office see where government is stuck and coordinate a response, but people must remain accountable for the decisions and their consequences. The ultimate test is whether governments can move from reacting to problems to anticipating them, and from announcing priorities to consistently delivering results for citizens.


Policy for the Future

Good policymaking has always been difficult. Even when the outcome is clear, governments must decide which interventions are most likely to work, drawing on international experience, lessons from past reforms and an understanding of their own context. Traditionally, this requires months of research, consultation and analysis, with limited ability to test different options before implementation.


AI is beginning to change this. In the UK, the government’s Consult tool analysed more than 50,000 responses to a public consultation on water reform in approximately two hours, followed by expert validation[6]. In Taiwan, AI-assisted facilitation supported a citizens’ assembly of around 450 participants discussing information integrity[7]. Both demonstrate how AI can help governments process evidence faster and engage citizens more effectively in policymaking.


The bigger opportunity is to move from policies designed largely on assumptions to policies that can be tested, refined and improved before they are implemented at scale. Imagine a government developing a national employment programme. AI could help assess how proposed training subsidies interact with existing welfare benefits, childcare support and local job markets, identifying unintended consequences before the programme is launched. It could also compare international approaches, model different scenarios and help design targeted pilots to test what works.


This would fundamentally change how policy teams operate. Rather than spending months gathering and synthesising information, they could spend more time evaluating options, consulting citizens and testing interventions. AI would support, not replace, expert judgement and real-world evidence. The result would be a more responsive approach to policymaking, where governments can learn faster, reduce implementation risks and continuously improve policies based on what works for citizens.


Delivery & Institutional Transformation

Even the best government strategy means little without effective delivery. Between a leader’s decisions and results for citizens sits a complex chain of ministries, agencies, districts and frontline teams, often working with fragmented data, competing priorities and limited visibility into what is happening on the ground. By the time implementation problems reach senior leaders, it is often too late to correct them.


AI has the potential to fundamentally change how governments manage delivery. Rather than relying on monthly reports and retrospective reviews, AI-enabled delivery units could monitor results from the field in near real time, from school attendance and medicine availability to benefit payments and job placements. AI could identify where implementation is falling behind, diagnose the underlying causes and recommend corrective action. By connecting data across ministries and departments, it could also identify dependencies, coordinate interventions across agencies and flag emerging problems before they become systemic.


This would also transform how leaders conduct delivery stocktakes. Instead of meetings dominated by presentations and progress updates, Presidents, Prime Ministers and Ministers could have a live picture of performance against their priorities, understand where delivery is failing and use AI-supported scenario planning to assess different courses of action. Decisions made during stocktakes could then be assigned, tracked and evaluated against actual changes in results, creating a continuous feedback loop between policy decisions, implementation and outcomes.


AI is also beginning to change the citizen experience of delivery itself. Abu Dhabi’s TAMM 4.0 has introduced AutoGov to automate routine government transactions, while India’s BHASHINI has supported multilingual access to public services[8][9]. In Alberta, a legacy heating-subsidy application that took five months to hand-code was rebuilt as a working prototype in four days, with a public portal the original lacked[10]. The next step is to connect these capabilities across government, allowing citizens to access services without having to navigate multiple departments themselves.


The real opportunity is not simply to automate existing government processes, but to build a more responsive delivery system: one where government can see what is happening on the ground, coordinate across institutions, anticipate problems and continuously adapt based on results.


Delivery should also be measured by the burden it removes from people: how often they must repeat their story, how many steps they must take, how long they wait and whether the experience preserves their dignity. These are not soft measures; they are part of whether government is working. In Government 3.0, Victor Dominello argues that the public sector should be as accountable for every minute it takes from citizens as it is for every dollar it spends[11]. Imagine a delivery unit that tracks the hours a jobseeker spends on forms and queues with the same rigour it applies to a programme budget, and treats every repeated question as a defect to be fixed.


Ultimately, delivery should be measured not by programmes launched or activities completed, but by the difference it makes in citizens’ lives.


Institutions & Team Building

AI is changing the skills governments need faster than traditional civil service training systems can keep up. Induction programmes, occasional workshops and classroom-based training are no longer sufficient. Governments need to rethink how they build capabilities and continuously upskill hundreds of thousands of public servants.


Some countries are already moving in this direction. Singapore’s Civil Service College has introduced an AI companion to help officials solve complex problems, while India’s iGOT Karmayogi platform is using AI tutors and role-based learning to identify and address skills gaps[12][13]. Alberta’s AI Academy trains public servants in cohorts across every level of seniority, on the published principle that the technology was the easier half of the problem.


The real opportunity is to move from periodic training to continuous, AI-enabled personalised learning embedded in everyday work. Public servants could use AI coaches to solve problems, simulate policy decisions, practise crisis management and develop new capabilities as their roles evolve, curated for each individual. Governments would also need to bring together policy, operational and technical expertise, building multidisciplinary teams capable of translating AI into real improvements in public services.


The goal is not simply to teach public servants to use AI, but to give them more capacity to apply judgement, empathy and expertise where these matter most. Automation must not lead to the atrophy of human skills or the diffusion of accountability: people must remain on the hook for public decisions.


Ultimately, preparing the civil service for the age of AI is about fundamentally changing how governments recruit, develop and equip their people to deliver better results for citizens.


Organising Government for AI

Governments are taking different approaches to organising AI. The UAE appointed the world’s first Minister of State for AI in 2017, placing AI at the Cabinet level. The United States and Australia require Chief AI Officers across federal agencies, while the UK has established a central AI engineering capability. Albania introduced Diella, an AI-powered virtual cabinet minister focused on public procurement. 


