AI and gen AI offer Romania’s public sector a transformative opportunity to enhance efficiency, improve services, and address labor shortages. Adoption of gen AI could boost the GDP of countries around the world by 0.6 percent annually, on average, adding a cumulative €30 billion to €50 billion to Romania’s GDP by 2040. Such adoption could also deliver annual public sector productivity gains of up to €1 billion, together with additional indirect benefits (such as facilitating the extension of digital and AI literacy among the Romanian population).
However, Romania’s Strategia Națională în domeniul Inteligenței Artificiale [National Strategy in the Field of Artificial Intelligence] (SN-IA) lists a number of challenges that hinder progress, including a lack of digital skills, a shortage of specialists, and infrastructure gaps. As of 2024, only 25 percent of Romanians use e-government services (compared with an EU average of 75 percent). Despite this, initiatives such as establishing the Authority for Digitalization of Romania (ADR) signal important progress and are reflected in Romania’s growing performance in the Oxford Government AI Readiness Index.
Global success stories from Singapore, the United Kingdom, and the United States prove the benefits of leveraging gen AI to modernize public administration. But to capitalize on gen AI, other countries, including Romania, need to take action on two levels: increasing their national preparedness and developing action plans for each ministry or agency that intends to adopt the technology. Both actions include investing in infrastructure, developing AI talent, and fostering trust. Increasing the national readiness for adopting gen AI in public administration and implementing the technology at scale hold a clear promise to unlock significant economic and administrative benefits.
The transformative potential of AI and gen AI in the Romanian public administration
AI and gen AI hold tremendous potential for both the public and private sectors. Gen AI has become a focal point of intense global discourse (see sidebar “What is gen AI?”). Once regarded as a futuristic innovation, it has swiftly transformed into a widely adopted and increasingly indispensable tool. According to the McKinsey Global Institute’s (MGI’s) 2024 report A new future of work: The race to deploy AI and raise skills in Europe and beyond, the integration of gen AI could automate up to 27 percent of work hours across all occupational categories in Europe by 2030.1 The European Union plans to adopt a Data Union Strategy, ultimately creating a single data market to accelerate AI innovation and help Europe succeed in the global AI race. This strategy will build on important existing pillars, such as the Open Data Directive, Data Governance Act, Data Act, and AI Act. While the private sector has propelled much of its adoption, gen AI holds immense potential to revolutionize the public sector as well, enhancing efficiency, service delivery, and decision-making processes.
Acting now to adopt gen AI provides governments with a critical window to address technical, regulatory, and workforce challenges while driving significant productivity gains. Amid ongoing technological innovation and, particularly in the European Union, a growing shortage of skilled workers, gen AI presents a wealth of opportunities to significantly enhance the productivity and value creation of economies, businesses, and organizations. According to estimates by MGI,2 the accelerated adoption and widespread use of gen AI could boost countries’ GDP by 0.6 percent annually, on average. For Romania, this impact translates to a total increase of approximately €30 billion to €50 billion by 2040,3 positioning it to thrive in an increasingly competitive global economy.
Gen AI can support public administration through extensive automation
On the journey toward digitalizing Romania’s public sector, previous digital technologies have already made a notable impact in reducing workloads. For instance, the creation of the ADR in 2020 contributed to the development and implementation of strategies and policies for digital transformation across the country, including efforts related to AI. The digitalization of life events in 2021, such as digital birth registration, also added to the seamless delivery of government services during critical moments in citizens’ lives. Moreover, in 2024, the Romanian government approved the SN-IA, signaling a growing eagerness to embrace and benefit from innovation.
