@woog09 5/5
For the record, I wrote this entire response in my native language, expressing my own thoughts, and then used AI to translate it into English.
Does that make this comment AI slop too?
@woog09 4/5
I’ve seen many researchers on X promoting these detectors lately, and I find this trend remarkably shortsighted. Please think more carefully about the implications. AI writing detectors can never be a reliable measure of research originality or quality.
@woog09 3/5 A researcher can develop an entirely original idea, design the methodology, and conduct all experiments, then use LLM to write the paper in polished English. LLMs have simply made traditional English editing dramatically cheaper and faster. Yet your detector labels AI slop.
@woog09 2/5
Second, AI detectors cannot distinguish between using AI to polish academic writing and using AI to generate the research itself. This creates serious bias against non-native English-speaking researchers, who often rely on language assistance to communicate original research
@woog09 1/5
Your classifier is yet another piece of AI slop, and it risks introducing serious bias into peer review. As you acknowledge, its accuracy is far from reliable. How such classifiers produce large numbers of false positives is a basic undergraduate Bayes’ theorem problem.
@DrKarleenG What if the authors are non-native English speakers who used AI only for grammar correction or language polishing? Would rejecting the paper based on an AI detector then disproportionately disadvantage non-native authors?
@Handers27600480@mungtungmun 다른 나라 통계에서 비교군이 근로자가 아니면 비교군이 잘못된거지만 저기에서는 모든나라가 근로자대비 소득수준을 보여주므로.전혀잘못된게아님. 저통계치는 각나라별 근로자대비 의사의소득이 얼마나 높은지를 보여주려는거 적절한 통계임