Using large language models to update data in mapping systems and applications

This invention describes a system for keeping digital maps up-to-date. A vehicle or machine uses its sensors to observe the real world, and a language model on that machine compares what it sees against its existing map data. If the language model finds differences, it creates a coded description of those changes, which is then sent to a central service. This service collects similar reports from many machines to confirm changes before updating the map.