Despite these advancements, significant opportunities for optimization and automation within Romania’s public administration remain untapped. According to the SN-IA, this is primarily due to a range of technical, legal, and financial challenges, including low levels of basic digital skills among the population, a shortage of specialists, and infrastructure limitations.4 Consequently, in 2024, the Digital Economy and Society Index (DESI) ranked Romania last among EU member states in the adoption of digital public services, with only 25 percent of the population using e-government services—well below the EU average of 75 percent.5
Nevertheless, automating citizen services is only one part of the equation. For public administration, AI and gen AI offer tremendous potential to optimize processes and thus more efficiently execute tasks, ultimately improving overall productivity. Across industries, the adoption of gen AI has been shown to drive productivity gains ranging from 0.5 percent to 9.0 percent of total industry revenues.6 For the public sector specifically, gen AI’s productivity impact is estimated to be up to about 1 percent of government revenue. If translated to Romania’s steadily increasing government revenues, which reached €120 billion in 2024,7 this presents a substantial opportunity of up to €1 billion annually in the next few years. Nevertheless, with the increase of gen AI use cases, this figure is expected to grow substantially in the future, underscoring the long-term transformative potential of this technology to drive efficiency and cost-effectiveness in public administration.
Part of those productivity gains are grounded in the additional automation potential that gen AI has unlocked across a wide range of tasks. Previously, only an estimated 15 to 25 percent of tasks requiring complex expertise, such as decision-making and collaboration, were considered automatable. However, our recent studies indicate that with the introduction of gen AI, this figure has surged to as much as 45 to 60 percent. The impact is even more pronounced in administrative functions, such as data management, where the automation potential could reach an impressive 80 to 90 percent in certain scenarios. These advancements highlight the revolutionary capabilities of gen AI in reshaping how work is performed, particularly in areas traditionally reliant on human input (Exhibit 1).8
In the public sector, gen AI could facilitate the automation of complex tasks that demand specific outputs, such as generating draft decisions based on application data, which can then be reviewed and approved by administrative staff. This automation would not only boost overall productivity but also alleviate employee workloads, enabling employees to concentrate on higher-value-adding responsibilities. In Romania, a large share of public sector employees are well aware of the opportunities of AI and gen AI. When consulted about the key benefits that AI can bring for their roles, Romanian public sector representatives identified the following9:
- improving the quality of public services provided to citizens
- optimizing process time
- enabling evidence-based decision-making
- enhancing organizational creativity
- increasing transparency and simplification
These improvements in quality and efficiency fully align with the gen AI use cases and initiatives that are steadily being adopted by public administrations worldwide.
However, it is important to emphasize that in terms of decision-making, the final decision should and will remain with the human individual. Hence, AI and gen AI models must be constructed in a way that allows humans to oversee processes and outputs to ensure quality control—that is, a “human in the loop” approach.
The first gen AI use cases have emerged in public administration
There already exists a wide range of potential use cases for gen AI in public administration. Several national and international success stories illustrate the potential impact of such use cases, as anticipated by Romanian public sector employees.
Citizen-oriented communication with chatbots
Gen AI offers numerous opportunities to enhance citizen interactions through personalized and automated communication. For example, gen AI–powered chatbots can provide 24/7 customer service, answering citizens’ inquiries in a personalized and automated manner. These chatbots can assist citizens in filling out forms, provide general information, or guide citizens through personalized customer journeys to improve their service experience. By handling routine inquiries and tasks, gen AI significantly reduces the workload for public administration employees.
- Germany. In the public administration of Heidelberg, the citizen assistant Lumi—a chatbot that answers citizens’ and visitors’ questions about the city—went live as part of a pilot project. Lumi goes beyond a simple keyword search and instead simulates human conversation. Since the beginning of the test phase, Lumi has had more than 4,500 conversations and answered more than 21,000 questions, including about basic administrative services (for example, the renewal of an ID card or the registration of a dog).
- Singapore. Singapore, a global leader in digitalization, introduced an AI tool called Virtual Intelligent Chat Assistant (VICA) in 2023 to handle citizen inquiries. VICA uses hybrid AI, which combines natural language processing and gen AI to balance automation and accuracy. Singapore has had remarkable results with the tool: More than 60 Singapore government agencies use VICA; combined, these agencies host more than 100 chatbots. Internal government agency use accounts for one-fifth of the chatbots, while the rest are used by the public. VICA chatbots receive more than 800,000 monthly queries, on average.10 This has led to significant reduction in call volumes, with Singaporean authorities reporting the lowest call volumes in the past five years.
