How to prevent Europe from being overtaken in the AI race by China and the United States


The first European summit dedicated to Artificial Intelligence takes place on April 18 at the Palais de Tokyo in Paris. A strong position must emerge from this summit.

A few weeks before the European elections, artificial intelligence is making its appearance in the political debate, following Emmanuel Macron’s speech in his recent forum on Europe. According to a McKinsey study, in 2016, Europe’s public and private investment was €7 billion compared to €38 billion for the United States and €20 billion for Asia. Since then, this gap has widened steadily. A worrying delay: AI is one of the technologies that will most likely impact our lives, as workers, as consumers and as citizens. It will also determine Europe’s place on the world map. It is not too late, but it is urgent to act now by pooling our efforts.

Despite low investment, Europe is not “outdated” as we sometimes hear. For example, European academic publications in AI between 2013 and 2017 are in the lead in terms of number of research publications (30%) before China (24%) and the US (17%)! Another example is the success of the Python library “Scikit Learn”, developed by INRIA, which is now the most widely used in Python for Machine Learning, ahead of the highly mediatized TensorFlow. However, Europe will not be able to be “first” in AI, faced with China, which has published a very aggressive roadmap, and the United States, supported by its technological locomotives.

It is therefore necessary to define the contours of the Europe of AI. In this vision, Europe must be modern, at the forefront of research, but business-oriented, and must establish itself as the guarantor of ethics.

  • Research and training are one of the strengths of the European system. Progress will have to be made in a number of key areas for European sovereignty, such as cyber security and health.
  • The value of AI lies in the productivity gains resulting from the integration of algorithms within organizations, both public and private. Uses must therefore be encouraged to “catch up” with the significant gaps between research and industrial application.
  • AI raises many unresolved ethical questions about its construction and use. Faced with a China that has taken a radical direction on the subject (deployment of the Skynet system and implementation of a “Black Mirror” style social credit, Europe must guarantee ethics and transparency.
  • To achieve this vision, a pragmatic roadmap must be put in place, building on European strengths and the history of our Union:

Pooling public and private investments

One of the virtues of reasoning on a European scale is to be able to concentrate investments in order to better structure and manage them. As such, the European Commission has shared a public and private investment ambition of 20 billion euros by 2020, then 20 billion each year in the following decade. This very ambitious pace seems to be the right one. It will have to be maintained and investments managed in a coherent way.

Creating European databases: this is one of the great needs of algorithmic learning, and it is one of Europe’s cruel shortcomings. Yet our old continent has given us the benefit of a unique social model, with health coverage and pensions, all of which are data that must be aggregated. As such, the “health data hub” proposed by Cédric Villani, and recently integrated into the health law, is a good start, which must be extended. The DGMP is an opportunity: we have a regulatory framework, which allows us to collect data, and to legally build databases on a large scale.

Building the European AI Agency on the ESA model

The European Space Programme is a model of its kind. At the origin of major scientific discoveries, ambitious programmes such as Galileo (the European GPS), but also of a huge commercial success: Ariane. In 2018, our launcher won more than one out of two contracts to put geostationary satellites into orbit. With a budget of €5 billion per year, ESA coordinates national efforts while respecting the agendas of each member country. The organisation of ESA could perfectly apply to AI: it is an excellent compromise between research and commercial ambition, short-term and long-term requirements, European sovereignty and national agendas.

Building bridges between research and business: The different actors must work together to build bridges between research and business.

It is necessary to get the different AI actors to collaborate. This is the ambition of “AI4EU”. Funded by the European Commission, this consortium of research institutes and private companies aims to build an open and collaborative AI platform. The official launch will be the “European AI Night”, the first European summit on the subject. It will bring together more than 2000 stakeholders (start-ups, research institutes, large and small companies, French and European politicians) in Paris on April 18. From their meeting must emerge a strong and clear position, which will be the beginning of a beautiful European union of the AI.

Vincent Luciani, Co-Founder and COO at Artefact


Getting here:
Palais de Tokyo
13 avenue du Président Wilson, 75116 Paris

Door Opening: 1:30 PM
Start: 2 PM
End: 11 PM

Metro: Iéna or Alma Marceau (line 9)

Bus: lines n°32,42,63,72,80,82,92

RER: line C (“Pont de l’Alma” station)