Because 73 percent of the large AI models come from the USA, around 15 percent from China, and the rest stays in Europe. “And that’s a huge problem, because these big AI models are going to change the world – even though I usually hate that expression,” says Daniel Abbouthe managing director of the German KI Federal Association compared to Ö1 Digital.Leben.
Rapid spread
Artificial intelligence is actually developing rapidly into “disruption” – i.e. a fundamental change in technologies and the economy. ChatGPT has reached 100 million users within two months. Compared to other technical innovations, this increase is unique. The landline phone took 35 years for this number of users, mobile phones around 17 years, Twitter around two years. And the generative language models alone have the power to completely change professions such as that of the lawyer and other medium-sized jobs.
Digital Junkyard
Daniel Abbou uses an example from Lithuania to demonstrate the extent to which artificial intelligence can influence value creation. There, a car junkyard was digitized. Next to Poland, Lithuania is the country in Europe where most used cars are dismantled. An AI system was trained on car parts and now evaluates the inventory at the junkyard.
“For example, if you remove an exhaust from a car, you take a photo of this exhaust, the system tells you it’s a 2002 VW Golf exhaust and it’s compatible with the Polo,” says Daniel Abbou, describing the benefits of the systems. In addition, it prints a QR code for inventory and suggests what price the part should sell for in the market.
Chip industry as a bad role model
For Daniel Abbou, the future competitiveness of the European economy depends on its own AI systems. “When these big models come from the USA and the Austrian, German, European economy enters their data there, then all the economic data and thus our secrets also flow to the USA.”
In contrast to private data, which has long been sent to Facebook or Twitter, large parts of our economic data are still here in Europe. And Abbou believes that they should stay here so that we don’t make ourselves dependent on other continents, as has long been the case in the chip industry, for example. But there is also a lack of understanding among the business bosses of the disruptive power of artificial intelligence.
Europe needs its own AI infrastructure
Die feasibility study of the German Federal AI Association suggests creating their own European AI centers, especially large computing systems, since artificial intelligence is very computationally intensive. The existing high-performance data centers would not suffice, since there is no quick computing time.
For Germany, the feasibility study puts the costs of an AI data center at 350 to 400 million euros. But centers like this are needed throughout Europe, and also across national borders, says Daniel Abbou. “The big things can only be solved at European level. Digital legislation does not take place in Berlin, Vienna or Paris, but in Brussels. And we have to stop thinking nationally. We have to think continentally when it comes to digital stories.”
