Building trust in administrative automation through the use of LLMs in the public sector of Sweden

Niclas Willem Fock

CEO Santa Anna IT Research Insititute, Sweden (Linköping University, Department of Electrical Engineering)

Introduction: Santa Anna IT Research Institute (“Santa Anna”) has through its membership in the consortium “AI Sweden” collaborated on the development of the GPT-SW3: LLMs. The GPT-SW3: models are since November 2023 open for further development, research and applications. Santa Anna’s position as a research institute in the “AI Sweden” consortium, allows us to work on the application of LLMs, and soon LMMs, in the Swedish public sector. In this project, Santa Anna will arrange a workshop in East Sweden aimed at building trust in LLM technology in the public sector. Further the workshop also aims at building an understanding of how to implement trustworthy AI-systems in the public sector for automation of administrative workloads. The theme for the workshop will be “Using generative AI and language models in the Swedish public sector”. The workshop is aimed at decision-makers, operationally and politically, as well as operational managers, within public sector responsible for project implementation. Background In 2023, Santa Anna, together with a group of public partners, identified some key challenges with generative AI and the application of language models: –
The field of Applied Artificial Intelligence is moving extremely fast and accelerating
– The potential for automation and savings with the help of AI in the public sector in Sweden is gigantic
– Management in the public sector has not sufficiently acquired knowledge about AI and the opportunities this brings, which has led to internal frustration within a number of public organizations. Not least this is about risks with, credibility and trust in, AI, as well as knowledge about the application of the new regulations that have been developed nationally and internationally
– Cooperation between different organizations and authorities is a key factor for success, but cooperation does not take place to a sufficient extent.
Purpose: There are today a number of use cases for language models (foundation models) that have been identified as possible to implement. The purpose of the Workshop is to point to such possible use-cases, but also to demonstrate completed and ongoing project examples to effectively illustrate what can be realistic goals, or possible projects to start as soon as the will and resources are available, and effects mapped within the respective organization. Through the workshop, a number of concrete cross-joint projects will be proposed. In preparation for the Workshop, a dialogue with the authority cooperation agency eSam ( will be initiated to propose how such cross-joint projects could be financed, realized and followed up.

Keywords: Understanding Generative AI, Public AI Infrastructure, Building trust in Large Language Models, Adapting Multimodal Foundational Models in Public Sector, Applying AI in the Swedish Public Sector, AI driven public organizations, AI for Exectuvie Managment

Scientific area: Applying Generative AI, Applying Large Language Models in Public Sector, Multimodal Foundational Models, Applying AI in the Public Sector, AI driven public organizations

Bio: Niclas Fock (born 1964) has a Master of Science within electronics, computer technology and image analysis, and a research background in Vocational Skills & Technology. In his licentiate thesis Adventure lust & risks (Dialoger, Sweden 2004), he examines using the dialogue seminar methodology, experience-based learning and judgment in the leadership required when conquering new areas of knowledge. Niclas has been working for the Swedish national AI initiative, AI Sweden ( since 2019 and is Research Arena Director in Wallenberg’s major research initiative on AI, WASP. He has previously, among other things, been CEO and director of RISE SICS East, part of the national research institute RISE Research Institutes of Sweden, and he is since 2023 CEO of the Santa Anna IT Research institute. Niclas was previously CEO of the research company Creo Dynamics (now Creo Faurecia) and worked prior to that, for a long time as a Head of Division in the management Combitech (a Saab Group subsidiary), with main responsibility for the Systems & Product Engineering parts of the company, where he also built up and led research and development of a learning organization through the so-called Combitech Learning Lab (now called Combitech Grow). In recent years, Niclas has worked with the development and implementation of courses within the Master’s program “Vocational skills and professional development” (120 credits) and “AI and professional skills” (15 credits) at Linnaeus University. Niclas works in leadership development and often lectures on the subject of Knowledge-intensive/Learning organizations and in AI and has published several articles in the journal Dialoger, in the book Dialogue, Skill & Tacit Knowledge (Wiley) and was the main editor of the book ‘Spänningsfält’ (Dialoger, 2016).

Visiting period: Workshop in June 2024