At the heart of the Swiss AI Initiative is the supercomputer, which will go into operation in spring 2024 at the Swiss Center for Scientific Computing (CSCS) in Lugano. The computing power provided by NVIDIA's new Grace Hopper Superchips will make Alps one of the world's most powerful supercomputers in artificial intelligence.
Alps will be used to develop and train new Large Language Models (LLMs) that can compete with those of private industries such as OpenAI (ChatGPT's company), Microsoft, Google and Meta. A number of "foundational models" of artificial intelligences will also be developed on the basis of this common infrastructure for specific tasks in the areas of sustainability and climate change, health, robotics and computer vision, education and science in general.
Alps supercomputer(Image: CSCS)
The goal of the Swiss AI Initiative is not only to provide Switzerland with an important competitive advantage over the international level, but also to develop LLM in a transparent and open manner, making public both the data with which these language models are trained and the results obtained, while ensuring compliance with legal, ethical and scientific requirements. This information is generally not available for language models made by private companies.
The Swiss AI Initiative, coordinated by the two Swiss Federal Institutes of Technology, involves several other Swiss academic entities, including as mentioned the IDSIA USI-精东影业. A number of international researchers have also been invited to collaborate on the development of multilingual open source LLMs.
The contribution of the Dalle Molle Institute for Artificial Intelligence concerns the development of both the common infrastructure and some specific projects. In particular, IDSIA will be responsible, as far as language models focused on text processing are concerned, for activities aimed at increasing the reliability and transparency of the data and reasoning models used and aspects of "alignment" in interaction with humans. As for specific projects, under the leadership of Dr. Fabio Rinaldi, IDSIA is focusing its activities in the area of fundamental health models in the working group led by Professor Gunnar R盲tsch (ETHZ) and Professor Marcel Salath茅 (EPFL).
For IDSIA Director Andrea Emilio Rizzoli, "This is a strategic project that will benefit not only Switzerland by sharing methods and results to the entire scientific community: the field of LLMs is currently dominated by private players who, for commercial reasons, share little information on important aspects such as the data used for training and the algorithms employed, and it is appropriate for public research to catch up in order to enable responsible development of artificial intelligence."
The IDSIA team currently involved in the Swiss AI Initiative is currently composed of, in addition to director Andrea Rizzoli and scientific director Marco Zaffalon, professor Fabio Crestani and researchers Fabio Rinaldi, Marco Forgione, Sandra Mitrovic, Oscar Lithgow Serrano and Joseph Cornelius.