#135 Course Correction on AI in the EU?
In this edition of Technopolitik, Bharath Reddy explores how the EU is shifting its AI strategy from a regulation-first approach to an industrial policy focused on promoting innovation and adoption to remain globally competitive.
This newsletter is curated by Adya Madhavan.
Technopolitik: Is Geopolitics Reversing the Brussels Effect
— By Bharath Reddy
Technology geopolitics seem to have reversed the Brussels effect. A year after the EU adopted its landmark legislation to mitigate the risks of artificial intelligence, the rollout is expected to be put on hold amidst heavy lobbying by industry and member states. In what seems like a course correction, the EU has just adopted the “Apply AI strategy”, an industrial policy of sorts to promote all things AI within the EU.
This development comes months after the US launched its AI Action plan to “achieve global dominance in artificial intelligence” and China launched its AI plus initiative that aims to “promote the extensive and in-depth integration of artificial intelligence with all industries and fields of the economy and society”. It builds on the vision of the “AI Continent Action Plan” launched in April with the goal of positioning the European Union as a global leader in AI.
The “AI race” narrative seems to be shaping policy priorities worldwide, and everyone wants to “win” this race, whatever that means. This includes removing barriers to scaling new technologies and creating a favourable regulatory environment that encourages innovation.
While the EU has relied on its market size as leverage to set norms for big tech, the current geopolitical climate has turned technology dependence into a vulnerability, potentially causing a shift in its strategy.
Key Pillars of the ‘Apply AI’ Strategy
The new strategy is designed to address these concerns by focusing on three key areas that signal a shift in the policy direction in the EU.
Targeted measures to boost AI adoption across 10 key industries and the public sector. This includes sectors such as healthcare, robotics, manufacturing, defence, mobility, communications, energy, climate, agri-food, cultural and creative sectors and public sector service delivery.
Support measures and actions to increase the EU’s technological sovereignty by tackling AI development and adoption challenges. These include initiatives such as innovation hubs, AI skilling programmes and regulatory sandboxes.
The creation of a new governance system that brings together AI providers, industry leaders, academia, and the public sector to ensure policy actions are grounded in real-world needs.
What Does This Mean?
The action plan’s policy instruments include direct project funding, discovery challenges to focus industry attention (such as drug discovery), infrastructure development (such as AI factories/datacentres), efforts to mobilize adoption by convening developers and users, skilling initiatives, and the setup of innovation centers and regulatory sandboxes.
While some of these are investments in areas with positive externalities (such as healthcare, drug discovery, climate, environment, and public service delivery), others are more clearly industrial policy measures meant to make EU companies more competitive and fast-track AI adoption across the bloc.
A fund of over €1 billion is being mobilised for these initiatives. A DARPA style focus on high risk high reward projects seems to be the objective. However, success would depend on whether the funding would be too dispersed to have the critical mass for breakthroughs and whether they can attract private capital. There are also risks of scope creep and AI washing in the push to go AI first in several sectors.
The EU strategy is clearly focussed on staying competitive but also attempts to ensure regulation keeps up, indicating a shift from the AI Act. In the race to stay competitive the speed limits for AI adoption (such as the capability-realiability gap, financial viability of projects, learning curves and organisational changes) that allow time for laws and norms to evolve are being reduced.
Note: Gemini was used for some of the research and copy-editing for this article.