Bitkom warns against overregulation of AI in Europe

  • Today, hearings for regulating AI kicked-off in the European Parliament
  • Only 9 percent of companies with 50 or more employees in Germany currently use AI applications

Berlin, 12th May 2020 - On the occasion of today’s hearings in the European Parliament’s Legal Affairs committee, Bitkom warns against overregulating artificial intelligence in Europe: It could deter European companies from using the technology and create a significant disadvantage compared to non-European competitors. After long preparatory discussions, during which EU bodies have mainly dealt with ethical aspects of AI, the Legal Affairs committee discusses a draft report on liability issues. It proposes the introduction of strict liability for artificial intelligence in high-risk applications. Liability would be carried by the operator of the AI application, for example the entrepreneur using a machine equipped with self-learning software. In addition, the operator shall be obliged to take out compulsory insurance for possible risks. "Instead of additional regulation, which tightens the liability rules and introduces compulsory insurance, companies need more incentives to invest in the development and use of the technology," says Bitkom President Achim Berg.

A recent survey among managing directors from companies with 50 or more employees in Germany commissioned by Bitkom shows that only 9 percent of companies are using AI applications. For two thirds (64 percent) of these companies, the use of AI is currently not an issue at all. Bitkom believes that AI systems can assist human activities and decisions in various ways, making our lives easier. AI systems are used, for example, in the quality control of substances or the diagnosis of diseases. They can also be used for text analysis, for example to determine authorship, support the maintenance of large plants or the design of complex industrial aggregates. "Artificial intelligence is one of the most promising future technologies," emphasizes Berg. "We must not rush to curb the enormous potential of this technology for economies and societies alike".

Bitkom warns of an uncoordinated package of new regulations for the use of AI. "The initiative being discussed in the Parliament initially concerned AI systems with particular risk potential, but further liability rules and other legal requirements for the development of AI have already been announced," said Berg. For example, there have also been discussions about tightening product liability at EU level in order to involve developers of AI systems more directly concerning liability. In its White Paper on Artificial Intelligence from February 2020 the European Commission with regard to AI high-risk applications proposed new documentation obligations and risk management requirements for AI manufacturers and AI users, mandatory conformity assessments for AI products and a change in the rules on the burden of proof at the expense of AI manufacturers. These new rules would add to the already existing liability for discrimination and data protection breaches. "Since no technology is perfect, we must effectively protect people from damage caused by such systems and no liability gaps should persist. Therefore, it is important to create a consistent framework that is also proportionate," says Berg.

Methodology: The information is based on a survey conducted by Bitkom Research on behalf of the Digital Association Bitkom. In the process, 503 companies from all industries with 50 or more employees in Germany were surveyed by phone. The survey is representative for the overall economy. The question was: „Inwieweit setzt Ihr Unternehmen bereits Künstliche Intelligenz ein bzw. plant oder diskutiert Künstliche Intelligenz zukünftig zu nutzen?“