Are you an aspiring judge/reviewer? Then here are all the tips you need to strengthen your review writing!
---
If you are reading this story on any other platform other than Wattpad you are very likely to be at risk of a malware attack and reading a...
I mean, just think of it this way: what if you run ten paragraphs through a detector and it comes back 20% AI-written? Seriously, why would someone use AI for two paragraphs but then not for the other eight. Doesn't make sense, right? Sure, it can happen, but at that point, it isn't even really an "AI-written" work and would be up for debate about if it's AI hybrid or just AI assisted, and you can debate that all day, but I have no interest getting into it. Let's just stay on the topic of detectors.
Think of it this way: if a plagiarism detector says your text has 20% plagiarism, but the things is flags are "I went to the grocery store" and other basic things like that, you wouldn't say it's actually plagiarism but rather a false flag, right? Same applies to low scores in AI detectors. Keep in mind humans, like myself, actually train these AI, so it's extremely possible for false positives, and they happen more than you think.
This isn't me saying this, by the way. My employers told me this (DataAnnotation), not to mention there are several AI detectors who have come out and said that their detection is based on likelihood, not "This is 20% AI written." Here is a screenshot from the Undetectable Blog, and Undetectable is a very popular AI detector:
Oops! This image does not follow our content guidelines. To continue publishing, please remove it or upload a different image.
The blog title is Can AI Detectors be Wrong? (Learn How To Avoid AI Detection). I would leave a link, but Wattpad links are about as reliable as AI detectors, so I will instead just leave the headline so you can direct yourselves there if interested.
The lower the percentage of AI is detected, the higher chance it is a false positive since it doesn't make sense for one or two lines to be AI written and the rest human written.
Keep in mind AI detectors are only able to boast high accuracy percentages due to loopholes where the only time a detector can be "wrong" is if it scans 0% or 100%. If an AI detector flags a human-written work as 100% AI, then it's wrong. But if an AI detector flags a human-written work as 99% AI, then it's right because of the 1% chance it's human-written. It technically said there was a chance for it to be human-written, so it's not "wrong."
That's why AI detectors are able to boast high accuracy ratings: it's a percentage scale, so it's technically only "wrong" when it flags a human-written work as 100% AI, or an AI-written work as 0% AI.
So why am I mentioning this in a book about reviews? The reason is because mindfulness is important when you're reviewing books, and please don't spread misinformation about AI and become over-reliant on AI detectors.
Along with this, AI detectors can have a bias against non-native English speakers. Undetectable says this about the topic:
Oops! This image does not follow our content guidelines. To continue publishing, please remove it or upload a different image.
