Why and how to evaluate Trustworthiness in AI?

While traditional performance indicators help compare AI systems based on technical efficiency, they fail to capture critical userrelated aspects such as perceived AI trustworthiness, which impedes acceptance and long-term adoption. Perceived AI trustworthiness offers avenues for user-centered evaluations of AI solutions across use cases or iterative design phases, but existing evaluation tools lack a clear and operationalizable definition of perceived trustworthiness. Thus, we present a simple framework with distinct measurable constructs (e.g., comprehensibility, technical functioning) grounded in established models. Additionally, we present an overview of existing trustworthiness assessment tools, analyzing their strengths and limitations. This work contributes to the broader goal of fostering AI systems that can be trusted, accepted, and effectively integrated into practice.