Content summaries and context-sensitive automation
Gen AI has the capability to process various data sources, including structured data such as tables and unstructured data such as text, images, and audio files. It can summarize content, prepare it for analysis, and pre-interpret it. By consolidating large source documents, gen AI can significantly reduce the workload for public administration employees. Additionally, the automated interpretation of data—such as application data—enables the execution of tasks that arise from these data changes. For example, when application data is updated, gen AI can automatically process the changes. This context-sensitive automation can also enhance citizen interactions, such as by automatically notifying individuals about events relevant to them.
- United Kingdom. The UK Pensions Authority employs a gen AI model to automatically process amendments for pensioners. This automation eliminates more than 230,000 hours of repetitive manual reviews annually, freeing employees to focus on more-complex and demanding tasks.
Creative generation of new content
Gen AI tools are capable of generating new content across various formats, including text, images, audio, and code. These tools can create translations, educational materials for citizen engagement, and context-specific content tailored to different audiences and languages. Gen AI can also generate dashboards, draft job advertisements, formulate interview questions, and create training materials based on predefined frameworks and the specific needs of public authorities.
For example, documents that previously required manual adaptation can now be automatically drafted by gen AI, at least in the initial stages, streamlining the process.
- United States. The Pentagon’s Chief Digital and Artificial Intelligence Office (CDAO) has developed a gen AI–powered tool called AcqBot for the US Department of Defense. AcqBot automates the creation of procurement contracts, formulates problem specifications, and optimizes contract processes. While experienced staff still perform final reviews and validations, the tool has significantly accelerated procurement processes within the CDAO.11
Romania’s readiness to adopt gen AI is growing
As seen from the use cases, there is significant potential for gen AI to enhance efficiency and productivity in Romania’s public sector, but its successful implementation hinges on several critical factors. Oxford Insights’ Government AI Readiness Index report sheds light on this, exploring the question “How prepared are governments to integrate AI into public service delivery?” The index evaluates countries based on more than 40 indicators across three key pillars—government, technology sector, and data and infrastructure—providing a holistic view of AI readiness. In the 2024 report, Romania ranked 13th in Eastern Europe and 58th globally out of 189 countries, with a readiness score of 58.1 out of 100.12 Notably, Romania’s AI readiness score has grown at a faster pace than the scores of many of its Eastern European peers between 2021 and 2024, underscoring the country’s upward trajectory in this area (Exhibit 2).
While there is always room for improvement, Romania’s steady progress, particularly in the government dimension, highlights a growing commitment to fostering the conditions needed for AI-driven transformation in the public sector.
Similar conclusions could be drawn from the recommendations given by the reports from OECD13 and European Commission14 discussing digitalization of public administration. While they mention progress in some areas—improved governance of digital initiatives through the Authority for the Digitalization of Romania (OECD) to adoption of necessary legislation and improved implementation of eID and government cloud (European Commission)—there are still areas that remain to be tackled, including the following:
- persisting deficits in relevant digital skills
- comparably low adoption and user-driven design of public services
- fragmentation and limited interoperability as well as prolonged implementation of some critical infrastructure projects (for example, for better interoperability and wider adoption of cloud services)
Nevertheless, country-level readiness is only one of the prerequisties for adopting gen AI in public administration. While the strategic steering comes from the government, the practical work in this direction is executed by individual ministries and agencies, which makes them a key stakeholder in the process. Therefore, the remainder of this article adopts a dual lens to explore what is needed for gen AI implementation in the public sector: improving national readiness at the country level and taking practical organizational steps at ministry and agency level.
