Contributing to the SAPEA: AI in Science and Research

We have contributed to the Commission’s call, “Successful and Timely Uptake of Artificial Intelligence in Science in the EU,” and to SAPEA’s report on AI in science. Our input focused on public and assistive technologies, related datasets, and research.

The European Commission has prioritized the strategic development of artificial intelligence (AI) through funding, infrastructure, and a harmonised regulatory framework. The EU AI Act, the world’s first comprehensive AI law, aims to ensure the safe deployment of AI systems in the EU single market. The EU’s approach to AI centres on excellence and trust. AI has been identified as one of the six key areas that support Industry 5.0, the transformative model towards a sustainable industry. However, the challenge is to strike a balance between innovation and responsible, human-centric AI deployment, while ensuring transparency and accountability.

The Group of Chief Scientific Advisors (GCSA) was given the mandate to provide scientific advice on how the Commission can accelerate the responsible uptake of AI in science. The mandate requests recommendations on what impetus AI could give to scientific productivity, and what benefits, incentives, and challenges AI-enabled research would bring to the European innovation ecosystem and society as a whole.

Summary

This Scientific Opinion from the EU's Group of Chief Scientific Advisors provides a strategic roadmap for accelerating the responsible uptake of AI in European scientific research. The report emphasizes that while AI presents unprecedented opportunities for scientific advancement, Europe needs a coordinated approach to maintain competitiveness while upholding European values.

Key Findings

Europe currently manages over 1,000 AI-related research projects under the Horizon 2020 and Horizon Europe programmes, receiving over €1.7 billion in EU funding. The European Research Council alone has funded more than 1,000 AI projects across all scientific domains since 2007. The European Innovation Council has supported 273 deep-tech companies developing AI-based innovations, with over €557 million in grants and €619 million in equity funding between 2018 and 2023.

Current State of AI in Science:

AI adoption in research is rapidly expanding across all scientific disciplines
Successful applications include protein structure prediction (AlphaFold), antibiotic discovery, and weather forecasting
Europe faces challenges in competing with the US and China in AI development and deployment
The EU currently lacks a dedicated, systematic policy for AI uptake in science

Major Opportunities:

AI can accelerate scientific discovery and enhance productivity
Potential for breakthrough innovations in healthcare, materials science, climate research, and social sciences
Can automate routine tasks, allowing researchers to focus on complex intellectual activities
Enables analysis of vast datasets previously impossible to process manually

Key Challenges:

Reproducibility and transparency issues with AI "black box" algorithms
Data quality concerns and potential perpetuation of bias
Unequal access to AI resources between public and private sectors
Skills gaps among researchers
Environmental costs of AI computing
Intellectual property and copyright concerns

Four Main Recommendations

First, the establishment of AI infrastructure and governance involves creating EDIRAS (European Distributed Institute for AI in Science) as a publicly funded distributed institute, complemented by EASC (European AI in Science Council) for flexible funding mechanisms, continuous monitoring systems, and environmentally sustainable AI initiatives. The proposed EDIRAS would provide high-performing computational power, sustainable cloud infrastructure, and repositories of high-quality, responsibly collected datasets to public scientists across all scientific disciplines, similar to how CERN revolutionised particle physics research collaboration.

Second, improving AI quality standards requires establishing protocols for high-quality, representative datasets, implementing fair access mechanisms across EU research institutions, developing public-private partnerships for data sharing, and supporting epistemic evaluation methodologies to assess AI system limitations.

Third, strengthening research infrastructure coordination involves integrating existing systems with EuroHPC and EDIRAS, fostering collaboration between AI developers and researchers, supporting European SMEs to compete with multinational corporations, and preserving European excellence across scientific disciplines.

Fourth, ensuring human-centric AI development encompasses comprehensive AI literacy training programs, promotion of interdisciplinary collaboration, talent retention strategies, and implementing AI systems that enhance—rather than replace—human assessment capabilities, while respecting cultural factors.

The strategic vision focuses on open science, addressing societal challenges, maintaining scientific integrity, and ensuring equitable access across member states. Implementation priorities include the immediate establishment of EDIRAS infrastructure, development of specialized scientific AI tools, comprehensive researcher training programs, and robust data quality standards.

• • •

References

¹ European Commission. "Commission receives scientific advice on Artificial Intelligence uptake in research and innovation." April 15, 2024.

² Scientific Advice Mechanism, Group of Chief Scientific Advisors. "Successful and timely uptake of Artificial Intelligence in science in the EU." Scientific Opinion No. 15. April 2024.

³ Scientific Advice Mechanism, Science Advice for Policy by European Academies (SAPEA). "Successful and timely uptake of Artificial Intelligence in science in the EU." Evidence Review Report. April 2024.

⁴ Publications Office of the European Union. "Successful and timely uptake of artificial intelligence in science in the EU." 2024.

⁵ European Commission, Directorate-General for Research and Innovation. "Artificial Intelligence (AI) in Science." 2024.

⁶ European Commission. "Trends in the use of AI in science." June 13, 2023.

⁷ European Commission. "Group of Chief Scientific Advisors." Directorate-General for Research and Innovation.

⁸ ALLEA (All European Academies). "SAPEA – Science Advice by European Academies." 2024.

⁹ CORDIS (Community Research and Development Information Service). "Science for Policy by European Academies | SAPEA." European Commission. December 11, 2024.