System and method for dynamic multi-party verification of generative artificial intelligence systems

A computer-implemented model verification system for verifying and evaluating a machine learning model in an enterprise, comprising

• a model aggregation unit for aggregating together a plurality of the machine learning models associated with the enterprise, wherein each of the plurality of machine learning models have metadata associated therewith,
• a cohort determination unit for determining and selecting a plurality of cohorts from a set of cohorts according to one or more cohort attributes to perform a verification process on one or more of the plurality of machine learning models to verify the one or more machine learning models and to generate cohort verification data, wherein the plurality of cohorts are blinded relative to each other, and wherein the cohort determination unit generates output cohort data indicative of the plurality of cohorts,
• an objective assessment unit for assessing a consistency between the plurality of cohorts when verifying one or more of the plurality of machine learning models based on the cohort verification data and the output cohort data and generating an assessment score that is indicative of a selected level of consistency of the cohort verification data between the plurality of cohorts,
• a model evaluation unit for applying an evaluation process to the assessment score and to model evaluation data for evaluating a performance of the machine learning model and for generating output model evaluation data indicative of the evaluation of the performance of the machine learning model,
• a result assessment unit for receiving and processing the output model evaluation data and for automatically assessing the performance and reliability of the machine learning model based on the output model evaluation data and for generating assessment results,
• a digital trust infrastructure having a blockchain for storing the machine learning models and associated metadata, the plurality of cohorts and output cohort data, the assessment score, the output model evaluation data, and the assessment results, and
• a model trust card generation unit for generating a model trust card from selected data stored in the digital trust infrastructure, wherein the model trust card sets forth model specific information of the machine learning models that are verified and evaluated.