LLMs are still not great at generating SQL. If Google had a breakthrough, it should be on the bird brain benchmark - (A Big Bench for Large-Scale Database Grou

LLMs are still not great at generating SQL. If Google had a breakthrough, it should be on the bird brain benchmark - (A Big Bench for Large-Scale Database Grounded Text-to-SQLs) https://bird-bench.github.io/ https://bird-bench.github.io/ At the moment GCP are at 76%, humans are at 93%.