I think "did AI write this?" is the wrong question. "What judgment did it pass?" is better.

There is a lot of discussion right now about AI slop: generated posts, articles, charts, summaries, and comments that look complete but feel empty. I think the useful distinction is not simply "AI wrote it" versus "a human wrote it." Humans make generic work too. AI can also be genuinely useful when it gets real context, examples, sources, constraints, and review. The problem starts when the output gets accepted without a human boundary for taste. By taste I do not mean aesthetics or personal preference. I mean judgment: Who is this for? What claim are we willing to stand behind? What evidence or context makes it specific? Which default examples or phrases are too generic? What should we refuse to say, even if it sounds polished? That last question catches a lot. Many weak AI drafts are not obviously false. They are just unearned. The tone is confident, but the examples are generic. The structure is tidy, but the actual decision is missing. The paragraph sounds reasonable, but nobody has decided whether it belongs in the final work. So I am starting to think "did AI write this?" is often the less useful review question. A better question is: What human judgment did this pass? For a chart, that might mean: is this the right comparison, or just the default plot? For a memo, that might mean: are the caveats and decision trace visible? For code, that might mean: can the team explain and maintain the change? For public writing, that might mean: does the piece have a real claim, a specific example, and a reason to exist? Curious how other people handle this. If you use AI for writing, reports, charts, code, or internal work, what review rule catches the most "looks fine but should not ship" output?