What’s the EU up to?
A report from the Meta Forum 2023

On June 27th, 2023, I attended the META FORUM 2023 which was not a Web 3 event as the name might suggest but a conference dedicated to the digitization of European languages and more specifically the state of these languages in LLMs and other AI.

It was a review of a 15-year, mostly academic effort sponsored by various EU programs including meta-net (hence the name), European Language Equality ELE and European Language Grid – ELG. These projects are coming to an end and their results have been documented in these two books:

Georg Rehm, Andy Way (Ed.). European Language Equality: A Strategic Agenda for Digital Language Equality. Cognitive Technologies. Springer, Cham, Switzerland, June 2023. [bibliographic information|PDF file with the full text of the book]

Georg Rehm (Ed.). European Language Grid: A Language Technology Platform for Multilingual Europe. Cognitive Technologies. Springer, Cham, Switzerland, January 2023. [bibliographic information|PDF file with the full text of the book]

The highlights for included the following contributions:

Kristine Eide from the Norwegian Language Council addressed the copyright issue of corpora. They solved the issue in collaboration with the national library which had the corpora, but instead of publishing it, they built a language which they published.

Magnus Sahlgren from AI Sweden spoke about their experience in building an LLM for Swedish and he encouraged everyone to just do it (building the models)  instead of only talking and writing about it.

 
Pedro Ortiz for the DFKI gave a very interesting talk about developing multilingual LLMs

The journey continues with European Language Data Space LDS. LDS aims at a single platform for sharing and also monetizing their language data and other language resources (e.g., language models).