Towards Specifying And Evaluating The Trustworthiness Of An Ai-Enabled System