The White Paper on Artificial Intelligence, recently presented by the European Commission, is a reflection document to place the EU at the forefront of these technologies. To achieve this, it is mainly based on two premises: creating an ecosystem of excellence, with greater investment in research, and creating an ecosystem of trust, with improvements to cover the possible risks associated with artificial intelligence.
In 2016, €3.2 billion was invested in Europe in researching Artificial Intelligence, while in the United States and Asia the investment was €12.1 billion and €6.5 billion, respectively. There is therefore no doubt about the need to increase this amount. Some of the proposals included in the White Paper are as follows:
- Ensure that at least one digital innovation center per Member State has a high degree of specialization in AI.
- Provide funding for innovative AI developments with a €100 million pilot plan in the first quarter of 2020.
- Establish a new public-private partnership in AI, data and robotics.
- Attract the best professors and scientists and offer world-class AI master’s programs.
- Create centers of excellence and testing centers that can combine European, national and private investments.
Regarding the «trust ecosystem», he proposes improvements to the legislative framework to cover potential risks associated with artificial intelligence. The EU acknowledges that these systems are complex and can generate mistrust, but stresses that the EU’s strict consumer protection rules, which protect personal data and privacy, will continue to apply.
For high-risk cases, such as those involving health, law enforcement or transportation, the AI systems – the text states – must be transparent and traceable and ensure human verification. Authorities should be able to test and certify the data used by algorithms, as they currently do for cosmetics, cars or toys.
For lower risk AI applications, the Commission foresees a voluntary labeling scheme in case they apply stricter standards.
Regarding the use of facial recognition for biometric identification (currently prohibited), the White Paper suggests initiating a discussion on the circumstances, if any, that could justify such exceptions.