AI can free up significant time in Norway’s public sector – if implemented right

 

The potential time saving from AI in Norway’s public sector amounts to 96,000 full-time equivalents in 2035. Realising a substantial share of that potential will require investment in tailored AI systems across the public sector to begin soon.

 

 

Key takeaways

• We estimate the potential of AI systems to free resources in the public sector at 96,000 full-time equivalents (FTEs) in 2035 – corresponding to an annual wage bill of NOK 83 billion at 2025 wage levels. The potential reflects a scenario where every public employee uses AI systems for every relevant task at full efficiency.
• More than half of this potential lies in municipalities and county authorities: around 53,000 FTEs, valued at NOK 43 billion a year.
• Around 26,000 FTEs of the potential could theoretically be achieved using secure, compliant off-the-shelf AI assistants. Of the 26,000 FTEs, we estimate that around 10,000⁠–⁠15,000 FTEs could be realised cost-effectively, at an annual licence cost of NOK 1.5⁠–⁠3.0 billion.
• Almost three quarters of the potential requires tailored AI systems. Realising much of this potential by 2035 would require systems covering significant parts of the public sector to be scoped, developed, tested, implemented and used efficiently by employees before then.

 

---

 

 

Norway’s 2024 digitalisation strategy sets an ambition for all public bodies to use AI by 2030.¹ In October 2026, the government, employer organisations and trade unions agreed to cooperate on responsible AI adoption in the workplace, emphasising employee involvement, skills development and increased productivity.² This paper asks how much public-sector capacity AI could free for other uses – and what realising that potential would require.

Our approach goes a step further than most of the academic literature. We evaluate not only whether AI systems could technically save time on a task, but also whether that saving is feasible in a public-sector setting, and whether it can be achieved with secure off-the-shelf AI assistants or requires tailored AI systems to safeguard the good administrative practice on which public-sector legitimacy depends. For off-the-shelf AI assistants, we go one step further by estimating how much of this theoretical potential can be realised cost-effectively.

In the coming years, a rising demand for services such as health and care will continue to put pressure on the welfare state, while public services must remain transparent, secure and consistent. With today’s systems and processes, the Norwegian public sector faces growing demand for staff.

Implementing AI systems in the public sector presents a unique opportunity to alleviate this trend, but unlike in the private sector, where competition makes AI adoption self-reinforcing, adoption in the public sector requires deliberate policy action. Early adoption would not only free considerable public-sector capacity to help meet these pressures; it could also shift considerable demand towards AI systems aligned with European values such as democracy and individual freedom.

Our results show a feasible automation potential equivalent to about 96,000 FTEs in 2035, based on assessments of around 13,000 tasks mapped to public employment. Freed time can create value through better public services, greater efficiency or improved job satisfaction (see Figure 1). How the capacity is used is a political and managerial choice.

Almost three quarters of that potential, however, is in tasks where automation requires tailored systems to meet the standards of good administrative practice; about half of the remainder lies in occupations such as childcare and education, where the gains could be difficult to consolidate and redirect.

Potential time savings are often much higher than realistic gains on the ground. We set up four scenarios to give a ball-park figure of the costs and gains of rolling out off-the-shelf AI systems to public employees. Of the 26,000 FTE potential, the scenarios yield between 10,000 and 15,000 FTEs after training, valued at NOK 8.2⁠–⁠12.5 billion a year, with annual licence expenditure of NOK 1.5⁠–⁠3.0 billion.

 

 

Note: Time saved through automation can translate into different outcomes across areas of public-sector work. Depending on the context, it could support higher output, better service quality, lower workloads, the ability to meet future demand with the existing workforce or, in some cases, lower staffing requirements. How the freed capacity is ultimately used is a political and managerial choice.

 

Realising a substantial share of the full potential therefore depends on tailored AI systems being scoped, built, tested and in daily use across large parts of the public sector by 2035 – no small feat. Such systems embed existing AI models within structured workflows that combine them with rules, data and other systems, with individual components or agents responsible for clearly defined steps. They are required not because today’s off-the-shelf AI assistants lack the technical capacity, but because a high level of consistency, transparency and security is required to maintain public-sector legitimacy.

