The EU requirements for “Trustworthy AI” emphasize high-quality data, compliance with legal regulations in data processing and storage, system control, and ensuring AI serves the well-being of users.
An AI system architecture for universities should integrate these principles from the outset.
It can be achieved through an algorithmic framework that embeds trust as a fundamental design element—a ‘trust-driven architecture’.
We propose the following architecture and align our development with it, as it combines user-friendliness on one hand and control/governance on the other.