A few days after the announcement of the 'Democratic Commons' at the Elysée Palace during the gathering of France's top AI talents around President of France Emmanuel Macron, Viva Technology event was the opportunity to present the first AI global research program to reinforce democracy.
Led by Make.org, Sciences Po, Sorbonne University and CNRS, This project has attracted the world’s leading experts in ethical AI: Hugging Face, Mozilla.ai, Aspen Institute, Project Liberty Institute, and Genci. It will bring together over 50 researchers and engineers over a two-year period. Its primary objective is to develop and share a social science scientific framework for determining democratic principles applied to AI, a model for evaluating the biases of large language models (LLMs) against these principles, debiased LLMs, and citizen participation platforms that adhere to these principles.
Marina Ferrari, Secretary of State in charge of Digital: "France defends an open, fair and responsible approach to AI. This is the purpose of the projects we launched for the "Digital Commons" on artificial intelligence. I'm delighted to announce the winners today, including the 'Democratic Commons' project launched by Make.org with Sciences Po, Sorbonne Université and the CNRS, which will provide solutions for assessing and correcting bias in AI systems to ensure their responsible use in democratic processes."
This unprecedented multidisciplinary approach combines advanced expertise in data science with in-depth knowledge of social sciences. It is supported by five leading partners in ethical AI: Hugging Face, Aspen Institute, Mozilla.ai, Project Liberty Institute, and Genci. To ensure the quality and relevance of the work, a Scientific Supervisory Board has been established, bringing together international luminaries such as Yochai Benkler (Berkman Klein Center / Harvard), Hélène Landemore (Yale), Karine Perset (OECD.ai), Asma Mhalla (EHESS/CNRS - Columbia), Raja Chatila (Sorbonne Université), Michelle Barsa (Omidyar Network), and Djamé Seddah (Inria).
This initiative is what we should have done 20 years ago regarding social media: harness their best features while ensuring they do not undermine democracy. In the future, thanks to the “Democratic Commons” program, all states and cities worldwide will have free access to an artificial intelligence that adheres to democratic principles.
Axel Dauchez and Alicia Combaz, Founders of Make.org: "The increasing electoral abstention reflects a deep crisis of trust in democratic institutions. This distrust is exacerbated by the rise of global conflicts fueling an informational warfare aimed at destabilizing democracy, and the emergence of generative AI, which amplifies their means of action. In this context, it was a priority for Make.org to invest heavily in building AI in service of democracy.”
Antoine Petit, CEO of CNRS: “The generative AI and large language models is revolutionizing a wide range of sectors, from justice and education to healthcare and the cultural industries. Anticipating the impacts of these technologies on society – and in particular on the exercise of citizenship in digital spaces – has become urgent. The CNRS brings to this project complementary expertise in computer science, political science and language science. This multi-disciplinary perspective is key to provide answers to the many scientific challenges posed by the study of the biases of generative AI and the control of their possible impacts.”
Nathalie Drach-Temam, President of Sorbonne Université: "Sorbonne Université demonstrates once again with "Democratic commons" the excellence of its research for a more democratic and safer AI. This project is all the more fundamental in view of the forthcoming European elections."
Jean-Philippe Cointet, Director Open Institute for Digital Transformations of Sciences Po: "At a time when AI is putting the democratic principles of our societies to the test, the humanities and social sciences have a decisive role to play in understanding and analyzing digital transformations. As a world-class research university, Sciences Po has the ambition to decipher and provide concrete answers to major contemporary issues. That's why our institution, one of the first universities to set up an OpenInstitute for Digital Transformations, is committed to developing these Democratic Commons, in partnership with the computer scientists who develop these technologies, and by observing in situ the uses they generate."
Paul-François Fournier, Executive VP in charge of innovation of Bpifrance: “Artificial intelligence has the potential to positively revolutionize the constructive engagement of citizens in public life, by reducing the obstacles to their participation. Trust and transparency are essential to its development, and to its inevitable appropriation by all stakeholders. We are delighted to support this global research program, which will enable us to identify democratic biases qualitatively and quantitatively, correct AI models and guide the virtuous development of these future models by making the results available in open-source format.”
The initiative is supported by « the Digital Commons » program of France 2030 and Bpifrance, which is funding the project to the tune of 6 million euros.