Open Source: a unique opportunity to create a European AI of trust
France's future is also shaped by lines of code and algorithms. Faced with the domination of the American and Chinese giants in the field of Artificial Intelligence (AI), Europe, and France in particular, have a crucial card to play.
It's time for France to take the lead and equip itself with a trusted AI, to strengthen its digital sovereignty and support its national players.
It is essential to understand that AI is much more than just a technological tool. It is the engine of innovation, a driving force behind the economy of the future. It is a strategic resource that requires national control to guarantee our independence. A sovereign AI is essential to ensure its acceptability and thus become a powerful lever for stimulating our economy, creating quality jobs and supporting strategic industries. Failing to guarantee a fully sovereign (self-sufficient) value chain, the major challenge is to secure a "chain or network of trust" to achieve a degree of strategic autonomy deemed satisfactory. This infrastructure of trust must lower the barriers to entry by pooling development efforts via Open Source digital commons governed by a multitude of InfraTech players.
What's more, by adopting an Open Source approach, we will be promoting open innovation and transparency, the fundamental values of the 3rd Digital Way, which is responsible, sustainable, ethical and respectful of European values.
Today, we are calling on the French government to rapidly implement an ambitious plan for a truly Open Source sovereign AI based on public and open bodies of learning data, documented algorithms to ensure that they can be explained, and a free, non-restrictive licence for use.
This plan must include several components, starting with massive support for French AI players. We have an incredible pool of talent and skills in both academia and industry in this field, which just needs to be encouraged and developed.
This plan should enable the creation and training of so-called "foundation" Large Language Models (LLMs), following the example of GPT, an essential tool for setting up effective conversational AI. This requires a joint and coordinated effort between French AI players to collect and create open data corpora that respect copyright and our regulatory framework (in particular the RGPD), a fundamental step in training such a model.
What's more, we must collectively invest in the development of tools that will enable conversational AI to be deployed on a large scale within French businesses. This is not just a question of competitiveness, but also of digital accessibility, inclusion and equity. The "confidence.AI" programme steered by France 2030 seems to be the ideal vehicle for such investment.
This large-scale deployment requires significant hardware resources (GPUs in particular), brought together in an accessible Sovereign Infratech to deliver results as quickly as possible, so that France can be at the top of the Open Large Language Model (LLM) world rankings.
This plan must also include massive support for the development of a hosting and service infrastructure capable of supporting the needs of our AI ecosystem, to enable the deployment and operation of large-scale sovereign applications.
This national hosting and service infrastructure will play a key role in supporting open source AI, not only ensuring its performance and efficiency, but also enabling the transparency and openness that are essential for widespread adoption.
Lastly, to ensure that this expertise is put to good use, it is essential to create a platform for harnessing French and European skills. The Hugging Face platform, whose success no longer needs to be demonstrated, must not be the only place for AI innovation! For obvious reasons of independence and autonomy, digital biodiversity must be the rule. It is therefore essential to set up an open innovation platform based on generative, sovereign and trusted AI controlled at national level.
It's a call for action, collaboration and national solidarity. Sovereign Open Source AI is not a utopia, but a strategic necessity. We have the potential, the skills and the determination to achieve it.
It is imperative for the French State to initiate this ambition and invest massively in this plan. This is our digital future: whoever masters AI will master their strategic, political and economic destiny.
The list of signatories can be found on the forum published by “Le Point”
- LINAGORA, Alexandre ZAPOLSKY
- Brightness.ai, Michel LEVY PROVENÇAL
- Digital New Deal, Arno PONS
- Fédération Francophone du Futur, Leïla MÖRCH
- Clever Coud, Quentin ADAM
IDFright, Jean-Marie CAVADA