TP1: Legal and Ethical Design of Trustworthy AI Systems

This WP is aimed at providing design methodologies and content for trustworthy AI systems along the different dimensions of trustworthiness, as set forth in the EU documents on trustworthy AI and in the AI Act. The goal of this task is to verify the compliance of AI systems with ethical and legal values (e.g., privacy, fairness, explainability, human control) in order to increase the trustworthiness of AI enabled systems.

Pervasive, autonomous, and intelligent computational entities, in fact, have become an essential aspect of today’s social ecology. By 2025, for instance, 95% of connected devices will be AI-enabled. The magnitude of this phenomenon dictates that AI research and development must comply with ethical and legal principles related to a human-centered AI. Compliance with ethical and legal standards becomes a foundational aspect of AI, as well as AI becomes an asset for reinforcing ethical and legal principles.

From this angle, the relationship of AI with ethical and legal principles is a transversal issue affecting all spokes. All spoke, albeit to varying degrees, will in fact be confronted with legal and ethical issues. Special care will therefore be needed in coordinating these aspects of the various WP. Three TP workshops are planned for this purpose. Depending on the different topics, the transversal project will address:

• Trustworthy and human centered AI in different settings: by-design, in-design, for designers
• Ethical models for trustworthy AI in uncertain multi-agent settings
• Identification and analysis of the ethical implications of algorithms in paradigmatic sectors and scenarios (resilient AI, adaptive AI, symbiotic AI, edge-exascale AI)
• Ethical Foundations for Adjustable Autonomy
• Methods for computable law and ethics: innovative technologies for the legal governance and ethics of AI
• AI act sustainability: reference and check of specific tasks against the principles of the proposed European regulation.