I was a TA for a community college psychology class once and one year at least 75% of the class turned in a very similar paper on anorexia. That wasn't the subject of the assignment. And I never found out why that particular paper was submitted by so many (this was 20 years ago so all I could guess was it was a previous students paper or a paper from another class they tried to get to do double duty as a paper in this class). It was SO stupid. I don't know what was worse, knowing so many people cheated or the fact that they were too dumb to realize that it would be caught when grading papers.

It's always the same story, because people don't use the very specific instructions given to chat gpt when you want something original. You say: 'write a story as if you were a (journalist/copywriter/fantasy book writer/reporter/columnist), write it in everyday normal English, avoid words that might make it sound too much like an ai generated story, use not more than 500 words, etc. The more specific you are, the more it looks like something you wrote. Especially when you let it read something what is typically you. Instead, people just say 'write me a story'. And then Elara pops up.

This is why we make the kids do any work that is for marks IN CLASS. That way they can't use AI at all.

And this is why we have so many functionality illiterate 🙄 😒 😤

AI will be the end of us. Watch Terminator Genesis movie, it's scary 😨 😳

Man, this teacher is a human AI detector? She better clone herself.

Had a biology teacher in high school - on his final he wrote the instructions on the blackboard and verbally said "read all of the instructions before starting your test". The instructions were about a page long, and about 3/4 of the way down the page was this line "If you put your name and date on this test and don't fill out any other information you can turn it in right now for 100% credit. But if you have filled out any other information/answers you must complete the test and have it graded." Only myself and one other student read all the instructions and turned in our tests about 5 minutes after the class started.

Got the same story, different name.

"Looks like the teacher found a cheat code to unlock 'AI detector mode'! Now students are sweating harder than the AI trying to generate original content!" 😆

I am a college English professor and have had to deal with Internet plagiarism for decades. My solution is to make all students write an in-class essay on the first or second day, and make copies of all the essays. That way you have a benchmark against which you can compare their later take-home assignments. You can reasonably prove the plagiarism.

Why not run suspected AI work through a couple of AI detection tools? That should provide enough evidence to challenge the suspect submissions.

Just asked chatGPT to "tell me a story" - no Elara-tale, but one about an orange cat flying a rocket.

I used to teach anthropology at a university. It was before the days of AI, but super easy for students to find essays on-line, or simply cut and paste from multiple sources. A lot of my peers relied on TurnItIn to screen for plagiarism, but I couldn't be arsed and instead made sure essay assignments had enough unique demands that it was easy to spot the cheaters. (I'd usually require a course specific lecture to be incorporated, plus a few assigned readings, and never repeated those in subsequent terms). I'm pretty ignorant about AI, but I wonder if an essay question with similarly esoteric demands wouldn't still work?

What is a "specialized high school"? One of the comments used the phrase enough times to be annoying.