Depending on the structure of devolution many countries have provincial level AI authorities. Alberta, Canada, has adopted a subnational model: a central technology ministry overseeing government-wide code, shared AI methods and training, and openly licensed guidance, with public-facing AI assessed and tested case by case before deployment.In February 2026, Punjab established its Office of AI, Pakistan’s first dedicated AI ministry, to drive AI transformation across government[14][15][16][17][18]. With a population of 130 million, including 80 million young people, Punjab has the scale, demographic potential, and ambition to become one of the world’s most significant real-world laboratories for AI-driven transformation—pioneering solutions, generating evidence, and shaping how low- and middle-income countries adopt AI to improve the lives of millions.


The case for a central Office or Ministry of AI is not simply better coordination, but ensuring that AI investments directly support national or subnational priorities. Because AI amplifies the system it finds, much of this work is foundational: connecting data, digital identity and core systems so that intelligent services have something coherent to build on. Rather than allowing individual ministries to pursue disconnected AI initiatives, a central institution can translate government priorities into a coherent AI strategy, identify the highest-impact opportunities and ensure the foundational infrastructure, data systems, policies and capabilities are in place to deliver them. It can also establish common standards for procurement, interoperability and evaluation, ensuring that AI investments across ministries build on one another rather than duplicate efforts, compete for resources or create fragmented systems.


France’s Albert API provides secure access to AI models across government, while Singapore’s AIBots enables public servants to build AI assistants using their own knowledge sources, with adoption across more than 100 government agencies[19][20]. These illustrate how shared infrastructure can accelerate adoption while reducing duplication.


Governments must also develop sovereign AI capabilities suited to their laws, languages and institutional needs, with appropriate safeguards for data security, accountability and human oversight. The goal should be centralised strategic direction, investment coordination and shared capabilities, combined with decentralised implementation and accountability for results.


Trust cannot be bolted on after an AI system is built. It must shape the design, procurement and operation of every public service. Citizens should understand how AI is used, seek human review and challenge decisions that affect them.

A Government That Learns and Acts

Governments have traditionally operated in linear cycles: set priorities, develop policies, implement programmes and evaluate results months or years later. By the time problems are identified, the opportunity to correct course has often passed. AI makes a fundamentally different model possible: a government that continuously learns, adapts and improves based on real-time evidence from citizens, frontline services and implementation results.

We are already seeing early examples. Brazil’s Alice uses AI to identify procurement risks before public funds are spent, while Singapore’s OneService uses AI to direct citizen complaints to the right agencies[21][22]. Yet adoption remains far ahead of impact. According to the OECD, 97% of surveyed countries use AI in government, but only 28% measure its impact[23].


The future of government is not about deploying more AI tools. It is about rethinking how government works: bringing strategy, policymaking, delivery, institutional capabilities and continuous learning together into a more connected and responsive system, organised around national priorities and measurable outcomes for citizens.


Return to Maya. In a government that learns and acts, her first weeks out of work would be spent preparing for her next job rather than chasing offices. Her income support would arrive before the first bill, her training place would be linked to a real vacancy and nobody would ask her to tell her story twice. Those results should be visible in people’s lives: less time lost to forms and hand-offs, more dignity in difficult moments and greater confidence that government will act in their interest.


This is the work Praxis was founded to do: helping governments build the leadership, institutions and capabilities needed to turn ambition into results in the age of AI.


Ultimately, the success of AI in government should be measured not by the technology it adopts, but by the results it delivers for its citizens.

Sources

  1. Government News: launch of the Australian Death Notification Service, August 2020.

  2. ERR News: Estonian parents no longer have to apply for family benefits, 14 October 2019.

  3. UAE Cabinet AI Advisor: approval and rollout, 2 September 2026.

  4. European Commission: Destination Earth and its first digital twins.

  5. ECMWF: AI development within Destination Earth.

  6. UK government: Consult’s Water Commission application, 16 October 2025.

  7. Taiwan Ministry of Digital Affairs: information-integrity assembly and AI facilitation, 2024.

  8. Abu Dhabi Media Office: launch of TAMM 4.0, 14 October 2025.

  9. India Press Information Bureau: BHASHINI at Maha Kumbh, 16 January 2025.

  10. Alberta Technology and Innovation (Velocity White Papers, No. 11): AI Factory case study, from five months to four days, 19 June 2026.

  11. Victor Dominello: Government 3.0: A Vision for the Public Service in the Age of AI, launch release, 9 October 2026.

  12. Singapore Civil Service College: Systems Thinking AI Companion, updated June 2026.

  13. India Press Information Bureau: AI tools for civil-service capacity building, 16 February 2026.

  14. UAE appoints Omar Al Olama as Minister of State for Artificial Intelligence, October 2017.

  15. US Office of Management and Budget memorandum M-25-21 on agency Chief AI Officers, April 2025 (Jones Day summary).

  16. Australian Department of Finance: Establishing Chief AI Officers for the APS, 2025.

  17. UK government: Incubator for Artificial Intelligence, part of the Government Digital Service.

  18. Global Government Forum: Albania’s AI minister Diella, 16 September 2025.

  19. France DINUM: Albert API, the state’s generative AI platform.

  20. Singapore GovTech: AIBots, updated July 2025.

  21. Brazil Controladoria-Geral da União: Alice and reported 2024 activity.

  22. Singapore GovTech: development of the OneService chatbot, 15 July 2021.

  23. OECD: Digital Government Outlook 2026, chapter 4, sections 4.2 and 4.5.7.

 
 
 

Comments


bottom of page