Realising the potential is thus, above all, an investment decision. Norway’s ambition for all public bodies to use AI by 2030 gives added urgency to investing in the systems and skills needed to turn adoption into tangible time savings. The paper first reviews what AI systems are capable of today, and why capability alone is not enough in public administration. It then presents our approach and results, compares them with previous studies, and turns to the costs, infrastructure and sovereignty implications of implementing AI across the public sector at scale before concluding.

 

 

AI in the public sector requires careful implementation

 

The capability of AI models is increasing fast.

Current AI systems hold considerable promise for raising productivity. Across a growing range of tasks usually associated with public-sector employment, AI systems can both make work faster and improve the quality of output, as several randomised controlled experiments have shown. Professionals given an AI assistant for writing tasks finished 40 per cent faster, while quality rose by 18 per cent.³ Customer-support agents resolved 15 per cent more enquiries per hour, with the least experienced staff gaining about 30 per cent.⁴ Another study showed that performance in legal drafting and analysis rose by between 50 and 130 per cent, with the largest gains on more complex tasks.⁵

There is reason to believe that the task-level gains could translate into productivity gains on a larger scale, even if they have yet to show up in macro data. Studies on economy-wide AI exposure generally show that public-sector work contains a high share of automatable tasks. The IMF estimates that around 60 per cent of employment in advanced economies is to some degree exposed to AI.⁶ Both the IMF and the ILO find that administrative and clerical work is particularly exposed,⁷ while the OECD finds that AI models are currently most useful in standardised, codifiable and information-processing tasks.⁸ At the same time, the OECD notes that the gap between current AI capabilities and those required for more complex and specialised tasks is narrowing.

 

Public-sector automation demands more than technical capability

Technical capability, however, is not enough to establish what AI systems can feasibly do in the public sector. As AI systems advance, public institutions will be expected to capture productivity gains. But they must do so while meeting the standards that make public administration lawful and trusted.

For administrative processes, this translates into concrete safeguards. Decisions must rest on documented reasons that citizens can understand and courts can review. Comparable cases must be handled consistently. The path from input to decision must be traceable, and personal citizen data must be handled lawfully and securely. These requirements are reflected in Norwegian guidance on digital administration and data protection.⁹ They are fundamental to legitimate public administration, and this paper assumes that any new AI systems used in public services must adhere to the same standards.

Off-the-shelf AI assistants are highly useful for a range of tasks, but on their own offer limited control over the consistency and predictability of their output. Where public administration requires consistency, transparency and oversight throughout the process, additional safeguards may therefore be necessary. The use of personal citizen data brings further requirements for data protection and security.

Norwegian experience shows why these safeguards matter. In a 2023 review of benefit decisions by Nav, Norway’s labour and welfare administration, Sivilombudet, the Parliamentary Ombud, found that some partly automated refusals of sickness benefits did not adequately explain the facts behind the decision.¹⁰ Separately, in a 2024 pilot, the Norwegian University of Science and Technology (NTNU) tested Microsoft 365 Copilot, an AI assistant that can use documents and other information available to each employee. The pilot highlighted how weaknesses in information management and access permissions could be amplified by the assistant, and NTNU concluded that it was not yet ready for a university-wide rollout, opting instead for controlled introduction in selected roles.¹¹ These examples show that useful technical capabilities must be accompanied by careful implementation.

 

The technical potential of AI systems is a poor measure of what the public sector can achieve

Technical potential captures the task time that current AI capabilities could in principle automate. In this paper we focus on the narrower concept of “feasible potential”: what can be automated while maintaining good administrative practice, which is fundamental to public legitimacy.

