Book page

Generative AI for Official Statistics

Default profile image
Magda CHMIEL • 6 February 2026

Course Leader

Dr. Marco Puts

Target Group

Statistical production units and methodologists of NSIs.

Managers and decision-makers involved in digital transformation and AI adoption strategies.

Staff who completed the “Introduction to AI for Official Statistics” course or have equivalent knowledge.

Entry Qualifications

Solid command of English. Participants should be able to make short interventions and to actively participate in discussions. 

The course is designed for staff from National Statistical Institutes (NSIs), including those working in statistical production units and methodological departments. It is also suitable for managers and decision-makers involved in digital transformation initiatives and the adoption of AI strategies. Participants should have completed the "Introduction to AI for Official Statistics" course or possess equivalent foundational knowledge of AI concepts and their application in the context of official statistics.

Objective(s)

The course aims to provide participants with the knowledge and practical tools needed to explore and responsibly implement Generative AI (GenAI) in the context of official statistics. Specifically, participants will:

  • Understand the principles of Generative AI and its relevance for official statistics.

  • Identify potential applications of GenAI in statistical production and dissemination, and prototype basic solutions.

  • Develop GenAI workflows tailored to the operational and institutional contexts of National Statistical Institutes (NSIs).

  • Critically assess the risks and limitations associated with the use of GenAI.

  • Apply appropriate governance, ethical, and legal frameworks to the deployment of GenAI technologies.

  • Formulate a practical roadmap for the responsible adoption of GenAI within their institutions.

Contents

Generative AI (GenAI) is revolutionizing how text, code, and even data are produced. For NSIs, it holds promise for automated metadata generation, summarisation, coding, and new dissemination tools, but it raises critical challenges: hallucinations, bias, intellectual property, and compliance with the regulations.

Foundations of Generative AI

  • From AI to GenAI: evolution and key concepts.

  • Overview of generative tasks: text, code, images, audio

  • Key GenAI models.

  • Open-source vs. proprietary approaches: opportunities and risks for NSIs.

  • Programming with AI: Agentic AI

     

Applications in Official Statistics

  • Text applications such as Automated classification of survey responses; Drafting metadata, methodological notes, or dissemination text; Summarizing consultations and feedback.

  • Code generation and assistance for data processing pipelines.

  • Image generation for communication and training.

  • Use of GenAI for data augmentation.

Data Management, Customization, and Integration 

  • Fine-tuning vs. retrieval-augmented generation (RAG) for domain adaptation. 

  • Building institution-specific/domain-specific assistants (Agentic AI). 

  • Managing knowledge bases: connecting GenAI to statistical documentation. 

  • Tools and platforms. 

  • Integration into statistical production: 

Governance, Ethics, and Regulation 

  • Ethical principles in GenAI use: transparency, bias, fairness, explainability. 

  • Specific challenges for GenAI in statistics: Hallucinations and factual accuracy, Confidentiality and protection of sensitive data, Risk of synthetic data misuse. 

  • Regulatory frameworks: EU AI Act, GDPR and data protection obligations, ESS and UN principles of official statistics. 

  • Institutional strategies: when and how to adopt GenAI. 

Future Outlook and Institutional Roadmap 

  • MLOps for GenAI: monitoring, updating, lifecycle management. 

  • Hybrid approaches: combining GenAI with traditional 

Expected outcome

 

 

 

Participants will gain a clear understanding of the transformative potential of Generative AI (GenAI) for National Statistical Institutes (NSIs), particularly in areas such as automated metadata generation, text summarisation, coding, and the development of innovative dissemination tools. They will also be equipped to critically address key challenges associated with GenAI, including hallucinations, bias, intellectual property concerns, and regulatory compliance.

Training Methods

  • Lectures

  • Programming exercises


 

Required Reading 

None

Suggested Reading

Possible References 

Ley, C. “Generative AI for Official Statistics” (NTTS 2025) 

Bender, E.M.; Gebru, T.; McMillan-Major, A.; Shmitchell, S. “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?” 

Bommasani, R. et al. “On the Opportunities and Risks of Foundation Models”. 

UNECE, “Generative AI for Official Statistics”. 

Required Preparation

n/a

Trainer(s)/
Lecturer(s)

Dr. Marco Puts (CBS Netherlands)

 

Rudolph Rössel (Statistical Office of the Slovak Republic)

 

Practical Information

Start date

End Date

Duration

Where

Address

APPLICATION VIA National Contact Point

05 October 2026

09 October 2026

5 days

Statistics Netherlands

The Hague, Henri Faasdreef 312

Netherlands

Deadline for application:

05/08/2026