Winter Course AI in Public Administration: Balancing Innovation and Responsibility
amsterdam, vrijdag, 24 oktober 2025.
The Netherlands School for Public Administration (NSOB) is launching a winter course on the application of artificial intelligence (AI) in public administration. The course focuses on practical implementation, ethical considerations and the impact on services for citizens. Experts and academics examine how AI can be used to modernise the public sector without losing sight of fundamental values such as transparency and the rule of law. The course starts on 19 January 2026 and is aimed at strategists, executives and policymakers who already work with AI or who will do so in the future.
Practical context: why a winter course on AI is relevant for administrators
The Netherlands School for Public Administration (NSOB) offers a two-day winter course that explicitly addresses the application of artificial intelligence in public administration, paying attention to both opportunities and risks and with a start date of 19 January 2026 described in the course offering [1]. At the same time, the national government and chain partners publish guidelines and programmes that shape digital governance — such as the Code for Good Digital Public Administration and broad activities on DigitaleOverheid — into which AI applications must be placed in a policy context [3][4]. This combination of training and policy underlines that governments must already make choices today about AI applications and governance, something NSOB explicitly responds to [1][3][4].
Personalised information provision: how AI reaches residents
AI makes it possible to personalise communication according to the needs of individual citizens or groups, making information more relevant and easier to process; this aligns with policy goals for accessible and personalised services of DigitaleOverheid [4]. In the public sector such personalisation techniques are often applied via chatbots and content routing that adapt information based on interaction history or profile data — a development that NSOB and policy instruments like the Code for Good Digital Public Administration frame both as potential and as a risk factor for public values such as privacy and transparency [1][3][4].
Chatbots and public services: concrete applications and considerations
Chatbots are increasingly used for first-line services — for example at municipal service centres, counters and telephone enquiries — because they can automate routine tasks and provide basic answers 24/7, a trend also noted in government strategies on digitalisation [4]. At the same time legal and practical sources warn that when deploying generative AI systems the GDPR and the European AI Regulation impose specific obligations, including transparency about the use of AI and risk-mitigation measures when processing personal data [5].
AI-driven information campaigns: measuring, optimising and risks
AI provides tools to measure the effectiveness of information campaigns (for example through A/B testing, behaviour analysis and real-time optimisation of content) and to adapt campaigns dynamically to response patterns, which aligns with goals for a modern, efficient digital government as mentioned by DigitaleOverheid [4]. However, this capability brings responsibilities: the Code for Good Digital Public Administration focuses on public values such as transparency and inclusion that must be reflected in such AI-driven campaigns, to prevent manipulation or unintended exclusion, for example [3].
Practical examples and organisations promoting public alternatives
There are initiatives and debates around public or open alternatives to commercial AI infrastructures; organisations such as Open Future advocate for digital commons and public infrastructure to counter concentration of power and to ensure AI applications deliver social value [2]. NSOB links practical cases and academic reflection in its course so that administrators learn how AI can be used for public value without losing sight of fundamental rule-of-law principles [1][2].
Benefits: reach, accessibility and agility of information
AI can make complex information more accessible through summarisation, translation, adaptive presentation (for example simplified language or multimedia) and profile-driven distribution, functions that can strengthen digital services and align with ambitions for accessible and personalised services on DigitaleOverheid [4][1]. With such tools, different target groups — think of low-literacy citizens, users with a different home language or people with disabilities — can receive information in a better-suited way, provided design and implementation are inclusive and responsible [3][4].
Challenges: privacy, inclusivity and reliability
Key concerns when governments deploy AI are compliance with the GDPR, transparency towards citizens, preventing bias in models and ensuring human oversight; legal advice and practice guidelines stress that prompts, training data and processing purposes must be critically assessed and that public authorities act as controllers under both the AI Regulation and the GDPR [5][3]. The Code for Good Digital Public Administration and other instruments emphasise specific approaches to protect public values, such as impact assessments, human supervision and responsible data choices [3].
Conflicting data and date mentions: observation and caution
There are inconsistencies in public notices about the scheduling of AI courses and events: while NSOB schedules the winter course to start on 19 January 2026, other government sites and agendas sometimes show different dates or brief mentions of ‘winter course AI’ or events on AI and digitalisation [1][4][3][alert! ‘different sources do not always give the same date or level of detail for courses and events; therefore precise planning should be verified with the organiser itself’]. This emphasises that administrators and communications professionals — both participants and providers of information — should contact organisers directly for current registration deadlines and programme changes [1][4].