Artificial Intelligence: guidelines for military and non-military use
On Thursday, the Legal Affairs Committee adopted guidelines on the use of Artificial Intelligence (AI) for military purposes and in the health and justice sectors.
MEPs reiterate their call for an EU legal framework with definitions and ethical principles. These rules need to ensure that human dignity and human rights are respected and that AI systems are subject to meaningful human control, allowing humans to correct or disable them in case of unforeseen behaviour. Humans should therefore be identifiable and ultimately held responsible.
Military use and human oversight
MEPs agreed that lethal autonomous weapon systems (LAWS) should only be used as a last resort and be deemed lawful only if subject to human control, since it must be humans that decide between life and death.
The text calls on the EU to take a leading role in promoting a global framework on the military use of AI, alongside the UN and the international community.
AI in healthcare and justice
The increased use of AI systems in public services, especially healthcare and justice, should not replace human contact or lead to discrimination, MEPs assert. When AI is used in matters of public health, (e.g. robot-assisted surgery, smart prostheses, predictive medicine), patients’ personal data must be protected and the principle of equal treatment upheld.
Judges use AI technologies more and more in decision-making and to speed up proceedings. However, safeguards need to be introduced to protect the interests of citizens. People should always be informed if they are subject to a decision based on AI and should have the right to see a public official. AI cannot replace humans to pass sentences. Final court decisions must be taken by humans, be strictly verified by a person and be subject to due process.
MEPs also warn of threats to fundamental human rights arising from the use of AI technologies in mass surveillance, both in the civil and military domains. They call for a ban on “highly intrusive social scoring applications” (for monitoring and rating of citizens) by public authorities.
The draft resolution was adopted with 16 votes in favour and 4 votes against.
Background
Parliament has already adopted a wide range of resolutions in the field of AI and most recently approved three reports on Artificial Intelligence in the areas of ethics, civil liability and intellectual property. Ahead of the Commission legislative proposal on AI expected in early 2021, Parliament also set up a new Special Committee on Artificial Intelligence in a Digital Age (AIDA).