For some tasks, the two potentials are close. Summarising documents, drafting text or extracting information can often be supported by a secure off-the-shelf assistant with limited additional infrastructure. Other public processes place much greater demands on consistency, traceability, data protection or integration with existing systems. Here, an off-the-shelf assistant would typically not be enough.

That does not put these tasks beyond automation with AI systems. A task that cannot safely be handed to an off-the-shelf assistant might still be handled by an AI embedded in a system that grounds its output in authoritative data, records the steps taken, constrains what the model can do and keeps personal citizen data within the required environment. This is an example of what we call tailored AI systems in this report.

Nesodden municipality provides a practical example. In June 2026, it introduced Saksassistenten, developed with Netcompany and Ikomm to support building-permit casework. The system helps staff navigate application documents, relevant plans and regulations, while caseworkers retain responsibility for assessments and decisions. The municipality estimates that staff time for a chimney-permit case could fall from four hours to two hours in the conservative case, or to 45 minutes in the optimistic case.¹²

Our results ultimately distinguish between two routes to automation: tasks that can be reached with secure off-the-shelf AI assistants and little additional investment, and tasks that require tailored AI systems to meet public-sector standards. How we measure these two routes is the subject of the next section.

 

 

Our approach – assessing feasibility task by task

 

To estimate the feasible automation potential of the Norwegian public sector, we map assessments of around 13,000 tasks from a US Department of Labor database to Norwegian occupations.¹³ We combine the task-level automation scores with Norwegian public-sector employment data for 2025 and project employment to 2035 using demographic demand, with growth rates for health and care occupations derived from Statistics Norway’s projected demand in its reference scenario. To translate estimated time savings into monetary values, we apply occupation-level wages and add pension and other costs.¹⁴

To do this, each task receives two assessments. First, we evaluate the share of the task that AI systems could currently perform given the right environment, tools and data – and while meeting the specific public-sector requirements discussed in the previous section. Second, we categorise the automatable tasks into two broad groups based on what level of investment would be required for reliable use in a public-sector setting:

• Off-the-shelf AI assistants.
  This category covers tasks that a trained employee could carry out with a compliant and secure off-the-shelf AI assistant, using public information or material that the employee supplies to it. This could include a caseworker turning meeting notes into a draft case record, a civil servant preparing the first draft of a committee paper or a teacher preparing material for a lesson.
• Tailored AI systems.
  This category covers tasks that require properties that off-the-shelf AI assistants do not inherently possess. Here, software is developed for the specific process in question with the aim of maintaining security and good administrative practice. Examples include handling record-based citizen casework, coordinating appointments across systems and processing permits for citizens and enterprises.
  

Figure 2 illustrates the distinction: off-the-shelf AI assistants respond to employees’ prompts, while tailored AI systems combine AI models with rules and data in a structured workflow, with different components or agents performing clearly defined steps. This allows AI systems to be used while meeting requirements for security, consistency and transparency across the process. The software will generally need to be adapted to the authority’s data, systems, procedures or professional safeguards. This may involve configuring and integrating an existing product or developing a purpose-built solution.

Our main estimate applies conservative assessment rules. It credits only capabilities that current AI systems demonstrably possess, while at least maintaining the current quality of public service. We find a conservative estimate much more useful from a policy perspective than speculative estimates of future AI capabilities, given that there is no theoretical limit to what AI systems might eventually be able to do.

The technical appendix provides further detail on the assessment rules, calibration, task weighting, occupational mappings, statistical tables, monetary valuation and demographic projection.

 

 

96,000 FTEs: the feasible potential for freeing public-sector capacity

 

We estimate the feasible automation potential in 2035 at approximately 96,000 FTEs, or 11.2 per cent of the civilian general-government workforce (see figure 3). Almost three quarters of the potential requires tailored AI systems adapted to public-sector data, workflows and safeguards.

The remaining quarter, or 26,000 FTEs, consists of work that can be automated using secure, compliant off-the-shelf AI assistants. About half of this lies in education, childcare and research, where gains are hard to measure and redirect, as we discuss below.

