LLMs are bullshitters. But that doesn't mean they're not useful

Note: This is a personal essay by Matt Ranger, Kagi’s head of ML