If you work in education, please please please educate your colleagues that "AI detectors" like TurnItIn are all scams and none of them can reliably detect AI-generated content. www.reddit.com/r/Teachers/c...
One of my son's professors had a note on a recent assignment: "If I believe you've used AI to generate the paper I will quiz you directly on the content." That seems like the only reasonable recourse, no?
I don't know if it's gotten better, but I remember turning papers in that Turn It In would ding as plagiarized. It would highlight the sections I allegedly plagiarized, which were inevitably always quotes that I had quoted and cited correctly 🙄🙄🙄
Teacher: "I didn't want to do my job, and so used an AI tool to do it for me. It produced results which seemed unlikely but verifying them would've been hard, and so I insulted an entire class and irrevocably undermined their faith in me instead."
Yeah, I had a student submit something via a platform I currently use for written work & the supposed built-in "AI detector" flagged the submission, which turned out to be almost entirely a copy-&-paste job from lecture materials that I wrote myself (which is of course a whole other problem).
It's possible to design LLMs to watermark their outputs**, and then have a detector that detects those, but that requires the cooperation of the LLM developer. ** One algo is to divide tokens into two bins, A and B. Boost the probabilities of A, then B, back and forth on alternating token gens.
Missing in this discussion is what kind of teacher takes this much glee in giving their students failing grades? Sounds like a failing teacher
Based on his Reddit comments, this is a guy who teaches English lit to 11th graders in Broward County, FL, watches cable TV on his laptop at work, had gallbladder surgery on or about March 21 and is planning treatment for gynecomastia. Sure hope no one figures out who he is and reports him!
I suppose it’s good for kids to learn early how often authority is arbitrary, cruel, and stupid.