I think after natural language models like GPT, semantic databases are more important. 1)act as database/knowledge base with query language etc 2)converted to natural language while keeping the rich semantics, that can be read from a chat bot, and use the infromation in natural language also, so people can use those data in natural language. (we could do the second with a database, but it has so poor semantics, compared to a knowledge base) Maybe i am wrong, maybe its the opposite, like we have natural language so no need for knowledge bases, What do you think?