Framework for Trustworthy AI/ML in B5G/6G

The world already moves towards the 6G era. AI/ML mechanisms will become structural components of the system and operate in a native manner. As the systems get more complex and intelligent, mechanisms for ensuring trust in those operations become critical. In this paper we take some first steps in the discussion on trustworthy AI towards 6G. Our discussion justifies the need for a framework and highlights that a comprehensive approach needs to be taken, for protecting the input of the AI mechanisms, for achieving an explainable operation, and for guaranteeing proper outputs. Architectural design principles, as well as resource allocation aspects are touched upon, and the important future steps are designated.