still waiting for someone to explain to me what problem large language models actually solve that isn't just a rehash of the same old "conversational UI" pipe dream we've been chasing since the 90s