Bibliothèque numérique de l'enssib

The six leading European research organizations, united within the G6 network – CNR (Italy), CNRS (France), CSIC (Spain), Helmholtz Association, Leibniz Association, and Max Planck Society (Germany) – representing 135,000 staff members, have published a joint statement calling for an ambitious European strategy for artificial intelligence in science

G6 | THE FUTURE OF EUROPEAN RESEARCH 
 
 
 G6 RECOMMENDATIONS FOR THE FUTURE EUROPEAN STRATEGY FOR ARTIFICIAL INTELLIGENCE IN SCIENCE  
 
Artificial intelligence (AI) is fundamentally changing the way scientific research is conducted. It enables 
the analysis of huge amounts of data, the automation of complex processes and the development of 
predictive models, leading to ground-breaking advances. These capabilities are having a profound 
transformative impact on the entire research landscape and revolutionising fields such as health, climate 
science, physics, materials engineering, among others.  
Developing AI and respective applications within the EU is essential for promoting competitiveness and 
autonomy, upholding European values and ensuring ethical and data protection standards .  
In order to ensure the effective application of AI across scientific domains to boost scientific 
competitiveness, the EU must also ensure that it remains at the forefront of AI development itself. 
Developing AI capabilities within the EU is a prerequisite to maintaining leading position. Strategic 
support for open infrastructures, talent attraction and training, and international collaboration will drive 
excellence in foundational and applied AI research. Investing in home-grown AI solutions reduces strategic 
dependencies on third countries, reinforces data sovereignty, and strengthens EU’s global leadership in 
responsible and cutting-edge scientific innovation. 
The G6 recognizes the need for a robust European Strategy for AI in Science. For each strategic priority 
identified below, the G6 presents a corresponding set of concrete measures, ensuring that every aspect 
of the strategy is directly supported by actionable initiatives. 
First of all, The EU must strengthen fundamental research in core AI disciplines. The success of a solid 
foundation in AI relies first and foremost on strong support for mathematics and computer science, which 
foster the theoretical and technical advances at the heart of AI. 
 
➢ To this end, the G6 calls on the EU to significantly increase funding for fundamental research 
disciplines that underpin all innovation in AI. This includes dedicated support for long-term, curiosity-
driven research in mathematics and computer science, as well as mechanisms to sustain the research 
environments—institutes, laboratories, and teams—where core AI breakthroughs are made. 
Strengthening these foundations is essential to ensure EU’s capacity to lead in the development of 
trustworthy, robust, and cutting-edge AI technologies 
 
To address disparities in expertise across regions, an intra-European mobility scheme modelled on 
successful programmes such as Marie Skłodowska-Curie fellowships should be implemented. This scheme 
would facilitate skill exchanges between Member States while targeting regions with less advanced AI 
capabilities. By fostering researcher mobility, this initiative seeks to bridge gaps identified in the 
Coordinated Plan on AI regarding uneven talent distribution.  
 
➢ To achieve this, the G6 proposes a comprehensive approach combining several measures: the 
implementation of Researchers Exchange Programmes to facilitate both short- and long-term 
research visits between institutions, with financial support for travel and accommodation; the 
organisation of Joint training and development activities—such as workshops, summer schools, and 
specialized training programmes—to promote skill-building and foster international networks; and 
the Joint supervision of doctoral students with mobility grants enabling doctoral students engaged 
in joint supervision arrangements to benefit from guidance by experts in both AI and other scientific 
disciplines, often across different institutions or countries. This should be organized in close 
collaboration with the existing ELLIS PhD co-supervision program 
 
G6 | THE FUTURE OF EUROPEAN RESEARCH 
 
 
A central aspect of this strategy must be the creation of interdisciplinary innovation platforms to 
integrate high-performance computing resources, sector-specific datasets, and expertise from academia 
and industry to address critical challenges such as healthcare advancements, energy efficiency, and 
cultural heritage preservation. For example, multilingual AI models could enable seamless access to 
Europe’s historical archives, ensuring the preservation of its cultural heritage across linguistic boundaries. 
Public-private co-financing mechanisms will support these platforms, ensuring their scalability and 
sustainability. 
 
➢ To support this, the G6 recommends establishing shared Computational Resources and Data 
Infrastructure to develop a federated European AI research infrastructure, including high-
performance computing resources and shared datasets, and interoperable platforms. This ecosystem 
should ensure compliance with EU data governance regulations. The G6 also advocates for expanding 
a Common AI Research Platform to serve as a central hub (such as the AI-on-Demand Platform) for 
accessing, sharing, and validating AI tools, datasets and services. This platform would connect 
seamlessly with the underlying infrastructures and provide standardised benchmarks, best practices, 
and support for both academia and industry. 
 