Feasibility does not mean that all this potential will be realised by 2035. In fact, realising the full potential is highly unlikely. Figure 3 presents the full potential along with two illustrative realisation scenarios, where 60 and 80 per cent of the potential is reached.

Realising most of the potential requires tailored systems alongside secure, compliant AI assistants.

 

Time savings vary across occupational groups – and lead to different outcomes

The automation potential spans four broad groups of public employment, of which “Administration” holds the greatest potential for AI automation at 38,000 FTEs, followed by “Health & Social Care” at 26,000, “Education & Childcare” at 20,000 and a residual “Other” group at 12,000 (see figure 4).¹⁵ The interpretation of their contributions differs markedly.

“Administration” consists of occupations such as office clerks, secretaries and legal professionals, which represent around a quarter of public employment and have the highest automatable share of work (see figure 5). The tasks carried out in “Administration”, in particular casework, are often procedural in nature and generally require processes to be auditable and transparent. Around 83 per cent of the potential therefore requires tailored AI systems to maintain good administrative practice.

“Education & Childcare” spans childcare and schooling through to teaching and research at universities. It accounts for around 197,000 public-sector FTEs in 2035, although a smaller share of its work can be automated. Most of the potential lies in preparation and other supporting tasks in education, as well as in research. Many of these tasks do not carry the same documentation and transparency requirements as administrative casework. As a result, the group accounts for around half of the total potential that can be realised with secure, compliant off-the-shelf AI assistants. The resulting time savings are likely to be spread across many employees and supporting tasks, while scheduled commitments to teaching, supervision and care remain.

“Health & Social Care” represents another large share of public employment with relatively low automation potential. But unlike in “Education & Childcare”, automatable work in this area is primarily documentation and administration, governed by strict rules and procedures. Around four fifths of the potential therefore requires tailored AI systems to maintain the quality of service.

Time saved through automation can translate into different outcomes across areas of public-sector work. Depending on the context, it could support higher output, better service quality, lower workloads, the ability to meet future demand with the existing workforce or, in some cases, lower staffing requirements. How the freed capacity is ultimately used is a political and managerial choice.

In “Administration”, automatable tasks account for a relatively large share of work, so the resulting time savings are more likely to add up to meaningful capacity at team or unit level. That capacity could be used to handle more cases, redirected towards other tasks or used to reduce overall FTEs.

In “Health & Social Care”, much of the automatable work consists of documentation and other supporting tasks carried out alongside patient and citizen contact. Time saved on these tasks is unlikely to be large per employee but could help staff spend more time with patients and citizens or fit existing responsibilities more easily into the working day. The gains would probably appear primarily as more face-to-face time with citizens and higher job satisfaction rather than redirected workload or possible staffing cuts.

Source: Own calculations based on O*NET, the US Bureau of Labor Statistics and Statistics Norway. FTEs in 2035. Totals may not sum due to rounding.

 

The same goes for “Education & Childcare”, where time saved on preparation and other supporting tasks is unlikely to allow more students or children to be served or to reduce staffing needs in the short term. The gains are therefore more likely to appear as improved quality than as measurable extra capacity.

Finally, some of the capacity freed by AI will be offset by new tasks and roles created by its adoption. Operating AI systems at scale will increase the need for cybersecurity, compliance, data monitoring and other forms of oversight that also require staff time.

With this in mind, policymakers should be explicit about what they want freed capacity to achieve. A policy meant to reduce public-sector staffing costs is likely to find the largest potential among administrative workers, while policies meant to free up time for citizen contact could focus more broadly. Realising most of the potential requires addressing more than one group and considerable software development beyond rolling out existing AI assistants.

 

The full potential of 96,000 FTEs corresponds to NOK 83 billion a year
The 96,000 FTEs correspond to annual wage expenditure of NOK 83 billion at 2025 wage levels, equivalent to 11.5 per cent of the corresponding civilian general-government wage bill, only slightly higher than the 11.2 per cent of hours automated.¹⁶ Around NOK 61 billion of this potential requires tailored AI systems, with the remainder accessible through secure, compliant off-the-shelf AI assistants. The NOK 83 billion can be interpreted as the cost of achieving the same capacity through hiring at current wage levels.

