The EU plans to invest €6 billion over the financial period 2021-2027 to build a single European data space, reports EURACTIV. The envisioned space will house personal, business, and “high-quality industrial data” (i.e. IoT/sensor data) and create the infrastructure for data sharing and use across businesses and nations.
The “European Strategy for Data,” published by the European Commission on February 19, aims to overcome the “legal and technical barriers to data sharing across organisations, by combining the necessary tools and infrastructures.” The proposal calls to establish data centers in nine strategic sectors: manufacturing, health, finance, energy, climate, agriculture, mobility, law and public procurement, and skills.
The Commission will also fund the creation of the following:
As a result of this work, data will be available across the EU in machine-readable format, for free.
Creating the legislative framework for data governance is listed as the first priority, followed by adopting an implementing act on high-value data sets in Q1 2021.
Data is the major prerequisite for AI. You need lots and lots of it, of good quality and high variety. Meaning, the more diverse the data used for AI/machine learning, the higher its value and potentially the benefits. Today, most data is held by private corporations, and a lot of it is controlled by Google and Facebook. A typical organization only has access to its own data sets plus some public (or partner) data and whatever is available as open source. It does not have access to, say, competitor information – outside of highly aggregated peer comparisons by third-party benchmarking services – even if both parties were eager to exchange their data sets.
In AI, federated learning has attempted to overcome this limitation by pushing learning to the edge, where an algorithm only has access to the local data on a device (say, a wind turbine for predictive maintenance) within a specific organization. Such local models are then integrated on a periodic basis into a global model, so that all wind turbine manufacturers using this model can benefit from the local learnings on their own devices as well as those of their competitors.
The emerging data marketplaces are another step in this direction. Similar to stock exchanges (or perhaps more like the old commodities exchanges), they aim to facilitate data trading and thus its monetization. (Most organizations use somewhere between 1 and 10% of the data they collect.)
Now, the “European Strategy for Data” promises to remove further obstacles to data integration, at least in Europe. By creating the technical, legal, and ethical infrastructure for data sharing, the EU will promote pan-European data exchange, increase data utility, its use, and ultimately its value, so that many more organizations and the public can reap the benefits from AI – all in a manner that promotes trust and respects fundamental human rights. The framework will also serve as a counterbalance to Google’s and Facebook’s dominance over data.
That is assuming it will be successful. This is a very ambitious, expensive, highly complex endeavour of enormous magnitude that will require close collaboration across all 27 EU members, many business sectors, and numerous organizations within and across the member states. I am reminded of an old joke: if you ask 10 lawyers for their opinions, you’ll get 11 back. This project will face similar challenges. We genuinely hope that it will be successful, though. Where there is a will there is a way, and a few recent EU initiatives to establish ethical safeguards for AI showed that it can move quickly.
Looking to start with AI? Download our blueprint Get Started with AI.