The European Commission sets out its approach to artificial intelligence

Background

The communication begins from a comparison of spending. In 2017, research and development investment in artificial intelligence in the EU was around EUR 4 to 5 billion, against EUR 6.5 to 9.7 billion in Asia and EUR 12.1 to 18.6 billion in North America ^[2]. That gap is the backdrop against which the Commission set out what it thought Europe should do.

What was adopted

On 25 April 2018 the European Commission adopted a communication titled "Artificial Intelligence for Europe" ^[1]. The document is COM(2018) 237 final, addressed to the European Parliament, the European Council, the Council, the European Economic and Social Committee and the Committee of the Regions ^[1]. A communication of this kind is not legislation. It states the Commission's reading of a problem and the direction it intends policy to take, and it is addressed to the other institutions that would have to act on it.

The document organised its proposal around three policy strands.

The first strand: capacity and uptake

The first strand is to boost the EU's technological and industrial capacity and the uptake of artificial intelligence across the economy ^[2]. In practice this means investing in research and in AI infrastructure, and widening access to that infrastructure. Two instruments are named for the access part: an AI-on-demand platform and digital innovation hubs, both intended to make the technology reachable for small firms and public administrations, not only for large companies or specialist laboratories ^[2].

The second strand: people and skills

The second strand concerns the socio-economic changes that come with the technology ^[2]. Three measures are listed under it: modernising education, helping the workers most exposed to automation to retrain, and training more AI specialists ^[2]. The pairing is deliberate — the same strand covers both the people whose jobs change and the people who will build the systems.

The third strand: ethics and law

The third strand is to ensure an appropriate ethical and legal framework ^[2]. Here the communication attached a date to itself: draft AI ethics guidelines were to be drawn up by the end of 2018 ^[2]. That is the most concrete commitment in the document, and it is a preparatory one — guidelines first, with the legal questions left to be settled afterwards.

The investment target

The communication sets a target of at least EUR 20 billion of combined public and private investment in AI by the end of 2020, and more than EUR 20 billion a year over the following decade ^[2]. The figure is deliberately framed as combined: it counts private money alongside public money, so meeting it depends on whether industry invests as well as on what the EU and national budgets do.

Framing and significance

The text argues that the EU can lead in developing AI "for good and for all" ^[2]. That phrase carries the document's self-description: an approach that treats ethics and inclusiveness as part of competitiveness rather than as constraints added later. Notably, the communication does not use the slogan "AI made in Europe", which belongs to later usage rather than to this text ^[2].

What the communication did was put three things that are usually discussed separately — industrial capacity, labour-market adjustment and rule-making — into a single agenda, and attach numbers to them. The investment target can be checked against actual spending; the 2017 comparison shows the size of the gap it was meant to close; the ethics deadline gave the regulatory strand a first deliverable. For readers coming to EU AI policy for the first time, this is the document in which the shape of the later debate — money, skills, rules — is already visible.