Go-ahead for the European AI language model OpenGPT-X

Press statement of the German AI Association on the launch of OpenGPT-X
20.01.2022

Ten German organizations from business, science and the media industry are developing the European answer to GPT-3. The German Federal Ministry for Economic Affairs and Climate Protection is funding the OpenGPT-X project with around 15 million euros.

Led by the Fraunhofer Institutes for Intelligent Analysis and Information Systems IAIS and for Integrated Circuits IIS, the OpenGPT-X project is taking off. It is one of the winners of the Gaia-X funding competition. The goal of the
project is to develop a large-scale AI language model for Europe.
“For business, industry and society, the disruptive potential of AI language technologies is enormous, and international competitors have already recognized this. A European AI language model such as OpenGPT-X is therefore imperative to ensure Europe’s digital sovereignty and market-based independence of Europe,” says Dr. Nicolas Flores-Herr, head of the project at Fraunhofer IAIS. During development, special emphasis is being placed on data protection as well as European language diversity.

Due to the high technical requirements, such as computing power, such powerful language models can so far only be implemented by large companies or consortia. For this reason, the project was able to win a variety of well-known partners.
While Fraunhofer IAIS will develop the AI models together with DFKI and the companies Aleph Alpha and Alexander Thamm GmbH, Forschungszentrum Jülich and TU Dresden will provide their high-performance computing capacities. IONOS SE will oversee the integration into the Gaia-X structure.
Domain-specific use cases for the large-scale language model will be contributed by partners including Controlexpert in the area of claims processing and WDR in the area of media. The KI Bundesverband is responsible for the communication of the project.

So-called big AI language models such as the American GPT-3 or the Chinese Wu Dao 2.0 use enormous amounts of data to create diverse AI models. These models write journalistic texts, for example, provide support in customer service or help to understand complex texts. The problem: data protection and an integration of European languages is not guaranteed in this process.

“With OpenGPT-X, European companies get the opportunity to access innovative language services beyond English and develop them as needed without revealing sensitive company or customer data,” explains Dr. Flores-Herr. “This also allows, for example, individual solutions for regional, small and medium-sized companies can be realized.”

“OpenGPT-X is a real opportunity for AI development in Europe,” says Daniel Abbou, managing director of the German AI Association, assessing the project. “We are pleased that the German AI Association is part of OpenGPT-X.”

Note: The American company OpenAI revolutionized the artificial intelligence market about two years ago with its GPT-3 language model. The model has been trained with over 175 billion parameters, outperforming previous models by a factor of ten. More than 300 applications already use the language model. These applications do not have free access to the model, but can only access an interface.