Successful launch of the “Swiss AI Initiative”
A collaboration between ZHAW, ETH, EPFL, and other Swiss universities and research organisations brings together Switzerland’s AI expertise. The researchers will use the new supercomputer “Alps” to build ethical and transparent AI systems that have real-world applications and benefit society.
The aim of the initiative, led by ETH Zurich and EPFL, is to position Switzerland as a leading global hub for the development and implementation of transparent and trustworthy artificial intelligence (AI). The supercomputer “Alps” of the Swiss National Supercomputing Centre (CSCS) provides the supporting infrastructure. It will go live in spring 2024 and, with more than 10,000 graphics processing units (GPUs), will be one of the world’s most powerful computers for AI applications.
Swiss know-how from ZHAW and beyond
More than 75 professors from all over Switzerland have committed their time and know-how to the initiative and will work closely together in project teams to develop new AI base models for use in different areas. ZHAW scientists Mark Cieliebak, Alisa Rupenyan-Vasileva, Thilo Stadelmann and Kurt Stockinger are part of the initiative and are contributing to topics like health, vision and robotics, education and multimodal AI models. “For example, in the field of computer vision for robotics, AI models are often trained on video footage showing how a task is performed from a third-person perspective. It would be beneficial to include the first-person view, showing the tasks through the eyes of the agent. The partners will incorporate such large datasets to improve the way robots approach a task,” says Thilo Stadelmann. In the future, this method can be applied in various industries.
Transparent and ethical systems for all
The AI models that will be developed must be transparent, deliver comprehensible results and ensure that legal, ethical and scientific criteria are met. This will guarantee benefits for society, industry, science and politics. “The idea is to start using the supercomputer as soon as possible and to develop models that create added value. I believe, that with this research initiative and large collaboration we are equipped to do so – in contrast to a single company. We can bring together expertise from many disciplines and existing data sets, and the investment will benefit many different use cases. Additionally, it will scale up the know-how and training of the next generation of researchers and entrepreneurs in Switzerland,” explains Thilo Stadelmann. Next to the development of new models, the initiative will focus also on fundamental questions in the development and use of LLM models, like the interaction between humans and AI, ethical frameworks, security, data privacy and energy efficiency.