Measured by the wage bill, the broad employment groups in figure 6 closely mirror the FTE picture in figure 4. The share of “Administration” becomes somewhat larger, as occupations in the group are better paid than average, while the opposite is true for “Education & Childcare”.

Source: Own calculations based on O*NET, the US Bureau of Labor Statistics and Statistics Norway. FTEs in 2035. Totals may not sum due to rounding.

Note: Employment and potential in 2035. Marker area is proportional to automatable FTEs
Source: Own calculations based on O*NET, the US Bureau of Labor Statistics and Statistics Norway.

Source: Own calculations based on O*NET, the US Bureau of Labor Statistics, Statistics Norway and NOU 2026: 5. Employment in 2035, valued at 2025 levels. Totals may not sum due to rounding.

 

Our estimate is a conservative upper bound – realised gains will depend on implementation

The estimated potential of 96,000 FTEs represents our best estimate of the working hours that could feasibly be automated using current technology within current processes. It is an upper bound, as it assumes that AI systems are used 100 per cent effectively for every relevant task across the public sector – an overly optimistic scenario. The scenarios in figure 3 show how much capacity could be released at different levels of adoption and effective use, recognising that implementation will be gradual. The discussion of costs and control gives more explicit scenarios for rolling out off-the-shelf AI assistants.

In another sense, the results are merely a snapshot of a fast-moving target. By the time AI systems have been broadly adopted in the public sector, the technology is likely to be far more advanced than today, and our results might therefore be understated. We also hold the task composition of each occupation fixed in the projection. Given how disruptive AI is, a reorganisation of public services could well reap significant additional benefits on top of the per-task savings we identify.

 

 

Local government can free capacity equivalent to 53,000 FTEs

 

For municipalities and county authorities, the feasible automation potential in 2035 is approximately 53,000 FTEs, or 9.8 per cent of a projected workforce of 539,000 FTEs. The estimates use the national calculation’s methods and assumptions.

Local government accounts for 55 per cent of the national FTE potential. In both sectors, most of the potential requires tailored AI systems. A larger share of the potential can be reached with general-purpose AI assistants in local government. This partly reflects the concentration of education and childcare in municipalities and county authorities, where many preparation and supporting tasks can be handled by these assistants.

 

Administration has the highest automatable share of work

Administrative occupations have the highest automatable share of work at 18.1 per cent, compared with 9.9 per cent in “Education & Childcare”, 7.3 per cent in “Health & Social Care” and 6.9 per cent in “Other” (see figure 8). “Administration” covers administrative and clerical occupations across municipal and county services, regardless of where they work.

Education and childcare have the largest FTE potential, at about 17,000 FTEs. Workforce size therefore matters as much as the automatable share.

 

The potential is valued at NOK 43 billion a year

Valuing the automatable work in municipalities and county authorities at 2025 wages plus pension and other costs gives an annual potential of NOK 43 billion. Higher wages make administration the largest group by monetary value, although education and childcare have the largest FTE potential.

Around 69 per cent of the FTE potential requires tailored AI systems. These systems account for NOK 30 billion of the annual value, with NOK 13 billion accessible through secure, compliant general-purpose AI assistants. These amounts represent the value of staff time that could be freed, measured at 2025 wages plus pension and other costs.

Note: Central government includes public hospitals; local government comprises municipalities and county authorities. Employment and potential in 2035. Totals may not sum due to rounding.
Source: Own calculations based on O*NET, the US Bureau of Labor Statistics and Statistics Norway.

Note: Local-government version of Figure 5. Employment and potential in 2035. Marker area is proportional to automatable FTEs.
Source: Own calculations based on O*NET, the US Bureau of Labor Statistics and Statistics Norway.

