Abstract: Systems and methods of improving AI governance are disclosed. One or more sub-contexts associated with a plurality of users are generated from one or more data sources. The one or more sub-contexts represent one or more changes in data that are relevant to assessing one or more risks associated with the plurality of users. One or more sub-contexts are provided as training data to a plurality of models. Each of the models is associated with a confidence score. A probabilistic assessment of the one or more risks associated with the plurality of users is generated based on an application of the plurality of models to additional data pertaining to the plurality of users received in real time. The probabilistic assessment is presented in a dashboard user interface, the dashboard user interface having user interface elements configured to provide insight into how the probabilistic assessment was generated.