Asking Chat GTP if it wrote an essay does not work if the screen is refreshed. It does not know it wrote it when copied and pasted into a new chat session.
Yes and no. While yeah they shouldn’t be given assignment after assignment that take ages to complete, if they don’t even have the discipline to write something then they won’t survive even a month in the real world
Ai detectors are very flawed, i wrote some actual stuff of my own, and then asked chat gpt to write something on the same stuff as well. i put both on their own into this and several other detectors and more than 50% of the results for all of them said my own writing was AI Generated, i tested it several times to see how serious the flaw is, my results ended up either proved detectors were even more flawed over my last results or remained unchanged i tried this again another time think maybe i just write alot like ai, so i had a friend also write some stuff and then compared it to actual ai generated stuff and results remained unchanged, either way, no mater how similar writing is to AI, the fact half the time ai generated stuff can come up as real and human written shows its flawed anyway
btw, if teachers suspected for some reason that a students writing was done via AI, and they check and it comes back as yes AI, how can we know that's right with how flawed it is, that student could fail because of a teachers mistrust and some faulty AI Detector
i use it to paraphrase and make parts of on essay based on a prompt, but beforehand i give my notes so that Ai can base the writing off of that. and the Ai detectors don't sense it as fake
The second method does not work (and the first one doesn't work either, for GPT-3). I tried to do that with an essay written by it, and it said it wrote it. So good so far. Then I gave it some essays I've written, without any help from the machine. Same result: it said that it wrote them all. It didnt :/ So I wouldn't count on the second method too much
Yes, there are more options available now which are more reliable. The technology to detect texts will probably always be behind the AI generators themselves.
Just in case people are still coming here: Any AI text 'detector' software has a high rate of false-positives. Some people naturally write in the form an AI may write in, depending on their writing abilities and what kinds of material they have read in their life. I, for one, am autistic and have read 90% technology stuff since I was 13 years old (I am now 39). My writing is probably coloured in that sense to be dry and monotonous. Don't rely on any artificial tool that can claim to detect AI!
I agree with you on this, as long as teachers show students the positives, we've nothing to worry about. Plus we need to get paid more I can second that!!