Improving national readiness for automation and gen AI adoption
As the global adoption of gen AI accelerates, countries around the world, including Romania, will begin to improve their national AI readiness. While pilot projects and proof-of-concept initiatives have already demonstrated the promise of gen AI across various countries, the time has come for the next step: moving beyond isolated experiments and embracing a coordinated, nationwide approach to scaling AI-driven solutions. There are several aspects that countries need to consider if they want to tap into the AI and gen AI potential on a national level, and Romania is already making steady progress.
Establishing a national AI vision and strategy
The stepping stone to a successful national push for gen AI is a clear vision and strategy that aligns with the broader national digital-transformation goals. This strategy should set measurable objectives, such as improving citizen satisfaction, unlocking productivity gains, increasing speed of service delivery, reducing workloads, uplifting the usage of digital public services, identifying high-impact areas, and exploring ethical and regulatory frameworks. Romania already reached a crucial milestone in this direction with the SN-IA, which was approved by the government in 2024. The strategy explores the benefits of integrating AI in both the private and the public sector while also assessing the boundaries and associated risks.
Centralizing AI and gen AI development efforts for efficiency and scale
To maximize the impact of AI, countries would benefit from adopting a centralized approach to deploying and managing AI development efforts across public administration. A centralized AI hub can help coordinate and pool resources, expertise, and infrastructure. While use case delivery should remain “close to the business” within the respective areas of administration, a centralized AI hub can enable smaller agencies to access gen AI capabilities with smaller investments and without duplicating efforts. The establishment of the ADR in 2020 has significantly contributed to centralized governance and the execution of digitalization initiatives across the country, including efforts related to AI. In the future, the ADR could become a host and driver for the creation of a centralized AI hub. On the other hand, the use cases themselves can and should be developed by the offices or agencies that own the corresponding process or value chain to ensure appropriate customization and buy-in. The centralized approach and technological or expertise backbone is needed to significantly enhance speed and capabilities.
Investing in infrastructure and data readiness
Scaling AI requires robust technical infrastructure and high-quality data. Countries that plan to adopt gen AI at scale need to consider establishing a national data platform to manage government-owned data sets, ensuring data quality, availability, and security. In fact, the SN-IA sets forth scalable IT infrastructure as one of its core objectives, stating that lack of funding, framework, and legislation are among the main challenges to building it.
Building a skilled workforce for AI and gen AI
A national workforce strategy is critical to ensure that public sector employees are prepared to adopt and use gen AI effectively. Partnerships with universities and private sector organizations can help develop AI-focused training programs and internships, while national upskilling initiatives could provide public sector employees with foundational skills. In this regard, there is huge potential in Romania’s future workforce, given that nearly 7 percent of university graduates specialize in information and communications technology, ranking Romania sixth within the European Union on this indicator.15
Promoting collaboration and public trust
Collaboration and public trust are essential for scaling gen AI at the national level. Interagency collaboration can be encouraged through shared platforms and knowledge repositories, while public–private partnerships can leverage expertise and resources from the private sector. An EU study16 found that in 2021, 59 percent of the Romanian public was positive about the changes that AI and automation could bring, and that number could be further increased through open communication and engagement in the digital transformation process as well as by adopting a citizen-centric design of public services.
Securing funding and policy support
A large-scale push for AI and gen AI requires sustained investment and policy support. Governments need dedicated funding for AI initiatives, streamlined approval processes for projects, and advocacy for EU-level support to access funding and expertise.
In sum, scaling AI at a national level is not a simple task—it requires strategic vision, coordinated efforts, and sustained investment. By addressing these key enablers and adopting a centralized, collaborative approach, countries (including Romania) can make significant strides in modernizing their public administration and improving citizen services. While this is not an easy endeavor, Romania is in good standing: the SN-IA concludes that the country is ready for a large-scale adoption of AI in most fields of activity.
