Major Danish players develop artificial intelligence: New language model to safeguard Danish language and culture

This post is also available in: Danish

We are regularly bombarded with news about how artificial intelligence is going to change everything in our society. For better or worse.

That’s why a consortium of Danish actors across the public and private sectors has developed a Danish language model to ensure that the Danish language and culture are not erased in the exponential development.

“It is a great strength for Denmark that innovative forces across the business and public sectors are joining forces to invest in developing Danish-based language models. Danish initiative and investment is important for us to reap the benefits of technology in Denmark and protect Danish language and culture. It is crucial that the development is done responsibly and securely so that we can have confidence in the models,” says Marie Bjerre, Minister of Digitization in a press release.

A collaboration between the public and private sector

The Alexandra Institute, IBM Denmark and the Danish Chamber of Commerce are at the forefront of the fight to protect Denmark, which, in addition to safeguarding Danish culture and language, must also protect Danes’ privacy.

The Danish Language Model Consortium will donate Danish data for the training of Denmark’s common language models built on theDanish Foundation Models‘ base model, Munin. A language model developed in collaboration between Aarhus University, University of Copenhagen, University of Southern Denmark and the Alexandra Institute.

“Common, free and responsible AI-based language models in Danish will become a national asset and ensure the continued digital development of Denmark. The role of the consortium is to contribute Danish data and use cases. “The development of the basic models is taking place under the auspices of Danish Foundation Models, which is a collaboration between Aarhus University, University of Copenhagen, University of Southern Denmark and the Alexandra Institute,” says our Head of AI, Jens Kaas Benner, who is one of the initiators behind the consortium.

Casper Klynge, Deputy Director of the Danish Chamber of Commerce, has made it clear in a statement to Berlingske that the language model is not intended to compete with Google’s Gemini or OpenAI’s ChatGPT.

“I’m sure the big language models will still be best for many things. But there is a huge demand for customized models with credible data protection guarantees. That’s what makes this model so special.”

The actors behind the Danish Language Model Consortium:

Aarhus Municipality, Aeven, ATP, Bagger Sørensen Group, Cbrain, Falck, HOFOR, JN Data, Kommunernes Landsforening, KOMBIT, Norlys, Ordbogen A/S, DLA Piper, Salling Group, SDC, Topdanmark, Visma Enterprise.