AI expert concerned about Germany's digital sovereignty
Ms Bertschek, Microsoft and other US companies are investing billions in data centers and cloud infrastructure in Germany. Is this good news for AI development in this country? Or should this not give us cause for concern, because we obviously cannot do it on our own?
Computing capacities are an important resource for training and applying AI models. In this respect, it is good that Microsoft is investing in this country. However, we absolutely need to do more ourselves to strengthen our digital sovereignty and avoid becoming too dependent on non-European providers.
The EFI report repeatedly warns against a loss of digital and technological sovereignty. What would the consequences be?
After all, digital and technological sovereignty means being able to maintain and further develop a technology yourself, and to participate in its standardization or to obtain and use this technology without being unilaterally dependent on other economic areas. It is therefore also important to have the necessary skills to be able to handle and control a technology. If we only rely on investments from Microsoft or other non-European hyper-scalers, this can lead to one-sided dependencies.
And what do we need to do to prevent this from happening?
We should invest in the development and expansion of computing capacities in Germany and Europe ourselves instead of leaving the field to non-European investors. Important elements here are cooperation between public and private players, initiatives such as the construction of an exascale computer at Forschungszentrum Jülich or European collaborations within the framework of European High Performance Computing (EuroHPC). Such computing capacities should then also be easily accessible for researchers, start-ups, and SMEs. However, investments should also be made in AI skills by further promoting basic AI research and by offering courses in schools, academia, and vocational training.
To what extent does German data protection prevent domestic AI developers from designing and training new models? And why is it so important to be able to access domestic data?
It is rather the interpretation of data protection – often even fragmented by federal state – that restricts their use. It is true that restrictive handling of data usage protects the rights of data owners. However, this can lead to AI models from non-European competitors being used in Germany and Europe that have not been developed in accordance with European values such as non-discrimination. We should therefore make every effort to improve the availability and use of data in this country. Not least in order to offer researchers and users of AI models attractive conditions and to be able to use AI to gain knowledge and develop good solutions, e.g. in medicine.
German AI start-ups often lack the funding they need to make progress. Do tax incentives need to be improved? Or is this a question of German culture getting in the way?
Providing stronger incentives for the activation of private venture capital, especially for the growth phase of start-ups, would definitely be an important step. But there are certainly also cultural reasons for this. In Germany, risk aversion is generally higher than in the USA, for example. However, this doesn't just apply to AI, but to the financing of start-ups in general.
The ZEW surveyed domestic companies on the use of AI. The large corporations have probably long since made the leap. But how are SMEs doing? And how can they be helped not to lose touch or become dependent on other countries?
Many SMEs are still unsure about the expected benefits of AI or have concerns about the maturity and reliability of AI. In addition, companies often lack the know-how. SMEs could be supported in getting to know the possible applications of AI through advice and use cases. Established companies can also benefit from their expertise by cooperating with AI start-ups.
