Bring us your LLMs: why peer review is good for AI models