While the Commission's communication is the long-awaited wake-up call for Europe to take the revolutionary innovation of our time more seriously, it lacks courageous new initiatives, did not perform an in-depth analysis of our current situation / capabilities and focuses too much on the potential risks posed by AI instead of concentrating on its enormous potential.
To avoid any misunderstandings, it shall be emphasized already at the beginning of this post that we do not substantially oppose the White Paper of the European Commission. We acknowledge its important role of incentivizing a new approach on Artificial Intelligence (AI) in Europe. Many conclusions, ideas and proposals find our full support. At the same time, we are however rather disappointed with several of its textual elements and - even with a lot of good will - are unable to identify a genuine strategic / geopolitical approach ... something this European Commission has promised to apply.
Ecosystem of excellence
The ambitions laid out in this section can only be achieved if they are backed up by massive investments. At least 20% of the European budgets (EU+MS) should be reserved for the digital transformation (a significant part alone for AI). Our digital infrastructure and education programs require for example very large investments. We should also better coordinate our excellent research centres and digital hubs and accelerate the knowledge transfer to the business sector. Common standards and a radical different competition policy would complete this column.
Ecosystem of trust
The most important point in this section is to execute the so far missing legislative analysis. We need to have a clear picture about which laws we already have and whether they are fit for AI. The worst thing that the EU could do while promoting AI-innovation across the continent, is to create legal uncertainty by creating overlapping laws or by accepting a fragmented Digital Single Market with contradicting national systems. We need to acknowledge that Europe has excellent technology-neutral and principle-based legislation in place that is already (to large extents) future-oriented. Where laws like the GDPR, PLD, GPSD etc. are lacking something, the necessary improvements should be made in a review of that specific law. Only in a few exceptional cases, we actually see the need for AI lex specialis.
Any new laws should thereby only cover high-risk AI-system, while AI must be narrowly defined, focusing on self-learning and autonomous technologies. If the search console of an e-commerce platform for animal food is using AI to generate better search results, this system does not create any harm or does require new legal safeguards. In the same spirit, we are rejecting several conclusions of the AI White Paper (e.g. ethical standards, biases, data, facial recognition). While there are of course potentially negative applications of AI, those should not be the prime target of the new legal proposals. Moreover, they are already addressed elsewhere.
If Europe continues to just concentrate on unrealistic worst-case-scenarios, it will never be able to fully embrace the potential of AI. At the same time, Europe needs to act much stronger against those existential threats that are already appearing in other parts of the world. Normative scoring of individuals - no matter if by governments or by private entities - is incongruous to our European values and traditions. We have a historical obligation to fight against such ideas or tendencies across Europe and should also categorically forbid the export of European AI-systems to third countries that would abuse them.
What are the next steps?
The European Commission will come out with its horizontal AI-framework in Q1 2021, followed by legislative changes to its liability regime at the end of the same year. The European Parliament will soon finalize most of its INIs / INLs on AI and will (after adoption in plenary) send those legislative blue-prints to the European Commission. The new special committee on AI (better known as AIDA) will address the impact of AI on our daily life but without commenting on concrete legislative proposals. As rapporteur of the AIDA report, we will therefore analyse AI more from a positive and future-orientated perspective and complement our findings with a comprehensive road map for a Digital Europe 2030.