Standards for the application of artificial intelligence: Science and industry in Baden-Württemberg welcome EU rules – Baden-Württemberg

From red / dpa / lsw

May 6th, 2021 – 8:18 am

—

Karlsruhe / Ditzingen / Brussels – Science and business in Baden-Württemberg welcome the planned EU-wide uniform rules for handling artificial intelligence (AI) – but they also have some critical questions. “Regulation alone will not help,” said Science Minister Theresia Bauer (Greens) of the German Press Agency. It must be prevented that technically unworkable regulations prevent the development of AI applications or that potential users are discouraged.

Europe wants to set global standards in the use of AI. The EU Commission recently proposed rules that are primarily intended to strengthen confidence in the technology. The higher the potential dangers of an application, the higher the requirements. High penalties are provided for violations of the rules.

“The signal is: we mean business now too.”

This uniform approach – if all countries really go along with it – could strengthen Europe as a business location, said Michael Decker, Professor of Technology Assessment, who heads the Computer Science, Economy and Society department at the Karlsruhe Institute of Technology (KIT). “The signal is: We mean business now.” The bureaucracy involved may only slow down the issue in the short term, but in the long term the project promotes innovations under the “Made in Europe” brand – combined with the values ​​that apply here.

At the automotive supplier Bosch from Gerlingen (district of Ludwigsburg), they say: “AI will only prevail if people trust it, that’s why we fundamentally support the Commission’s proposals.” Last year, the Swabians adopted an internal AI code. The human being should therefore always be the controlling authority.

Exports within the EU are made easier and cheaper

From the point of view of the Ministry of Economic Affairs, regulation will increase the trustworthiness of European high-risk applications, “which should sustainably improve the general acceptance of the technology, but also the specific sales opportunities and the competitiveness of European providers”. Exports within the EU are also made much easier and cheaper in view of the same standards.

The machine builders at Trumpf assume that equal opportunities in competition can be preserved and that high-quality and safe AI products will come onto the market. “However, we should avoid investments in promising AI solutions outside the EU due to excessive regulation in Europe and the necessary AI expertise not being built up in this country,” said a spokesman in Ditzingen (Ludwigsburg district).

KIT researcher Decker sees a need for clarification

“Regulatory requirements can represent a major hurdle, especially for small and medium-sized companies and start-ups,” says the Ministry of Economic Affairs. AI innovations should not be made unnecessarily difficult or delayed. The regulation should therefore be limited to those cases in which the use of AI can have serious effects on the safety and health of people and on fundamental rights, said a spokeswoman. Most of the possible AI applications are “absolutely uncritical”.

KIT researcher Decker sees a need for clarification as to which application is to be classified and how. Toys, for example, are on the EU’s index that can encourage children to behave in a dangerous manner – but not an AI-supported video game. “A learning system could drive the player to increasingly aggressive behavior.”

It will be a few more years before the rules apply

The EU states and the European Parliament must negotiate the proposals. It will be a few more years before the rules apply.

Jessica Heesen, Head of Media Ethics and Information Technology at the University of Tübingen, also sees some catching up to do with consumer protection: “Individual complaint options and an ombudsman are practically irrelevant in the regulation,” says her statement for the Science Media Center.

For Science Minister Bauer, it is important that AI should not restrict freedoms or create discrimination. In order to handle data responsibly, she sees the need for basic research: “AI methods must meet the highest standards; among other things, they must be robust and secure.” From Decker’s point of view, the discussion has long since influenced research. Aspects such as data protection and security would be considered much earlier than a few years ago. “We have now reached a level of discussion where clear rules are needed.”

– .