Our AIDA draft report rejects the common view that AI is an unpredictable threat and instead argues that it is a powerful tool that could bring many benefits for the EU. As we are however falling further and further behind in a global AI competition, we suggest to urgently implement an ambitious Roadmap that would not only help regaining our technological leadership but also create a genuine European approach to AI.
The political groups in the European Parliament decided on 18 June 2020 to set up three special committees on (1) cancer [BECA], (2) foreign interference including disinformation [INGE], and (3) artificial intelligence in a digital age [AIDA]. In principle, we welcomed this step as special committees offer the unique chance to address horizontal topics, such as Artificial Intelligence (AI), with a cross-committee approach that involves various MEPs with different backgrounds. This seems necessary since AI will influence almost each aspect of our daily life as well as sectors of our economy. The decision to set up three special committees can also be seen as realization by the European Parliament that the existing structure of standing committees is incapable of dealing with horizontal topics. However, the AIDA committee faced major obstacles from the beginning:
Broad but also heavily restricted mandate: the committee's mandate (have a look) explains that the MEPs shall analyse the future impact of AI, further investigate new challenges, analyse the approach of third countries, respond to the 2020 communications of the Commission (AI White Paper, Data Strategy, Shaping Europe's Digital Future), and eventually draft a Roadmap to make Europe fit for the digital age. At the same time, the AIDA committee is not allowed to talk about any topics that were already addressed by the seven previously adopted AI own-initiative reports by the European Parliament in 2020 and 2021. It is still unclear what this means in practice as basically all AI-related issues were to some extent touched upon before.
Out of sync with the EU policy cycle: when the special committee finally started its work in late 2020, the discussions on the AI White Paper from March 2020 were already fading away in Brussels. By scheduling the AIDA draft report for Q4 2021 and the adoption for Q2 2022, the special committee also gave up any chance to shape the Commission's proposals on the Data Governance Act (Nov 20), the AI Act (April 21) or the Data Act (Dec 21) before they were published.
No legislative powers: even after the Commission presented its legislative proposals, the AIDA committee was not allowed to respond directly to them as only standing committees have the necessary legislative powers. How can the AIDA committee fulfill its mandate without talking about the regulative angle, which - by the way - is the central piece of the EU strategy on AI?
Declining interest: when they are created, special committees regularly lead to strong interest among stakeholders as well as journalists. Moreover, most MEPs are trying to be involved, leading to intensive battles between the political groups about influential positions like the Chair or the Rapporteur. However, when the actual work begins, the general interest is often steadily decreasing. The declining number of press requests or the lack of presence of MEPs in committee sessions showed that it is also the case for the AIDA special committee.
Make the most of AIDA
Despite those obstacles, we decided to use our AIDA Rapporteurship to draft the first holistic report of the European Parliament on AI. Our report goes thereby beyond the Commission's AI policies by putting forward a comprehensive plan on how to reach the EU's goal of becoming a global leader in AI in practice. It is divided into five chapters:
Chapter: observes that the fourth industrial revolution has triggered a global tech race that could lead to a global power shift away from the Western world but that could at the same time also generate major risks for our democratic systems and economic prosperity.
Chapter: illustrates that AI is an umbrella term that incorporates numerous old, new and hypothetical technologies, from which the vast majority is completely risk-free. Explains that it is also not a specific AI technology that is risky but certain use cases.
Chapter: examines that the USA is the global leader in AI but that China is quickly catching up. Notes that while the EU is losing more and more ground, it still has the chance to regain lost ground and to even leverage a first-mover advantages by establishing a future-oriented regulatory framework.
Chapter: concludes that if the EU wants to become a global leader in AI, it needs to swiftly implement a comprehensive Roadmap based on seven key areas: (a) Favorable regulatory environment, (b) Complete the Digital Single Market, (c) Digital Green Infrastructure, (d) Ecosystem of excellence, (e) Ecosystem of trust, (f) Industry Strategy, and (g) Security and military deterrence.
Chapter: summarizes the report and - as long as no horizontal digital committee with legislative powers exists - assigns the standing committees with the task of implementing parts of the EU Roadmap.
We hope that the European Parliament uses this Roadmap to strengthen and complement its position on the AI Act for the upcoming inter-institutional negotiations with the Council. Establishing a European approach on AI, which would allow us to take full advantage of AI technologies while shaping international standards, will not be possible by focusing on regulation only. On the contrary, a successful AI strategy needs to be holistic. Our AIDA draft report fills this gap in the existing EU plans on AI. Compared to most other pieces of the European Parliament, our text reads more like an essay than a parliamentary resolution. Moreover, it is based on the findings from numerous AI studies as well as our own public consultations with feedback of over 100 stakeholders. Consequently, the observations on AI as well as the listed policy recommendations in the EU Roadmap are based on well-recognized research and on consisting observations across all Member States.
The next steps
Due to the limitations on characters, the draft report that is presented today is not complete. Many paragraphs of Chapter 4 (EU Roadmap for AI) are missing. They will however be tabled before the deadline for AMs on 6 December 2021. After writing the compromise amendments, we will start to negotiate with the other political groups in January 2022. The vote in the AIDA committee is scheduled for late March 2022, adoption in plenary for April 2022. We expect tough negotiations. While the Shadows seem very reasonable and pragmatic, our draft report features highly political points that will lead to lengthy debates and strong counter proposals. Significant cross-committee and cross-party divisions will further complicate the search for compromises.