If you are an enthusiast of new technologies, you may already know the White Paper on Artificial Intelligence, the document published by the European Parliament to manage and promote the possibilities and risk of the implementation of artificial intelligence in our lives in general and in the European common market in particular.
This book is intended to be a guide on the future position of the European Union in the face of the “challenges and opportunities” that the implantation of artificial intelligence implies, always basing these actions on “European values”. In this way, two clear approaches are proposed to the member states for the development of AI, on the one hand regulation and on the other investment.
Based on these two objectives, the artificial intelligence white paper is divided into two sections: creating an “ecosystem of excellence” and establishing a “trusted ecosystem”.
Ecosystem of excellence
In order to develop an ecosystem of excellence, the EU intends to increase investment in R&D on artificial intelligence, to match the levels of Asia or the United States. With this, the EU intends to become a benchmark in AI, which would be achieved through the following policies:
- Creation of centers of excellence and tests within the framework of the Digital Europe Program.
- Creation of a collaborative network between the main knowledge centers to attract the best scientists on the subject.
- Creation of at least one digital innovation center per member state of the European Union.
- Creation of a financing plan for pioneering projects in artificial intelligence.
- Creation of a public-private partnership on AI, robotics and data.
- Contact and dialogue with the different social groups involved, paying special attention to the health and transport sectors.
To create this “ecosystem of trust”, the European Union undertakes to create a legislative framework for the different economic actors related to AI, since it understands that the implementation of artificial intelligence in certain sectors can be considered a “high risk” advance. For these cases, the European Commission poses a series of requirements (not yet well defined) that should delve into:
- Data retention and management: create a record of all data that has been used to train artificial intelligence.
- Training data: use a large and secure amount of data in a way that avoids discrimination, paying special attention to private identification data such as facial recognition.
- Inform users: clearly that you are interacting with an AI and of the possible data collections.
- Robustness: in such a way that the AI
is not affected by possible failures and errors while it is kept running.
- Human supervision: in such a way that the results given by an AI have to be approved or at least supervised by a human being.
To improve cooperation between different national authorities, the European Commission is committed to creating a European governance structure on AI. This structure will be in charge of “guaranteeing the maximum participation of the interested parties […] consumers and social partners, companies, researchers and civil society organizations – they must be consulted on the application and further development of the framework”.