Learning Swedish is perhaps the most important thing for someone who comes to Sweden new to be able to get a job and become self-sufficient. Swedish for immigrants (SFI) is offered to everyone, but for many it takes far too long to complete SFI. We are now investigating whether it is possible to develop a tool that with the help of artificial intelligence can help with language learning.




Conversations and workshops with SFI students show that one of the biggest obstacles to learning Swedish is that there are too few opportunities to speak Swedish. Above all, many felt that they wanted to meet more people who speak Swedish. For many, it can take a very long time to get through SFI (Swedish for immigrants) and for some you never succeed.

Learning a new language in adulthood is an enormous challenge and at the same time we know that it is absolutely crucial to be able to get a job and establish oneself in society. Therefore, everyone must have the conditions to learn the Swedish language and also do it as soon as possible. We need to ensure that it is possible to learn Swedish in the way that suits everyone best, and then several different ways and methods are needed.



We are now investigating the possibility of developing a test bed where AI-driven language learning will be made possible in collaboration with development companies, researchers, the idea-driven sector, students and staff. The goal is to be able to give newly arrived and foreign-born residents better conditions to learn Swedish in order to be able to establish themselves more quickly in society and in the labor market.

In recurring loops, we test dialogue with end users to develop a service that meets actual needs. In a first prototype, we focus on dialogue in a grocery store. This prototype has now been tested repeatedly, which has resulted in a basic model for simple language training in the grocery store. Tests from this model are now being scaled up by entire teaching classes doing recurring tests for two weeks. After these weeks, there should be a complete basic model for the food environment, which can then be duplicated in other everyday environments.

Impact target

More people are learning Swedish faster.

Impact target

Faster establishment in society and working life.

Resident involvement

End users are involved in all phases of the project, in needs analysis and in the development of dialogues.


Name: Peter Ekström
E-mail: peter.ekstrom@helsingborg.se