Leading experts make recommendations for a Swiss artificial intelligence strategy

A group of experts in artificial intelligence (AI) has published its recommendations for a Swiss AI strategy. They advocate more intensive use of the technology and the creation of national data platforms, the latter being an essential prerequisite for efficient algorithms. Without appropriate efforts, Switzerland risks losing ground in this technology of the future. Social acceptance is a key factor in the strategy. That's why the public needs to be made more aware of the opportunities and risks of artificial intelligence.

Leading experts in artificial intelligence and members of the "Artificial Intelligence" theme platform of theSwiss Academy of Engineering Sciences (SATW) have drawn up recommendations for a national artificial intelligence (AI) strategy. Among the authors are Hervé Bourlard, Director of the Idiap Research Institute, Joachim Buhmann of the Machine Learning Institute at ETH Zurich, Alessandro Curioni, Vice President Europe and Director of IBM Research - Zurich, Ernst Hafen of the Institute for Molecular Systems Biology at ETH Zurich and co-founder of MIDATA, and Jana Koehler, Chairwoman of the Executive Board of the German Research Center for Artificial Intelligence DFKI.

Five areas are at the heart of the report and its recommendations: Switzerland needs to set up a national data platform; it needs to create an "AI verification and certification body"; society's trust in the technology needs to be strengthened; research on and with AI needs to be encouraged and its use in higher education enhanced; finally, businesses, especially SMEs, need to be able to use AI to their advantage.

 

Data platforms for the common good

Data is the foundation of AI systems. Organizations with large databases therefore enjoy a significant competitive advantage in algorithm development. In order to be able to compete, for example, with the American giants (GAFA), Switzerland should create national data platforms. Citizens could store their data there and share it for the benefit of society, for medical research for example. Citizens must have control over their data at all times, and full transparency over how it is used. This requires democratic control of the platforms, for example by a cooperative organization. The MIDATA health data cooperatives can serve as a model in this respect.

 

Artificial intelligence in education and research

Computational thinking has transformed science: computer-aided modeling is now an indispensable empirical method. AI is producing increasingly impressive results, for example in medical diagnostics based on imaging techniques. However, proving the quality of these results in a guaranteed way remains a challenge. This will require major research efforts. The emphasis must be on understanding algorithms and probabilistic models. This knowledge must be imparted in computer science courses and, increasingly, in other disciplines. Future graduates must be prepared for a future in which they will need to be able to assess and judge the quality of decisions made by AI.

 

National AI strategy "very important" according to experts

To date, Switzerland has not pursued a national AI strategy, unlike many other countries. The EU has also launched a supranational initiative, the Digital Europe Program (DEP), which defines AI as one of its five pillars. 2.5 billion of the program's 9.2 billion euros will go to AI. The program also concerns Swiss research ("Horizon 2020"). The State Secretariat for Education, Research and Innovation SEFRI therefore conducted a survey on the DEP among key Swiss players, and published the results in July. 150 institutions and researchers took part. Those surveyed felt that it was very important for Switzerland to participate in the EPD and develop a national AI strategy.