A practical guide for drawing public sector organizations toward AI adoption
As noted above, improving national readiness to adopt AI is the stepping stone of such efforts; however, the next step is the actual transformation. After an AI strategy is defined, including a vision for infrastructure, funding, and talent, it is up to individual public sector organizations, ministries, and agencies to operationalize this strategy and execute it. Therefore, the second level of gen AI adoption after improving national readiness lies at the organizational level and requires public sector institutions to take specific, tangible actions to integrate gen AI into their operations. The following five key actions provide a practical guide for public sector organizations to successfully adopt and scale gen AI.
Create a clear implementation plan
Public sector organizations need to outline detailed plans for integrating gen AI into their operations. This includes identifying and prioritizing specific processes or services that can benefit most from automation or AI-driven insights. Instead of focusing on isolated tasks, they should think in terms of domains to optimize entire workflows and maximize impact. They should pick one domain and assess multiple use cases and technologies to reimagine that domain; only then should they prioritize the next organizational unit.
Train employees, and hire talent for AI and gen AI adoption
In Romania only 1.5 percent of businesses use AI, compared with an EU average of 8.0 percent.17 Therefore, to ensure successful implementation in the public sector, organizations must equip their employees with the skills needed to work effectively with gen AI tools. This includes offering hands-on training for specific roles such as securely handling data, interpreting AI-generated outputs, and using AI tools to improve decision-making. Additionally, organizations should actively recruit talent with expertise in AI and data science to complement internal upskilling efforts and to bring specialized knowledge to accelerate gen AI adoption. Nevertheless, our research suggests that people are already using AI to a much greater extent than their employers assume, which means that employees are ready for the AI revolution—they just need to be given the right tools.
Build a strong data and technology foundation
Public organizations must ensure that their data is ready for gen AI applications. This involves cleaning and standardizing data sets, investing in data management tools, and ensuring compliance with data protection regulations. Additionally, IT systems should be upgraded to support gen AI, with a focus on integrating AI tools into existing technologies securely and seamlessly. When adjusting the IT architecture, use an agile approach to ensure that the necessary foundations are in place and will be gradually improved, as opposed to investing years into perfecting the tech stack and falling behind with the actual implementation of AI. Based on international examples, the most beneficial approach to such a transformation is one that focuses on use cases, thereby avoiding an otherwise impossibly broad task (for example, attempting to clean the entirety of the data estate). The key is to identify the priority use cases and use case domains and to run focused transformations, enabling targeted impact through AI and gen AI.
Adapt organizational structures and processes
Public organizations must work closely with data protection officers to ensure compliance with regulations such as the General Data Protection Regulation in the European Union. Governance frameworks should be established to guide the responsible use of gen AI, including processes for approving AI models and validating data inputs. Lessons from pilot projects can be used to refine risk management practices. Agencies should actively collaborate with one another and the government when creating an organizational AI framework, and they should work together to mitigate common risks.
Communicate benefits, and build trust
Effective communication is critical to fostering a culture that embraces gen AI. Organizations should clearly articulate the benefits of AI adoption to employees, stakeholders, and citizens. Highlighting success stories and using tools such as gamification (for instance, certifications or recognition programs) can help build enthusiasm for and trust in the technology. People already use easily accessible gen AI solutions on a daily basis to optimize their productivity; it is simply necessary, then, to translate that broader enthusiasm into the context of the public sector.
By taking these five actions, public sector organizations can effectively integrate gen AI into their operations, delivering tangible benefits to both employees and citizens. When combined with national-level orchestration, these efforts will help tap into the full potential of gen AI to modernize Romania’s public administration (for more insights on gen AI in the public sector, see sidebar “Further insights”).
When deployed strategically and responsibly, AI and gen AI have the power to transform public administration by enhancing efficiency, reducing costs, and improving the quality of services. Such adoption represents a critical step toward strengthening states’ ability to act while delivering services that are more citizen-focused and business-friendly. By combining national-level orchestration with a structured organizational approach, Romania can unlock the full potential of AI to modernize its public sector and meet the evolving needs of its people.