Interview with Nicole Grobert, Professor of Nanomaterials, Department of Materials, University of Oxford, United Kingdom, and Chair of the Group of Chief Scientific Advisors to the European Commission (part of the Scientific Advice Mechanism). As such, she helped provide independent scientific advice to the European Commission on the uptake of artificial intelligence (AI) in EU research.
Many scientists are already using artificial intelligence (AI) in their fields. What benefits do you see for materials science?
Nicole Grobert: In the study of nanomaterials, thousands of images have to be screened to assess their structure and chemical composition. Here AI can help with screening, and with comparing the nanomaterials against benchmarks and ranking them accordingly. For example, we are currently testing and training a language model to inform, aid and speed up our chemical synthesis methods and other physical experiments. It is quite clear that AI applications will speed up our discovery processes in science.
With its AI Act, the EU is the first governmental body worldwide to enact legislation on the use of AI. The law categorises AI systems according to the intensity and scope of the risks that each AI system can generate. What do you think was the main driving force behind the regulation?
Nicole Grobert: There is a lot in the media about the possible harm and dangers entailed by AI. The rapid development and spread of AI models was probably the main driving force behind the regulation, which focuses on AI safety. We in the Group of Chief Scientific Advisors to the Commission recommend that the technology be explored, applied and integrated in a timely manner into research across all disciplines, including the physical, life and social sciences and the humanities, so that we remain competitive and at the forefront of international leading research.
The recommendations of the Group of EU Chief Scientific Advisors are based on evidence review reports and scientific opinion papers. How should the EU support their swift take-up in science, and what should be implemented first?
Nicole Grobert: One priority for all scientists is the quality of data. For example, if you feed AI machines with ‘rubbish’ they will spit out ‘rubbish’. The output of the machines is only as good as the source they are trained with. This output cannot be taken at face value and must still be verified by experts in the field. We advisors have clearly stated that data reliability is key to using AI. We must ensure that data are of high quality, responsibly collected and meticulously curated, and that European researchers and innovators across all fields have fair access to such data. Another of our priorities is to ensure a people-centric approach. AI is all around us, so we need to promote AI literacy within society. AI applications and their benefits could be prioritised in a range of areas, such as healthcare, the climate, chemistry and advanced materials science.
The Group of Chief Scientific Advisors also has a European ‘AI Institute’ on its ‘wish list’. What role do you envisage for such an institute?
Nicole Grobert: The advisors have a preference for a distributed but coordinated institute, which we have called the European Distributed Institute for AI in Science (EDIRAS), with a focus on AI for research rather than on AI research itself. EDIRAS would have EU-wide operational nodes providing ‘massive’ high-performing computational power, high-quality data, talent training and a substantial and sustainable cloud infrastructure. The institute’s nodes would be located in different places across the EU with a view to being inclusive and ensuring the widest possible access for all researchers. The Group envisages two roles for this institution: first, to boost high-level European research with and into AI. Second, to provide a platform for collaboration with tech developers and companies to help them find the right data and models to match their needs. To function effectively, the institute will require significant funds, so we further recommended establishing an AI for Science Council alongside EDIRAS.
On what areas should funding for AI in research be focused?
Nicole Grobert: There is a particular need for more AI in research in general. In addition, we must ensure that we avoid the opacity typical of the commercial AI sector. Research transparency is key to obtaining the reproducible scientific results that are essential to robust science in an open society.
Clear guidelines are therefore essential. Research development and funding should not be left to the private sector alone, which may collect user data from us without our knowledge. The EU should prioritise AI-powered research in areas where large amounts of data are available but difficult to interpret, such as personalised healthcare, social cohesion and the green and digital transitions. This will bring the greatest benefits for EU citizens.