The whole point is, it's 10x, 100x more slop now, if no one looks at LLM output. If no one reviews it. Context.