Interest in artificial intelligence (AI) is huge: according to Google Trends, „AI“ is searched for about as often in Germany as Taylor Swift. Worldwide, this query ranks on a par with searches relating to the computer company Apple and its product range. The fact that this technology is on everyone's lips, that many fear it and others look forward to it, is due to its potential to change society.

In contrast to earlier waves of automation, which were primarily able to replace repetitive human activities, AI is now also being used for its intellectual capabilities. Although the algorithm still has no imagination or inspiration, it can reassemble learned connections in language, text, images, and knowledge, draw its own conclusions, recognize trends and form correlations. This makes it possible to improve existing products such as computer tomographs, create new products such as images and essays, and take processes and analyses to a whole new level. Many industries will therefore have to adapt to changes. This applies, for example, to medicine, the justice system, the back office of companies and even creative professions. Old jobs are disappearing or changing, and new ones are emerging. Whether Germany will emerge stronger from this technology revolution and whether the number of job losses on the labour market can be limited will depend crucially on whether the country is able to embrace this technology as quickly as possible in competition with others.

But there seems to be a similar problem in this field as there was with the first wave of digitization: Germany was at the forefront of basic research, held important patents and created global standards with MP3 music compression. But the products, the computers, the network and storage technology, the chips and many applications came from elsewhere. Companies such as Microsoft, Apple, Google, Sony, and Samsung have been setting the tone ever since.

The subsequent waves of digitalization (internet, social media, cloud) also saw European licensees and users pump billions into the corporate accounts of Meta, Google, Amazon and – again – Microsoft. Small domestic seedlings such as telebuch.de or StudiVZ disappeared. Not a single domestic company was involved in the digital platforms (Facebook, Instagram, WhatsApp, YouTube). Instead, American and Chinese companies have created digital ecosystems and tapped into new user groups. In the financial sector, no German bank has stood out with digital products either, instead, American payment system providers such as Mastercard, Visa and PayPal dominate the market.

If Germany were to be left behind again in the AI revolution, it would be fatal for the entire economy. Not only would another part of the added value be created elsewhere, but expertise would also migrate abroad, because the foreign AIs that are used have to be fed with it. New dependencies would arise. OpenAI/Microsoft and Google would then not only be used in the dashboards of German cars and tap into information, but would spread throughout the entire engine room of the German economy and increasingly take over the products for their own purposes.

Nothing less than „technological sovereignty“ is at stake, warns the government's Expert Commission for Research & Innovation (Expertenkommission Forschung & Innovation – EFI). According to the report presented to Federal Chancellor Olaf Scholz, this sovereignty „presupposes that Germany and Europe can maintain and further develop AI technologies themselves and participate in their standardization; or have the means to obtain and apply these technologies without unilateral dependence on other economic areas“.

The Commission is currently „concerned“ that Germany and the EU are falling further behind in international comparison, both in the area of scientific AI publications and in the registration of AI patents.

The region is also „not an international leader“ in the development of significant machine learning (ML) models. In 2022, around half of the major language models came from the USA, 21.9% from the UK, 8% from China, 6.3% from Canada and 5.8% from Israel. Germany contributed only 3.1%. Nevertheless, there are a number of European names in the key AI sectors that can also ensure technological progress in product development: Germany with the GPTNeoX-20B model or with Luminous from Aleph Alpha, and with Mistral 7b v0.1, a French company also joined in.

And why is Germany increasingly being left behind by other countries? First of all, there is a lack of computing infrastructure: this is the basis for the development of AI models. These are tested in large cloud services. Capacity requirements are growing rapidly with the increasing „intelligence“ of these algorithms. Up to now, Amazon and Microsoft have been the main providers of this kind of scale.

The fact that they are currently increasing their capacities in Germany is therefore not only due to the desire to do more business with domestic providers in the data business, but also to gain access to their knowledge and retain them in the long term. At the moment, corporations are taking on a kind of gatekeeper role for AI.

The EFI Commission also warns against a concentration of power in a few US companies and the resulting opportunities to virtually infiltrate domestic companies. In addition, European values could be jeopardized if basic models lead to distorted results in applications that are based on them and this cannot be remedied independently.

Many scientists are therefore looking for salvation in „open source“, where a community of developers works on open models, but must share their findings with everyone. The intelligence of the swarm as an opportunity to outdo the corporations? But there is a catch.

The „free“ developers at the French AI company Mistral are currently experiencing just how quickly „open source“ can turn into „closed source“. It is one of the great hopes for Europe's AI ecosystem because Mistral has further developed a generative AI model that can compete with OpenAI's GPT-4. However, Mistral has now entered into a cooperation with Microsoft. And similar to Microsoft's „cooperation“ with OpenAI, the previously largely open program has suddenly been sealed off from the open source community.

There is no lack of human AI potential in Germany. Scientists and developers from German universities and institutions are often leaders in artificial intelligence. But they should at least find such good conditions in Germany that they are not drawn away. Irene Bertschek, researcher at the ZEW – Leibniz Centre for European Economic Research, warns that she has met many German software engineers in Silicon Valley who have left Germany because of the better offers and development opportunities. And it's not just the earning opportunities, but above all the technical possibilities and infrastructure available there, which are also easier to access in the USA. Bureaucracy and restrictions on accessing research funding in Germany also play a major role, as this restricts freedom and demotivates developers.

The core problem of the German developer scene is the lack of venture capital for start-ups. According to research by the EFI Commission, a total of 79.6 billion US dollars in venture capital was invested in American AI start-ups per year between 2021 and 2023. In China, the figure was 28.5 billion dollars, in the EU only 12.4 billion dollars and in Germany a pitiful 3.6 billion dollars. According to the ZEW, there are currently around 500 relevant AI start-ups in Germany, all of which need funding in order to be able to survive in the further phases of growth and achieve a stable market position.

US corporations, which have built up a large financial buffer through their platform strategy and oligopolistic status in the past waves of digitization, are happy to provide financing. But not out of altruism, of course. In the past, they have always invested in smaller companies in order to take them over and incorporate them so that they could not grow into competitors. Far too late, the competition authorities took notice of this tactic. And often it can merely watch this hustle and bustle because the corporations create facts and the courts only make the final decision years later.

The AI wave is now threatening the next wave of market powering. And the intrusion of large corporations into the value chains of industries that are actually foreign to them is making things even worse – and is pouring more profits into their pockets. Especially as it is now possible for them to take their market position with them into the AI age and expand it even further.

Whether Germany can claim the AI revolution as a success will therefore be decided in the present: more money for computing capacities, cloud infrastructure, legally compliant data material for training the models, and a high level of attractiveness for the domestic developer community are the prerequisites. If considerable sums of venture capital are added to this and the antitrust authorities ensure competition in the markets, domestic companies can look forward with confidence to the AI age, which may then turn out to be a utopia rather than a dystopia.