I didn’t mean to argue against the usefulness of LLMs entirely, they absolutely have their place. I was moreso referring to how everyone and their dog are making AI assistants for tasks that need accurate data without addressing how easy it is for them to present you bad data with total confidence.