LLMs are not a replacement for human judgment, let's not pretend like they're going to magically fix all our problems just because they can generate some decent text.