Note: Local-government version of Figure 6. Employment in 2035, valued at 2025 wages plus pension and other costs. Totals may not sum due to rounding.
Source: Own calculations based on O*NET, the US Bureau of Labor Statistics, Statistics Norway and NOU 2026: 5.

 

Our feasibility criterion places our estimate below most previous estimates

 

The studies differ in sector coverage, assumptions and what they measure, so their estimates are not directly comparable.

Menon Economics (2023) estimates that full use of the language models assessed in 2023 could free the equivalent of 155,000 FTEs in public administration, education and health and social services, nearly 18 per cent of employment in those industries, private providers included.¹⁷

Implement Consulting Group (2025) estimates that widespread AI adoption could yield an annual productivity gain valued at NOK 26 billion in public administration, defence and social security, equivalent to 10 per cent of gross value added in these activities. This represents the potential to deliver more with the same resources. For non-market public services, gross value added is measured mainly by staff costs and depreciation.¹⁸ Its broader 2024 study estimates an annual gain of around NOK 100 billion at widespread adoption across a larger group of service industries, including education, health and social services and private providers.¹⁹

The Tony Blair Institute’s 2024 UK study is the closest in design, as it also divides the potential by the technology required. Regrouping its categories, free or low-cost tools account for about 55 per cent of its 20 per cent time saving and bespoke systems, sensory hardware and high-cost equipment for about 45 per cent, against almost three quarters requiring tailored AI systems in our estimate.²⁰

 

 

Using AI in public processes raises questions of cost and control

 

This section turns to two questions that any government implementing AI at this scale should ask: What does it cost, and what does it mean for control and sovereignty?

We examine the former using illustrative scenarios for rolling out off-the-shelf AI assistants. Their relatively standardised nature makes a simple licence-and-training calculation possible. Tailored AI systems, by contrast, depend on the specific processes, data and systems in which they are deployed. Costing them would require a comprehensive mapping of public-sector processes and is therefore beyond the scope of this analysis.

The latter question – realising the productivity gains while retaining sufficient control over critical functions, data and infrastructure – is a complex matter. A comprehensive analysis of the strategic implications of these dependencies is beyond the scope of this paper, but policymakers should recognise that laying the foundation of future public processes in the age of AI is deeply intertwined with geopolitics and questions of control and sovereignty.

 

Optimal investment in off-the-shelf AI assistants: Four scenarios

To move from theoretical potential to cost-effective investment in off-the-shelf AI assistants, we need the number of users for whom a licence pays off and its annual cost. Our occupation-level data provide time-saving potential and labour costs. We combine these with assumptions about licence prices, effective use and training.

We set up four scenarios along the axes of licence price and efficiency of use. We assume an annual licence cost of between NOK 3,000 and NOK 6,000, at 2025 prices. For effective use, we assume 80 per cent of the off-the-shelf potential in the generous case and 60 per cent in the conservative case.

For all scenarios we assume that users need two days a year to learn how to use the systems, corresponding to about 20 minutes a week, covering both on-the-job training and formal courses.²¹

We include an occupation-by-subsector group when the value of its gross time savings exceeds licence expenditure plus the value of training time. Figure 10 shows the resulting four combinations of annual licence expenditure and net time savings. The shaded area spans the four scenarios.

Of the approximately 26,000 FTEs theoretically accessible through secure, compliant off-the-shelf AI assistants, the scenarios yield around 10,000–15,000 FTEs after training. The time released is valued at NOK 8.2–12.5 billion a year, before annual licence expenditure of NOK 1.5–3.0 billion. Additional organisational and technical implementation costs are not included.

Note: 2035 employment and 2025 prices. Labels show annual licence cost and effective use. Time savings are net of training; expenditure covers licenses. One licence per FTE is assumed.
Source: Own calculations using the preferred employment, automation and labour-cost estimates.

 

The four scenarios suggest that off-the-shelf assistants can free staff time worth substantially more than their licence and training costs, even though they realise only a modest share of the total 96,000-FTE potential. Realising substantially more entails tailored systems becoming part of daily public-sector processes by 2035.

