Paola Cinnella

An expert in fluid mechanics and high-performance computing, Paola Cinnella is taking the helm of the Sorbonne Cluster for Artificial Intelligence (SCAI). Far from the fantasy of automation, this Franco-Italian researcher advocates for AI grounded in reality, which has become a driving force behind scientific discoveries. 

Portrait of a committed director who places ethics, technological sovereignty and dialogue with society at the heart of the future of scientific research.

When Paola Cinnella discusses the future of artificial intelligence, she begins neither with large language models nor with the promise of automation. She begins with physics, climate, medicine and scientific computing. Above all, she speaks of research. For the new director of SCAI, the key challenge for AI lies in its ability to profoundly transform the way in which scientists produce knowledge.

No one would have predicted that this Italian engineer would become one of the leading figures in AI at Sorbonne University. Gifted in applied mathematics, she focused on fluid mechanics from the start of her studies, a discipline at the heart of numerous scientific and industrial challenges, ranging from transport to energy. After beginning her academic career in Italy, she completed her PhD in France before continuing her studies, notably at the École nationale supérieure d’Arts et Métiers and then at Sorbonne University, which she joined in 2020 as a professor at the Institut Jean Le Rond d’Alembert.

Even though her current work combines high-performance numerical simulation, physical modelling and AI, she likes to point out that she did not come to AI via the traditional route. “I wasn’t originally trained in AI,” she explains. “As a fluid mechanic and numerical analyst, I have long worked with scientific computing and high-performance computing, but these involve different algorithms from those used in AI.”

Her unique background as a researcher is a fundamental element of her vision. “I began to drift towards AI during my time as a visiting professor at a statistics laboratory in Chile. At first, it was difficult for statisticians and physicists to understand one another. But once we’d found a common vocabulary, it really helped the work to progress.” This discovery of a collective intelligence emerging from interdisciplinary collaboration has become a primary driver of her work. “On one hand, I learnt a great deal from the statisticians, and on the other, they found food for thought from the problems I brought to the table.” A lesson she now applies on a larger scale.

Taking on a crucial ecosystem

As director of the Sorbonne Cluster for Artificial Intelligence (SCAI), she intends to build on the momentum gained in recent years by her predecessor, Gérard Biau, who made SCAI a major driver for AI at Sorbonne University and beyond. “He successfully fostered a high-calibre scientific community centred on AI,” she summarises. Her ambition is not to change course, but to build on this success. “I want to preserve this scientific dynamism and the high standards that have enabled SCAI to support top projects in fields as varied as the humanities, medicine, astrophysics and mathematics.”

Following this period of initial establishment and rapid expansion, she is now seeking to transition toward a phase focused on consolidating and structuring the existing framework.

AI as a driver of scientific discovery 

For Paola Cinnella, the future of SCAI will focus on developing what the scientific community defines as ‘AI for Science.’ “AI is truly transforming today’s research methods. It is no longer simply a data-processing tool, but a tool for scientific discovery that enables us to tackle problems of far greater complexity than was possible with traditional methods. It is a powerful catalyst for transformation in research, provided it is developed in close collaboration with the relevant disciplines.” There are numerous examples: “This is already the case in high-energy physics, quantum chemistry, the design of new molecules, medicine with new diagnostic methods, climate science and even the humanities and social sciences,” she explains.

Fostering dialogue between scientific communities

As director of SCAI, her aim is to encourage exchange: “At Sorbonne University, we have an exceptional concentration of expertise. The challenge is to forge more lasting links between disciplines to generate high-impact research. For example, an issue of concern to climatologists may be of interest to physicists. A method developed in one field can be applied to another.”

This commitment is reflected in several initiatives, such as the launch of a new monthly seminar dedicated to ‘AI for the sciences.’ “The idea is to invite scientists from various disciplines to present their work in a way that is accessible enough for everyone to understand the issues at stake,” she explains. “There is nothing more effective than bringing scientists together around a table to build a joint project. It forces them to discuss, compare approaches and exchange ideas.”

This philosophy also underpins the development of PostGenAI@Paris, a flagship project that has enabled SCAI to achieve the national AI Cluster status. The challenge is to build on the momentum generated by this programme into lasting collaborations and provide a greater scope to the most promising projects. “My ambition is to build a community of researchers from all disciplines who work together, support one another and use AI to improve our understanding of the world.”

A deeply European conviction

The other major priority of Paola Cinnella’s term of office lies at European level. Her personal background, spanning Italy and France, partly explains this perspective. “I feel deeply European,” she says. Accustomed to international collaboration from a very early age, she believes that the future of AI research cannot be envisaged at the level of a single institution or even a single country.
Faced with the massive investments being made in the United States and China, Europe must strengthen its scientific cooperation. “If we don’t pull together on a European level, we stand no chance of remaining competitive. The infrastructure needed to train or use large models is becoming so costly that no single institution can reasonably develop it on its own,” she asserts.

She also calls for greater responsiveness in research funding. “In some cases, it takes us two or three years to launch a project. Yet in the field of AI, three years is a century!” For the director of SCAI, the ability to provide rapid support for emerging initiatives is a major strategic challenge.

Training and educating to maintain control

A primary focus for Paola Cinnella is education and training. She maintains that universities play a vital role in cultivating skills that will remain essential as algorithms increasingly automate various functions. “Above all, what we need to develop is critical thinking.” Simply mastering the tools is not enough. “We must be able to ask the models the right questions, but also to understand, interpret and verify their answers. If our students do not have a solid understanding of texts, concepts and their discipline, they will be at the mercy of the models,” she warns. In this context, she aims to expand the course offer provided by SCAI further and “develop a comprehensive training catalogue, featuring various proficiency levels, ranging from AI literacy to advanced model development.”

This requirement for understanding also applies to society as a whole. Paola Cinnella observes growing polarisation in public debate. “Part of society sees AI as an absolute danger, whilst another sometimes places excessive trust in it.” Positioned between these two extremes, she advocates for an approach rooted in knowledge, critical thinking, pedagogy, and accountability. “We have a role to play in educating, explaining and engaging in dialogue with society. We must warn of the risks, but also explain, demystify and guide its use.” This also involves reflecting on ethical, environmental and technological sovereignty issues. “A key role for SCAI will be to facilitate the development of responsible and informed AI applications, while integrating considerations regarding ethics, scientific reproducibility, environmental impact, and sovereignty,” she concludes.