Cross-domain expertise is central to this effort. The participating organisations bring diverse strengths 
across fields such as healthcare, climate science, physics, and materials engineering. This diversity enables 
advancements in fundamental AI research to directly inform real-world applications while ensuring that 
applied innovations are grounded in robust theoretical foundations. By contributing insights to 
policymakers, this collaboration aims to shape governance frameworks that support innovation while 
maintaining EU’s leadership in responsible AI development. 
 
➢ To enable this, the G6 proposes Collaborative Research initiatives which would be a financial support 
for joint European research projects with shared infrastructure and datasets, focusing both on the 
fundamentals of AI and on AI applications, such as the use of AI in science. Additionally, the joint 
supervision of doctoral students will further promote interdisciplinary projects and training by 
enabling candidates to benefit from expertise across multiple institutions and scientific fields. 
 
Another priority is the establishment of EU-wide AI research standards to support the European Green 
Deal. This includes systematically assessing the environmental impact of AI projects using standardised 
audit tools and adopting best practices for optimising neural architectures to reduce energy consumption. 
Shared methodologies, infrastructures, and ethical guidelines should be developed to foster collaboration 
across Member States while supporting the responsible deployment of AI technologies. 
 
➢ To address this, the G6 advocates for the development of a Policy and Ethical AI Framework, 
complementing existing efforts on the EU level, such as the Living Guidelines on the responsible use 
of generative AI in research. The G6 will engage in developing guidelines, standards and 
recommendations for the responsible development of AI, organising forums for policy contributions 
at EU level, and developing frameworks for integrating AI into scientific decision-making processes 
while ensuring transparency and fairness. Finally, the use of G6's cross-domain expertise can serve as 
an ideal testing ground for the implementation of the European Strategy on AI in Science.  
 
Finally, the use of AI in science contributes to the EU’s efforts in reducing strategic dependencies on third 
countries, reinforcing data sovereignty, and consolidating its position as a global leader in cutting-edge 
scientific innovation. Building on the foundation of trustworthy AI, the integration of AI into scientific 
research will enable researchers to process complex databases and accelerate breakthrough discoveries. 
G6 | THE FUTURE OF EUROPEAN RESEARCH 
 
 
By translating results from fundamental AI into practical tools for science, the EU can unlock 
breakthroughs in key strategic areas such as health, climate change, energy, and digitalization. AI-based 
models enable scientists to analyse large amounts of data with high speed and accuracy, allowing them 
to find hidden patterns and knowledge and helping researchers make discoveries. Moreover, AI acts as a 
catalyst for accelerating the transfer from laboratory discoveries to industrial and societal applications, 
thus facilitating the valorisation of research results and fostering technological innovation across Europe.  
 
➢ This ambition is embedded in each of the proposed measures, ensuring that the benefits of AI for 
science are fully realised throughout the European research ecosystem in line with EU values and 
strategic priorities. 
 
The G6 recognizes that genuine AI innovation requires societal trust. This depends on widespread digital 
literacy and robust data protection. It is only when citizens understand AI and are assured that their data 
rights will be protected that they will be supportive of innovation in AI. This underscores that European 
AI development demands more than technical expertise. By embedding social sciences at the core of AI 
development, we can create technologies that are not just technically sophisticated, but culturally 
intelligent and socially responsive. By prioritizing trust and human context, the EU defines a new paradigm 
of excellence in AI development. 
Through this strategy, the EU will enhance its global competitiveness in AI research, fostering innovation 
and solidifying its position as a leader in both foundational and applied AI research, while ensuring 
sustainable, ethical, and impactful AI development. 
                EXISTING CAPACITIES WITHIN THE G6                                                        
The following examples illustrate the distinctive expertise and capabilities of the six participating 
organisations. When leveraged collectively, these assets have the potential to deliver significant 
benefits for the EU as a whole. 
ELLIS is a pan-European network of the leading AI sites across the continent. ELLIS aims to strengthen 
EU’s sovereignty in modern AI research by establishing a multi-centric AI research laboratory 
consisting of units and institutes across EU. It was established by G6 scientists, and it has quickly grown 
into a network comprising 41 Units and two Institutes, 16 research programmes, and a pan-European 
PhD programme attracting talent from all over the world, with a stream of cross-cutting events across 
the continent. ELLIS includes core sites of CNRS (Paris Unit), CSIC (Barcelona Unit), Helmholtz (Munich 
& Saarbrücken Units), Max Planck (Jena, Saarbrücken and Stuttgart Units, Tübingen Institute), and 
CNR (Modena Unit). Since its inception, it has led a sequence of successful EU networks of excellence 
in AI (ELISE, ELIAS, ELSA, ELLIOT, openEuroLLM). 
The Max Planck Society and the Helmholtz Association are featured among the “Top rising institutions 
in AI in Nature’s 2024 list. In 2011, the Max Planck Society founded the Max-Planck Institute for 
Intelligent Systems, now one of the largest institutes of the Society and an international hotspot for 
modern AI research. Already in 2002, Max-Planck scientists co-founded the series of Machine Learning 
Summer Schools (www.mlss.cc), which have since been organised more than 40 times on six 
continents. AI core methodology is developed across all of our computer science institutes in Bochum, 
Kaiserslautern, Saarbrücken, Stuttgart and Tübingen. Work on applications of AI across the sciences 
pervades the Max Planck Society, and a dedicated AI PhD programme is under development. 
The Helmholtz Association has set up its Information and Data Science framework (HIDS) since 2016 
and, e.g., invested significantly in its AI and data science ecosystem with its “Helmholtz AI Platform”. 
This deep structural capacity building fostered collaborations and underlies concerted efforts such as 
the Foundation Model Initiative (HFMI). The Helmholtz Association operates one of the largest GPU 
clusters in Europe, which is in the process of being extended to the first exascale machine, and it 
played a major role in setting up the LAION initiative for developing open source datasets and models. 
G6 | THE FUTURE OF EUROPEAN RESEARCH 
 
