maybe OpenAI have a terribly inefficient data ingestion pipeline? (wild guess) basically taking in new data is tedious so they do that infrequently and keep usi

maybe OpenAI have a terribly inefficient data ingestion pipeline? (wild guess) basically taking in new data is tedious so they do that infrequently and keep using old data for training.