 

AI in critical public processes creates sovereignty dilemmas

Operational resilience requires that critical public functions do not depend on technologies, models or compute that external actors can unilaterally restrict or withdraw; data sovereignty, that governments retain control over sensitive public data; and autonomy, that they can switch technologies or providers when circumstances change.

Dependencies take several forms. Systems hosted outside Europe may be exposed to decisions by foreign governments, and even infrastructure in Europe may be operated by companies under foreign jurisdiction. Limited competition and vendor lock-in can make switching costly, and access to the models themselves can be restricted, as the temporary US export controls on Anthropic’s leading models in June 2026 illustrate.²²

Control over advanced chips, computing power and AI models is increasingly intertwined with economic and foreign policy, which makes these dependencies geopolitical: a commercial dependency can become a strategic vulnerability when regulation or international relations change. The safeguards discussed in this paper must therefore be accompanied by scrutiny of the infrastructure underlying AI systems.

European providers hold only around 15 per cent of the European cloud market, and Europe depends on foreign providers of leading AI models and advanced chips.²³ Expanding Europe’s options requires substantial investment in computing, data-centre and energy capacity; growing public-sector demand could help create a market for European alternatives.

Such capacity will take time to build and will not remove the need for global technology. Governments must therefore continue to assess where control is essential and where dependencies can be accepted and managed – diversifying providers and keeping models and infrastructure substitutable where possible. Infrastructure, access and control will thus be central to how Norway captures the productivity gains while maintaining a legitimate, trusted and resilient public sector.

 

 

Norway can free substantial capacity – if it builds for it

 

Norway’s welfare state faces growing demand for its services. Our results suggest that AI systems can help meet these pressures: a feasible potential of 96,000 FTEs in 2035, valued at NOK 83 billion a year in 2025 wages plus pension and other costs. Capturing a substantial share requires broad implementation of what is already feasible with today’s technology under today’s administrative standards.

It also requires the right kind of effort. Almost three quarters of the potential depends on tailored AI systems built to the standards that make public administration lawful and trusted. While the remainder is technically reachable with off-the-shelf assistants, going from theoretical potential to realised gains reduces this further. Our illustrative scenarios yield 10,000⁠–⁠15,000 FTEs after training, valued at NOK 8.2⁠–⁠12.5 billion a year, with annual licence expenditure of NOK 1.5⁠–⁠3.0 billion.

Most of the gains from off-the-shelf AI assistants are in education, research and health, where time saved is inherently harder for management and policymakers to consolidate and redirect. Policymakers should therefore be explicit about what they want freed capacity to achieve. Redirecting capacity to new tasks or reducing staffing needs may point to automating administration, whereas more face-to-face time with citizens could also be gained in education, care or health. Either way, realising a substantial share of the potential requires broad implementation, with the necessary systems scoped, built and in daily use well before 2035.

Finally, because these systems will sit inside critical public processes, control over the infrastructure beneath them becomes part of the policy decision. Overreliance on vendors outside Europe puts Norway at a strategic disadvantage and creates new geopolitical pressure points. Public-sector investment in Norwegian computing infrastructure and AI capabilities can reduce that dependence and give Norway greater control over the technology underpinning its public services.

Welfare states are under pressure everywhere, and Norway’s is no exception. AI technology offers a rare chance to shrink the time absorbed by administration while preserving the quality and legitimacy of public services. Seizing it requires no technological breakthrough – only broad and sensible implementation, in which AI systems are made to meet the standards of good administrative practice rather than allowed to redefine them.

 

---

 

 

Endnotes

 

[1] Norwegian Ministry of Digitalisation and Public Governance, national digitalisation strategy 2024⁠–⁠2030, “Utnytte mulighetene i kunstig intelligens”, 26 September 2024.

