Method of and server for training a machine learning algorithm for estimating uncertainty of a sequence of models

This invention describes a system that uses a special artificial intelligence (AI) to figure out how confident a series of other computer models, which might also include AI, are about their predictions. This uncertainty-estimating AI is trained on various pieces of information, including the original data, the predictions made by the main models, and specifically, details about the format of the inputs and outputs, as well as other background information not directly used by the main models. The system uses two separate computer servers, one for the main models and another for the AI that estimates uncertainty.