The use of AI in science must be facilitated to accelerate Europe’s R&I efforts and reduce strategic dependencies
CSC welcomes the Commission’s initiative to develop a comprehensive AI in Science strategy for the EU. The combination of artificial intelligence, data, and powerful computing resources is transforming the traditional paradigms of scientific research, and the EU’s commitment to step up the use of AI in science is not only timely but essential for strengthening Europe’s research capacity, driving innovation and competitiveness, and reducing strategic dependencies.
Fulfilling the potential of large-scale AI-models in science needs access to supercomputing and well-managed data. It is essential to continue pooling resources within the EuroHPC framework for world-class European AI-enabling infrastructures – including supercomputers, AI factories, data platforms, open European web index, and high-speed research networks – and their development as a mutually reinforcing ecosystem. Equally important is promoting their interlinkage and connection to the scientific community, to provide researchers with the best possible tools to accelerate research.
Empowering researchers with the right skills and expertise is critical to advancing AI-driven science. Promotion of joint projects between AI or ML specialists and domain-specific scientists could yield valuable results. In addition to AI-specific skills, it is critical to create understanding about technology as a horizontal skillset that is needed in all domains. The AI in Science Strategy should be in line with the Union of Skills Strategy to ensure an integrated approach to competence development, and the Apply AI Strategy for synergies between research and industry.
Alongside access to computing, the ability to develop AI models relies fundamentally on data. Successful application of AI/ML for different uses requires models to be reliable and well-assessed, and high-quality open data is needed to train the models. Therefore, a strategy for AI in science cannot be isolated from building Europe’s sovereignty in data. European data must create value for Europe. Additionally, developing the regulatory environment and legal support for research teams is key to ensure seamless use of training data in AI modeling.
Finally, AI must be used in science sustainably in a broad sense. This includes the ecological footprint and handprint of AI development, as well as upholding European values in AI-based research.
Irina Kupiainen
Irina Kupiainen is responsible for CSC’s Public Affairs.