[2] Norwegian Ministry of Digitalisation and Public Governance and the Prime Minister’s Office, “Regjeringen og partene i arbeidslivet inngår KI-avtale”, press release, 2 October 2026.

[3] Shakked Noy and Whitney Zhang (2023), “Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence”, Science 381(6654), pp. 187⁠–⁠192.

[4] Erik Brynjolfsson, Danielle Li and Lindsey Raymond (2025), “Generative AI at Work”, Quarterly Journal of Economics 140(2), pp. 889⁠–⁠942.

[5] Daniel Schwarcz, Sam Manning, J.J. Prescott, Patrick Barry, David R. Cleveland and Beverly Rich (2026), “AI-Powered Lawyering: AI Reasoning Models, Retrieval Augmented Generation, and the Future of Legal Practice”, Journal of Law and Empirical Analysis 3(1), pp. 220⁠–⁠250.

[6] IMF (2024), Gen-AI: Artificial Intelligence and the Future of Work, Staff Discussion Note SDN/2024/001, January 2024.

[7] Paweł Gmyrek, Janine Berg and David Bescond (2023), Generative AI and Jobs: A Global Analysis of Potential Effects on Job Quantity and Quality, ILO Working Paper 96, August 2023.

[8] OECD (2026), The OECD AI Exposure Measure: Mapping the OECD AI Capability Indicators to Occupations.

[9] Sivilombudet, Digital forvaltning; Datatilsynet, Å lykkes med åpenhet: Rettslige krav til åpenhet.

[10] Sivilombudet, Begrunnelser i Navs vedtak, 21 September 2023 (SOM-2022-5350), main findings. The review concerned automated case handling, not generative AI.

[11] Datatilsynet, NTNU: Copilot med personvernbriller på, November 2024, updated October 2025.

[12] Nesodden kommune, Nesodden lanserer skreddersydd KI-plattform for saksbehandling, 23 June 2026; Arne Kongsnes, Nesodden og KI: Halverer saksbehandlingstiden og sparer millioner, Plan, 6 August 2026

[13] US Department of Labor, O*NET Resource Center (2023), O*NET Database, release 27.2

[14] Authors’ calculations using SSB tables 11418, 11655 and 11658 for 2025, SSB’s population projection and Reports 2026/18, table 6.1. See the technical appendix.

[15] “Administration” covers administrative and clerical occupations across the public sector – office clerks, secretaries, economists, lawyers, HR staff and similar – regardless of the subsector in which they work.

[16] SSB table 11418, monthly wages for 2025 multiplied by 12, with pension and other costs estimated from the subsector cost ratios in NOU 2026: 5, Grunnlaget for inntektsoppgjørene 2026, Norwegian Ministry of Labour and Social Inclusion, appendix 6, table 6.5. See the technical appendix.

[17] Trygve Svalheim, John Oskar Skjeldrum, Sebastian Winther-Larsen, Kristoffer Midttømme and Jonas Erraia, KI: Betydning for arbeidsstyrken, Menon-publikasjon nr. 127/2023.

[18] Implement Consulting Group, The AI opportunity for eGovernment, commissioned by Google, July 2025. See pp. 7 and 51 for the productivity estimate and method. For the national-accounts definition, see SSB, Concepts and definitions in national accounts, “Value added”.

[19] Implement Consulting Group, The economic opportunity of AI in Norway, commissioned by Google, May 2024, p. 16. The grouping covers NACE O–R and U, including arts, entertainment and recreation. The estimate is expressed at 2023 economic levels on an approximately ten-year adoption horizon.

[20] Isabel Atkinson and James Browne (2024), The Potential Impact of AI on the Public-Sector Workforce, Tony Blair Institute for Global Change, July 2024.

[21] Calculations assume a 220-day working year and one licence per FTE. Part-time staff may require more licenses per FTE.

[22] Anthropic, Redeploying Fable 5, 30 June 2026, updated 1 July 2026.

[23] Synergy Research Group, European Cloud Providers’ Local Market Share Now Holds Steady at 15%, 24 July 2025.