This Swiss research initiative aims to make AI more transparent

The Federal Institutes of Technology in Lausanne and Zurich want to position Switzerland as a world-leading location for artificial intelligence (AI). With this goal in mind, they launched an initiative called ‘Swiss AI’ on Monday.

As part of the initiative, new so-called large language models (LLM), such as ChatGPT, are to be developed and trained, as the two universities (ETH Zurich and EPFL) announced on Monday. Unlike the major language models that are publicly available today, the initiative is based on transparency.

“It must be clear to everyone how and on what data the models were trained and how they arrived at their results,” emphasizes Jan Hesthaven, Academic Vice President of EPFL in the statement.

Swiss supercomputer

In February 2024, the new supercomputer ‘Alps’ will be put into operation at the Swiss National Supercomputing Center (CSCS) in Lugano. According to the universities, this will be one of the most powerful computers in the world.

This gives Switzerland a major competitive advantage, the statement said. Due to the rapid development of generative AI, supercomputing infrastructure worldwide is scarce and largely owned by a few large companies.

In the ‘Swiss AI’ initiative (AI for Artificial Intelligence), the two Federal Institutes of Technology want to combine the expertise of a dozen Swiss universities, technical colleges and research institutions. According to the universities, more than 75 professors have already been recruited.

Sources

(dsc/sda)