 
AIHUB-CSIC, established by the CSIC, integrates 76 research groups from 35 centers across 17 
scientific disciplines. Its interdisciplinary approach has enabled projects such as AI4HealthyAging, 
which develops tools for early intervention in neurological and degenerative diseases affecting aging 
populations. The CSIC also emphasizes ethical principles through initiatives like HumanE-AI Network, 
which focuses on consensus-building systems that respect diverse perspectives. 
Similarly, the CNRS has established itself as a leader in EU’s AI landscape through initiatives 
like AISSAI (AI for Science, Science for AI), which fosters interdisciplinary collaboration between AI 
experts and scientists from fields such as biology and climate modeling. The CNRS also hosts EU’s 
leading AI-dedicated supercomputer Jean Zay, supporting over 1,400 projects annually with 
applications ranging from cancer research to climate simulations. Its nine national AI clusters further 
integrate research with training and industry collaboration while addressing critical challenges such 
as sustainability and data privacy. 
The CNR plays a pivotal role in aligning Italy’s AI research and applications, particularly in science and 
technology, with broader EU priorities. The CNR leads the FAIR (Future AI Research) initiative, the 
largest national AI programme that brings together more than 20 universities, research institutions, 
and private companies to develop human-centered, trustworthy, and sustainable AI systems. In 
addition, the CNR coordinates the National PhD Programme in AI, a federated programme involving 
61 universities and research centers. The availability of the Leonardo supercomputer - one of the 
world’s most powerful supercomputer - represents a major asset for advancing AI research and 
innovation. 
The Leibniz Association accords high priority to application-oriented research into and the use of AI. 
The Leibniz Research Alliance ‘Health Technologies’ integrates AI potential into medical imaging and 
interpretation, while the Leibniz Research Network ‘Mathematical Modelling and Simulation’ has 
been running AI schools for several years to promote AI skills. The institutes are strategically 
positioning themselves in this field, e.g. in the areas of AI knowledge graphs and data science 
infrastructures, digital materials research and AI hardware research. The decentralised structure of 
the Leibniz Association enables the institutes to apply AI in a targeted manner to the respective 
research questions. In addition, the Leibniz Association is involved in cross-institutional exchange on 
the use of AI in research in accordance with good scientific practice. 
 
The G6 network unites six large multidisciplinary Research Performing Organisations located in Europe, 
the Consiglio Nazionale delle Ricerche (CNR), the Centre National de la Recherche Scientifique (CNRS), 
the Consejo Superior de Investigaciones Científicas (CSIC), the Helmholtz-Gemeinschaft Deutscher 
Forschungszentren (Helmholtz Association), the Leibniz-Gemeinschaft (Leibniz Association) and the Max-
Planck-Gesellschaft (MPG). 
 
This paper was prepared with the contribution of experts representing the six organisations: Massih-Reza 
Amini (CNRS), Fabrizio Falchi (CNR), Carles Sierra (CSIC), Fabian Theis (Helmholtz Association), Harald Sack 
(Leibniz Association), Bernhard Schölkopf (MPG). 
 
 
 
 
 
 


National Superior School of Information Science and Libraries

Not a member yet

G6 recommendations for the future european strategy for artificial intelligence in science

17,334

49,771