LLM’s can search the web or query a db with the law.
That helps a bit but doesn't counteract training. The LLM still reasons through the web query/db response with it's trained parameters
That helps a bit but doesn't counteract training. The LLM still reasons through the web query/db response with it's trained parameters