i'm still waiting for someone to convincingly explain to me what actual problem large language models are solving that humans couldn't do better and faster with a few simple tools.