ChatGPT Cheats

[Neapolitan Ice Cream]

Various online courses have been blighted with ChatGPT cheats, making the online AI tool do their work for them. This is annoying for anyone who has actually done the leg work. Interestingly, Universities/Accredited Bodies now have an "AI Detector" tool that will even pick up AI generated text even if you paraphrase it. Score one for the awarding bodies, big problem for the cheats!

 

[fla4fun]

Various online courses have been blighted with ChatGPT cheats, making the online AI tool do their work for them. This is annoying for anyone who has actually done the leg work. Interestingly, Universities/Accredited Bodies now have an "AI Detector" tool that will even pick up AI generated text even if you paraphrase it. Score one for the awarding bodies, big problem for the cheats!

I wonder how they pick it up if you paraphrase the generated text? It seems like that would be more difficult to find accurately. I would hope that no diligent scholars get caught up with the cheaters. It would be a good idea to keep a thorough bibliography and source list in case you had to prove you really did the work. I assume AI doesn’t provide any of the sources it used to compile the generated text, correct?

I am glad they are finding ways to crack down on AI cheaters, but I do want them to be accurate about it and have an appeals process in place if a student is falsely accused.

 

[Neapolitan Ice Cream]

I wonder how they pick it up if you paraphrase the generated text? It seems like that would be more difficult to find accurately. I would hope that no diligent scholars get caught up with the cheaters. It would be a good idea to keep a thorough bibliography and source list in case you had to prove you really did the work. I assume AI doesn’t provide any of the sources it used to compile the generated text, correct?

I am glad they are finding ways to crack down on AI cheaters, but I do want them to be accurate about it and have an appeals process in place if a student is falsely accused.

I agree. Well, not much more to add really!

[ETA] Just noticed, ChatGPT has just been updated to a far more human version, far more intricate and detailed. This could pose a threat to the detectors.

 

[hrhdhd]

I don't need an AI detector. The stuff I get that's generated by AI is clearly not the work of a college freshman, which is the level I teach. And my students don't use AI to have something to paraphrase; that would mean additional work, which is what they are trying to avoid in the first place. They simply copy and paste.

 

[jazzsmith]

I used to think it is impossible to detect a paraphrased text ... It's definitely frustrating to see people using ChatGPT to cheat on their coursework. It's good to know that universities now have AI detectors to catch this. Even using a free writing AI generator won't help if the text is easily detectable. This should encourage students to do their own work and learn the material properly. It's a fair move to maintain the integrity of education and reward those who put in the effort.

 

[Neapolitan Ice Cream]

It is impossible to detect a paraphrased text

Ah, QuillBot. Interesting one that. Rather clever IMO. It does tend to repeat itself though. But if you carefully proof read everything that it gives you and iron out all the kinks, it could thwart the detectors...

 

[kdonnel]

I don’t have an issue with students using AI. It is just another tool. In my work I don’t know everything. I have to research and find answers every day to help me solve problems at work. Knowing how to craft a great AI prompt is a very valuable skill. I would rather work with or hire people who have the skills to find answers and get work done rather than people who can only operate with their limited list of memorized facts.