Time to turn the tide: towards a European model for AI that places humanity at the heart of it – Europe can become a global AI leader a new focus paper reveals.
“Europe can lead rather than follow, shaping a future where technology serves humanity rather than the other way around.” This is a key message emerging from a focus paper on a European Model for Artificial Intelligence published by members of the Expert group on the Economic and Societal Impact of Research and Innovation earlier this month. The paper investigates key aspects of creating a unique European approach to AI, tackling the challenges AI faces, and leveraging its strengths to overcome them.
Structural challenges (infrastructures, skills, and data) and external challenges (poly-crises and the Open Strategic Autonomy imperative) of the current AI landscape have led to the EU lagging behind the US and China. Not only does the rapid pace of technological developments further exacerbate these divergences in terms of growth and productivity, but it threatens European values and priorities, as non-European players are setting cultural and ethical standards in their AI systems. And although US and Chinese dominance may paint a bleak picture of Europe’s role in shaping the AI regulatory landscape, the tide can still be turned. Europe is home to cutting-edge supercomputers and talent – key elements in becoming a competitive AI power. Whilst these resources have, in the past, been under-utilised, there is growing recognition of and support for building and strengthening these capacities to place Europe at the forefront of the AI landscape. Initiatives have already been launched to support a powerful European AI strategy, such as establishing the EU AI Office and AI Factories. Moreover, the Competitiveness Compass provides a concrete roadmap for the EU’s ambitions, and the freshly announced InvestAI initiative to mobilise €200 billion for AI investment will further support Europe’s transformation into an AI continent.
The EU prides itself on standing “at the forefront of championing an AI regulatory framework that takes into account people, the planet and prosperity.” By leveraging and developing EU strengths in science and government technology, focusing on self-sovereign infrastructures, and striving for an innovative-friendly legal framework that is human-centric, a unique European approach to AI will ensure that the EU remains a forerunner in this field.
However, AI stakeholders are concerned by the lack of available venture capital for AI startups in Europe – a problem that persists, as highlighted by the stark contrast between private investment figures in the EU and US. With only 5% of global AI funding going to EU27 start-ups and 64% and 16% to the US and China respectively, it is unsurprising that AI startups in Europe struggle to compete. The EU must therefore rethink its approach to investment, to achieve its ambitions of becoming an AI leader. Notably, though, when comparing cities and regions rather than countries, European ones can compete with those in the US or China. Paris, for example, receives funding comparable to Boston. This gap, apparent at country-level, is mainly explained by Silicon Valley and Beijing. However, increasing and scaling up investments is imperative to leverage Europe’s competitive potential.
The rapid developments and changing landscape of this innovative technology necessitate a comprehensive regulatory framework, one that is ethical, secure, and ensures the equitable use of AI. Although noting the European Commission’s AI legislation (see SwissCore article), the paper highlights two “glaring gaps” that exist in the European legislative effort, as well as in several parallel national and international initiatives. First, many legislative tools are preoccupied with narrow AI and its present impacts, disregarding the potential negative ramifications of Artificial General Intelligence (AGI). Thankfully, the latest version of the AI Act goes in the right direction, by empowering the AI Office to oversee General Purpose AI systems. Second, fragmentation of legislative efforts undermines the EU’s attempt at creating an effective framework imbued with EU values; AI transcends national borders, and international cooperation is thus necessary to establish a global regulatory framework.
Although challenges exist, the ongoing work both at national and regional levels provide much hope for Europe to lead the way in shaping the future of technology. European R&I has the potential to play a critical role in supporting an AI sector that can deliver on societal goals. A coordinated AI policy is crucial, as fragmentation across Member States is cost-inefficient, lacks economies of scale and knowledge-sharing, and results in missed opportunities. Focus must now be on creating a more robust and competitive AI ecosystem. Not only does such an ecosystem require developing an EU-wide computing infrastructure, but adequate data and high-quality and diverse datasets are key to ensuring accurate and unbiased AI models. At the heart of creating such an infrastructure, though, is ensuring the availability of the talent needed to develop these technologies. Although Europe has the talent, much of it migrates towards the US and/or is scattered across countries and institutions with low mobility. Bold initiatives are thus needed to ensure talent-retention and provide mobility opportunities to foster an environment conducive to collaboration and knowledge-sharing. With these elements in mind, R&I has a valuable role in developing an AI policy that can transform the European tech landscape.
The paper concludes by emphasising the two complementary policy elements that can be used to achieve the desired goals: the “moonshot” and the “demand-side” approaches to AI. The former approach aims at developing technological, infrastructure, and workforce capabilities required for AI advancements. The latter approach supports AI industrial applications and needs that contribute to societal goals. Successful integration of both will create a “balanced and comprehensive strategy for AI development.”