Throughout 2018, the European Commission’s approach to Artificial Intelligence (AI) was gradually ticking over with a non-legislative Communication on AI for Europe, and the ongoing work of the grandiosely titled ‘High-Level Expert Group Artificial Intelligence’ who were trying to define ethical principles for a ‘human-centric’ AI. However, this was up-ended in July 2019, when the then President-elect of the Commission, Von der Leyen, used her political guidelines to declare that she would “put forward legislation” on AI in her first 100 days in office. Industry, Commission DGs, and even newly designated Commissioners expressed varying degrees of surprise at this new expedited timeline.
However, from the explicit promise of ‘legislation’, policymakers gradually began referring to a ‘legislative initiative’, and we’re now met with the prospect of a non-legislative ‘White Paper’ for public consultation. With the 100 day deadline coming to a close on 10 March, the Commission is expected to publish this White paper on AI on 19 February.
As outlined in our corresponding video, the key contents of the EU’s approach to AI will be around the regulatory framework, stimulating investment, and increasing data access.
Sketching the regulation on the horizon
The White Paper on 19 February will not bring immediate hard legislation. Instead, it will identify priority areas and open further discussions on AI’s impact for safety, accountability, liability, and transparency. Naturally, the Commission rarely starts such discussions without having an end goal in mind and we can therefore expect a Directive on Civil law liability for AI by the end of 2020. In the longer term, we could see mandatory requirements for still-to-be-defined high-risk AI applications. These are all topics that the Commission will be very keen to better understand during the subsequent consultation on the White Paper.
Opening the vaults and coffers
Whilst often overlooked in the policy heavy world of Brussels, stimulating investment for AI will be an important benchmark for the Commission’s success on AI. The European Innovation Council will look to fund projects that can create sovereignty for European AI, and will thus hope to invest 10 billion in AI from 2021-2027. In addition, the yet to be agreed Digital Europe programme will deliver further investment opportunities, as will Public-Private partnerships. More broadly, the Commission could bring AI investments into a discussion on infrastructure, and further look to create a dedicated European PHD programme for AI.
Pushing the data sharing agenda
Every event, discussion, and announcement on AI in Brussels has included references on the need for more data. To that effect, Commissioner Thierry Breton has stated that AI is ‘first and foremost’ a data issue. Whilst a non-legislative Communication on a European Data Strategy is expected the same day as the White Paper on AI, the lines between the two are quite blurred with AI being a potential ‘trojan horse’ to advance increased data sharing and access. In the coming months, AI will become even more synonymous with data access, and could thus be a central part of the Commission’s ‘Data Act’ that is expected by the end of 2021.
Whilst the promise of ‘legislation’ on AI by 10 March 2020 is no longer an imminent threat, it’s clear that the political drive on AI means that the topic won’t escape the cross-hairs of EU politicians. Parliament will continue to push for action, and the White Paper will lay down deadlines and actions for the next five years. Whilst the prospect of AI may impact the whole economy, the Commission has seemingly focused their priorities around regulation, money, and data.