Why EU plans to limit the use of artificial intelligence
The European Commission has proposed tightening control over Artificial Intelligence technologies in order to make AI safer for society. A bill is to be drafted that will protect personal data and interests of citizens of the European Union. This law is a continuation of the GDPR regulation, which protects personal data and prohibits the use of many algorithms for collecting and analyzing data without the direct consent of users.
AI technologies will be categorized by risk tolerance, introducing different degrees of restrictions:
— Unacceptably high risk. Algorithms that manipulate people’s behavior and influence their choice in a particular area fall into this category. For example, the authors of the proposal want to prohibit a social scoring system that evaluates a person according to their social characteristics and predicts their behavior based on social media.
— High risk. This includes online biometrics collection systems, hiring scoring systems, crime prediction systems, robotic surgery, and drones. The EU plans to strictly control this category and limit it if necessary.
— Moderate risk. This category includes seemingly safe voice robots and chatbots. However, to use them, developers will need to notify users that they are communicating with AI, not a human.
— Minimal risk. This includes all kinds of spam filters and gaming services, which will not be regulated for now.
An exception will be made for AI solutions used in the armed forces, security, anti-terrorism, and search for missing children and criminals. They will not be affected by the law.
A specially created European Council for Artificial Intelligence will monitor the enforcement of the law. The council will be composed of one representative each from the EU member states, as well as a representative of the European Commission and the European Data Protection Supervisor.
Companies that provide inaccurate information about their developments or fail to cooperate with these bodies could be fined up to 4% of annual revenue.
How will this affect the AI market?
An important result of the adoption of this law in the EU will be the need for an assessment and certification procedure for high-risk technologies. When deploying AI applications in the European market, companies will have to spend more to comply with the new regulations, which will drive up prices. At the moment, the document is under discussion, and its development may take years.
