Artificial intelligence (AI) is at the heart of worries and fantasies. Its simple definition is a problem in itself. Therefore, it is essential that we can contribute to it in order to be able to offer the world a European, that is, a humanistic, vision of this complex subject, which is becoming more and more structuring in economic, social, environmental and democratic fields.
Therefore, Europe and France, led by AI for Humanity, must be able to think and manage an AI of Enlightenment that is transparent, safe and responsible, refusing to let technological biases become ideological biases with all the consequences we know. For AI to be the guarantor of these values in the service of citizens, we must demonstrate unchanging ethical and political standards towards it and, therefore, make European trusted AI a global benchmark.
We cannot allow opaque AI to guide our choices without being aware of it and without being able to appeal. The general public and economic players should be able to “trust” these AIs that have a measurable impact on their lives (access to credit, job recommendations, health, etc.).
Faced with the algorithmization of our lives and its potential abuses, we really need to take control of our digital destinies by making AI a political entity. This pressing need should force us to transform our values into global standards through the European regulatory package (DSA, DMA, DGA, DA, AI Act). They allow you to impose a trust system to promote your product or service in the EU, making our standards de facto extraterritorial in nature, like the GDPR, which has been adapted in most countries. We firmly believe that by making reliable AI a benchmark, Europe can become a trustworthy, attractive and digitally competitive third way, provided of course that this time everyone seizes the opportunities these new rules create.
In our view, reliable AI is the real scout for our digital sovereignty and our industrial competitiveness.
Sovereignty first. In this report, we define sovereignty in terms of our autonomy, that is, our ability to choose the level and nature of our dependency: “Strategic autonomy is the ability to create and protect the ecosystem of trust that organizes our interdependencies.” Digital sovereignty is an ideal, strategic autonomy is a political goal, and dependency is a measure.
Like energy sovereignty, which forces the state to guide its choices so as not to depend on one supplier or single source of energy, digital sovereignty also seeks to control its dependencies. Without being able to guarantee a fully sovereign (self-sufficient) value chain, the main problem is to provide a “chain or web of trust” to achieve a degree of strategic autonomy that is considered satisfactory.
This robust ecosystem is based on the Cloud, Data, and IA triptych, so we recommend deploying collaborative governance between data and IA, combined with a collaborative approach to use across dataspaces (Gaia-X).
Then competitiveness. Robust AI is an opportunity to (re-)conquer our industries that have been so hard hit by deindustrialization. But for our excellence in AI research to translate into industry leadership, regulation alone is not enough. We absolutely must create industrial cooperation at the European level in order to meet our ambitions. To do this, we propose the creation of “Infratec”, i.e. an ecosystem of European software players, often start-ups and SMEs, that come together to offer a comprehensive software infrastructure offering. He must internally carry European values and rules, harmonized norms or common standards. They are based on a digital asset that will need to be collectively valued in the marketplace, thereby irrigating all of Europe’s strategic sectors (employment, education, health, finance, tourism, mobility, governance, etc.). Finally, it will automatically lower the barriers to entry into the trusted AI marketplace:
“Infratech, foundation of trust and scalability lever for Frenchtech”.
Another bet advocated in this report is to capitalize on the European culture of critical systems, where the requirements in terms of trust are strongest, to make it a competitive advantage and a guarantee for citizens. Today’s AI developments are heavily “driven and influenced” by B2C players like GAFAM. However, their culture is based on use and constant experimentation. On the other hand, Europe could adopt an evidence based, performance rather than process based approach based on risk analysis and the stringent requirements that characterize critical or “high risk” systems as defined under European rules. . This is what citizens and users expect. In this way, we will de facto merge AI and trusted AI, which will allow us to enter the global AI market (231 billion euros) and make trusted AI a real “force” for Europe.
Trust is a defensive weapon against the monopolies we suffer and an offensive weapon for the cooperation we choose. The European Union must provide itself with all the means for its ambitions, both regulatory and industrial. We are convinced that it is by relying on these levers that the EU will be able to build its strategic autonomy and establish its own extraterritorial conditions for the world market, thus providing our companies with the opportunity to export. To do this, Europe must not repeat the mistakes of the past. “Internet case law”, with all the implications we know, is forcing us to be more proactive and more collaborative so that building trust in AI is at the heart of the standards of tomorrow.
(1) Julien Chiaroni, Confiance.AI (Grand Défi IA) and Arnaud Pons, Digital New Deal (Analyze Centre), co-authors of the note “Reliable AI”.
See the full version here: Robust AI: A Strategic Opportunity for Industrial and Digital Sovereignty
SUMMARY OF PROPOSALS
Seize the opportunity for robust AI to build our industrial sovereignty by:
– Exploiting the idea that AI and trusted AI will become almost synonymous.
- Adjust so that all items become critical (social/business critical)
- Aim for the 50 billion strong AI market leader to conquer the 230 billion global market.
– Implementation of trust in AI as a global benchmark
- Rely on our rules to export our legal criteria
- Benefit from our competitive edge through our Mission Critical Systems culture
– Establish standardization as a strategic priority
- Become a standards setter
Give yourself the means to ensure that robust AI becomes the vector of our industrial competitiveness by:
– Funding InfraTech
- Mutual costs to contain and, above all, conquer the big technology market.
- Create an intermediate layer of complex packaged sovereigns
- Support for digital asset creation
– Creation of a European industrial AI alliance (as on Data with Gaia-X)
– Deployment of the overall IA & Data strategy
- Favor a usage-driven approach through dataspaces
- Create a collaborative management of trusted digital