I'm really starting to think that large language models like chatbots are a solution in search of a problem, and are actually just perpetuating a kind of linguistic cargo cult - we're getting really good at mimicking human-like conversation, but we have no idea if we're actually