Systems and methods for generating artificial intelligence models and/or rule engines without requiring training data that is specific to model components and objectives

This invention describes a system to automatically create or improve software rules for AI systems, like those used in decision-making. Instead of needing extensive, specific training data, it starts with an existing set of rules. An AI model then rewrites these rules, and the system checks the new versions for mistakes or unexpected changes by comparing them to other automatically generated alternatives. If a change is found, it determines if that change is actually an improvement, specifically for rule engines.