Trustworthiness assessment of generative AI in high-stakes applications using circular q rung orthopair fuzzy Hamacher EDAS model

This paper presents a novel multi-criteria decision-making (MCDM) framework for evaluating the trustworthiness of Generative Artificial Intelligence (AI) systems in high-stakes applications under uncertain, imprecise, and conflicting information environments. The proposed approach is based on Circular q-rung Orthopair Fuzzy Sets (Cq-ROFSs), Hamacher aggregation operators, and the Evaluation based on Distance from Average Solution (EDAS) method. First, new Circular q-rung orthopair fuzzy Hamacher weighted averaging and weighted geometric operators are developed, and their fundamental properties are investigated. These operators are integrated with the EDAS model to effectively handle circular uncertainty while preserving hesitation and flexible membership and non-membership information during the decision-making process. The proposed framework is applied to a real-world case study for assessing the trustworthiness of Generative AI systems in high-stakes environments using multiple qualitative and quantitative criteria. To validate the effectiveness of the model, a comparative analysis is conducted with TOPSIS, GRA, CoCoSo, WASPAS, Similarity Measures, and MABAC methods. Furthermore, sensitivity analysis with varying values of parameter q and Spearman Rank Correlation Coefficient analysis are performed to evaluate robustness and ranking consistency. The results demonstrate that the proposed Cq-ROF Hamacher EDAS framework provides a flexible, stable, and reliable decision-making tool for complex uncertain environments.