Jörg Bienert goes back to 2002 when he wants to explain the technological upheaval that is about to take place in the world.

At that time, he says, there were really doubts as to whether Google would actually ever have a functioning business model.

In Europe, it was then observed for a very long time that later attempts to set up their own search engine failed.

"I see great parallels to what is happening with Chat GPT right now," says Bienert, who is President of the Federal Association for Artificial Intelligence (AI) - an association of companies and experts who work on or with AI.

Maximilian Sachsen

Editor in Business

  • Follow I follow

For almost 2 months, the software from the American company Open AI has been causing such a sensation that it is reminiscent of the launch of the iPhone or Google.

Chat GPT is a so-called chatbot that can use AI to answer a wide variety of inquiries - and in a previously unknown quality.

The technology is based on so-called deep neural networks.

This deep learning is an even more sophisticated version of machine learning, in which special algorithms are trained with huge data to recognize certain relationships and later to be able to apply them to unknown data.

Chat GPT can write love letters, programming code or homework, could change society and the economy in the long term.

Only: Once again it is an American company that seems to be able to trigger a digital revolution.

Google and Microsoft are also investing enormous resources in the topic.

"There is once again the danger of monopoly-like structures," warns Bienert.

73 percent of deep AI models come from the United States, and 15 percent from China.

Europe or even Germany?

negligible.

Bienert sees digital sovereignty in terms of AI as "acutely threatened." In order to train advanced AI models, huge amounts of data have to be processed.

This requires appropriate computing capacity.

And that is exactly what is lacking in Germany and Europe, says Bienert.

Germany is looking for the supercomputer

He wants to change that with the LEAM initiative, which the Federal AI Association launched.

LEAM stands for "Large European AI Models".

They want to “enable the German economy to develop large AI models” according to European standards: high demands on data protection, transparent algorithms, CO2-neutral.

In concrete terms, this means above all the construction and operation of a high-performance data center where AI start-ups and industrial companies alike can develop, test and adapt advanced AI models.

There are already supercomputers in Germany, for example at the Jülich research center.

However, these are mainly intended for scientific use and are often fully booked for long periods of time.

In addition, AI models not only consume computing capacity during their training, but also during their subsequent operation.

However, the research computers are also required for projects in physics, chemistry or biology.