With our partner Salesforce, unify sales, marketing and customer service. Accele your growth!
Just a year ago, in March 2024, the Ministry of the Armed Forces formalized its ministerial strategy in defense artificial intelligencemarking an acceleration in the integration of AI technologies within the French armed forces. Two months later, theMinisterial Defense AIA (AMIAD) was born, with the mission of structuring the development and adoption of algorithmic tools in a sovereign framework. But in an area where national security is at stake, can the state really rely on private actors without jeopardizing its technological and strategic mastery? This is one of the points discussed by Olivia Penichou, Defense Information and Communication Delegate and Bertrand Rondepierre, director of the Ministerial Agency for Artificial Defense Intelligence (AMIAD) in a press point last week.
A delicate balance between autonomy and collaboration
Upstream of the summit for action in AI, Bertrand Rondepierredirector of the Amiad, laid the foundations for a hybrid industrial policy Articulating internal development and collaboration with the private sector, an industrial political policy in terms of defense AI, which will be developed more precisely in the coming weeks, once the summit has passed. Far from a rigid doctrine, he insists on a pragmatic approach:
“In areas that are eminently sovereign and sensitive, even too confidential even to talk about it, there are things that we can neither delegate nor share with manufacturers. »»
AI applied to intelligence, strategic command systems or cybernetic operations is one of the areas where The state keeps the handout of external interference. On the other hand, more open fields, as logistics, predictive maintenance or operational data analysiscan benefit from advances in the private sector.
AMIAD, a strategic tool for digital sovereignty
It is with this in mind that the ministry launched The Ministerial Defense Artificial Intelligence Agency (AMIAD) in May 2024. Designed as a Central pillar of the French Defense AIthis entity’s mission is to ensure the rise in power of the capacities in the armed forces. She plays a role of coordination between the different branches of the ministry, operational forces and industrial actorsin order to accelerate the development and adoption of artificial intelligence technologies in a sovereign and secure setting.
The friend is not limited to the Management of internal projectsit also acts as A bridge between civil innovation and military needs. Its role is to identify the areas where the army can benefit from advances in the private sector without compromising its strategic requirements. It thus makes it possible to adopt a differentiated approach: Some technologies can be outsourcedwhile others, more sensitive, remain under the exclusive control of the ministry.
Partnership with Mistral AI: a controlled cooperation model
The recent Collaboration between Amiad and Mistral AIannounced by the ministry, illustrates this dynamic of Technological co-construction Between public power and private enterprise. The objective is not to subcontract the defense AI, butIntegrate advanced models while ensuring close control over their adaptation to military needs.
“What we bring is all the knowledge of the ministry and the very specific uses that we have, and on which they are not competent,” explains Bertrand Rondepierre. “In exchange, Mistral AI makes us benefit from models in the state of art, which avoids reinventing what already exists. »»
This logic is based on a Structuring triptych ::
- Business expertise of the ministry on military issues,
- Access to data To train models on specific scenarios,
- Secure infrastructure allowing these algorithms to be executed in a sovereign setting.
A strict evaluation grid: cost, delay, performance and confidence
But how far can this controlled outsourcing ? To avoid any risk of excessive dependence or dilution of sovereign control, The ministry applies a rigorous evaluation grid each project. Four criteria are systematically taken into account ::
- The costto guarantee budgetary sustainability,
- The deadlinein order to meet operational needs without excessive inertia,
- Performanceto ensure the effectiveness of the tools developed,
- Confidencecentral criterion to determine the level of private involvement.
“There is no question of creating rent annuities or entrusting strategic prerogatives to industrialists. Each decision is made according to a strict framework, where the imperative of sovereignty takes precedence over any other consideration, “ underlines Bertrand Rondepierre.
Towards a European defense AI?
Beyond the national question, the Industrial Cooperation in Defense AI Also is part of a European and international logic. The displayed ambition of the ministry is to Support the emergence of a competitive ecosystem on the continent levelso as not to be dependent on American (Openai, Palanta) or Chinese (Huawei, Sensetime) actors.
“We must support our industrial partners internationally, while taking care to preserve our strategic interests. The dynamic IA is global, it would be naive to think that it is limited to national borders, “ Recalls Bertrand Rondepierre.
But this opening also includes risks: What part of the data and business expertise can we share with foreign companies? How far can France go without losing control over its own technologies? Many questions are still unanswered.
