Trustworthy AI: a fuzzy-multiple method for evaluating ethical principles in AI regulations

In this study, we investigated the ethical principles of trustworthy AI and differentiated five prime factors essential for developing trust in AI and most widely presented in regulatory guidelines worldwide. By utilizing Fuzzy Logic Toolbox in MATLAB 9.4, we evaluated the impact of primary ethical principles on trustworthy AI systems in a systematic and structured manner. We discovered that the principle of Fairness and Non-discrimination is the most influential for the development of trustworthy AI, as it is the most represented in the regulatory guidelines. The proposed model offers two main benefits for developers and deployers of AI systems, including predicting the potential public trust in AI systems and assessment compliance with the regulatory frameworks. To ensure the continued trustworthiness of AI systems, the model should be used at all stages of the software life circle, including during development, before placing the system on the market, and at the stage of use to monitor compliance with the safeguards declared to users.