“The development and use of AI has permeated public service design and delivery in recent years, with 67% of OECD countries using AI to improve this function” concludes the recent OECD study Governing with Artificial Intelligence (OECD, 2025). From drafting policy briefs and analysing large volumes of data to answering citizens’ questions and supporting frontline staff, AI systems are becoming embedded in everyday knowledge work. This transformation also has the potential to alleviate the shortage of skilled professionals and understaffing in public service, as argued by the McKinsey survey on Empowering People to Unlock AI’s Full Potential.
AI introduces powerful new capabilities into this environment, but it also raises profound organisational, ethical, and governance challenges. Understanding how to blend human and artificial intelligence effectively is therefore essential for modern public administration.
Implications for managing professionals in the age of AI
As AI becomes embedded in daily work, managing knowledge work is no longer only about allocating tasks and ensuring compliance with procedures; it increasingly involves orchestrating a productive relationship between human expertise and algorithmic support (North and Kumta, 2025).
One key implication concerns job design and role clarity. AI takes over parts of information processing, drafting, and analysis, which can blur boundaries between junior and senior roles. Managers need to explicitly redefine what constitutes value-added human work: interpretation, ethical judgement, stakeholder engagement, and responsibility for decisions. Without this clarity, staff may feel deskilled or uncertain about expectations.
A second implication is performance management. Traditional metrics such as output volume or turnaround time may become misleading when AI accelerates routine tasks. In assessing and managing performance of professionals, greater emphasis will have to be put on the quality of judgement, the ability to challenge AI outputs, and the responsible use of tools. Evaluating how staff collaborate with AI systems – rather than how much they produce – becomes essential.
Third, learning and professional development take on a heightened importance. AI changes workflows faster than formal training programmes can adapt. Managers must promote continuous learning through peer exchange, communities of practice, and safe spaces for experimentation. Encouraging staff to share both successes and failures with AI tools helps build collective intelligence and reduces individual risk-taking.
Fourth, trust and psychological safety are central management concerns. Knowledge workers must feel empowered to question AI recommendations without fear of appearing inefficient or resistant to innovation. Managers play a critical role in signalling that critical thinking is expected, not discouraged, and that responsibility ultimately rests with humans.
Finally, leadership and culture matter more than ever. Managers who treat AI as a purely technical add-on risk undermining morale and professional identity. Those who frame AI as an augmenting tool – one that supports public service values and professional judgement – are better positioned to retain talent and legitimacy.
Experiment first, scale up later: learning to make AI work in the public sector
Successful use of AI in public sector knowledge work is less about deploying cutting-edge technology and more about thoughtful integration into organisational practices. Public administrations are realising that AI works best when they approach it as a learning process – trying things out, experimenting, and building trust – rather than just focusing on the technology itself. The JRC report on Artificial Intelligence for the Public Sector concludes that AI in the public sector works best when public organisations are willing to “experiment first, scale up later”, allowing controlled pilots and sandboxes to test ideas and discover what actually delivers results. Making AI work well also depends on involving multiple stakeholders, from policy experts and technical teams inside government to civil society and private sector innovators, so that solutions are trustworthy, ethical, and aligned with public values. Equally important is upskilling staff so they can interpret AI outputs and make informed decisions.
Towards a balanced future of public sector knowledge work
AI will not replace public sector professionals, but it will change what it means to be one. Analytical skills, ethical reasoning, collaboration, and the ability to interpret and contextualise information will become even more important. AI can augment these capabilities, but only if introduced thoughtfully.
The real transformation lies in creating socio-technical systems where humans and AI learn from each other. This requires leadership, governance, and a culture that values reflection as much as innovation.
For a balanced approach to public sector value creation by blending human and artificial intelligence, consider the following recommendations:
- Start with public value, not technology. Define clear, mission-driven use cases for AI in knowledge work.
- Keep humans accountable. Ensure human-in-the-loop designs and clear responsibility for decisions.
- Invest in AI literacy. Equip public servants with the skills to critically and confidently use AI tools.
- Build strong data and ethics governance. Address data quality, bias, transparency, and compliance from the outset.
- Avoid lock-in and over-automation. Maintain strategic autonomy and protect areas where human judgement is essential.
EIPA’s Seminar – Knowledge Management in Public Sector Organisations: Embracing Digitalisation and AI, 8-9 June 2026
The discussions above offer a snapshot of the questions public administrations are grappling with in relation to how AI affects the creation, sharing, and use of knowledge. Building on these reflections, EIPA will host a seminar on Knowledge Management in Public Sector Organisations that puts these themes into practice. Participants will work with real-life cases and practical tools they can adapt to their own organisational context, and leave with a concrete action plan covering relevant roles, processes, and next steps. They will collaborate with peers, reflect on their experiences, and experiment with practical exercises, which will help them translate ideas from this article – such as clearer role definitions, structured knowledge sharing, and ongoing learning – into measures that can be applied immediately to their organisations. In doing so, it supports public servants in moving from understanding AI’s potential to using it thoughtfully and responsibly in their daily work.
More information about the seminar and registration can be found here.
Generative artificial intelligence tools have been used to review and correct wording and potential spelling errors in this blog. The final analysis, arguments, critical insights, and conclusions are the result of the authors’ work and remain under their sole responsibility.
