Anecdotally, but I know a number of people who do university assignments for money. Many of the clients for the folks I know are at the university level are folks with poorer then average English language skills and are usually in intro writing courses. I'd be terrified if I were one of them right now.
GPT-3 would be a godsend for cheaters, but still requires a human to jump in and rewrite whole sections.
No, if you want to REALLY want to cheat using AI, you should most likely utilize either 1. Abstractive Summarizers (e.g. Pegasus) or 2. paraphrasing tools (e.g. like at https://quillbot.com/). I believe that Quilbot is primarily powered by MLMs like BERT rather than CLMs like GPT-2 (but someone who works there can enlighten me more).
Copy and paste a text that you want rewritten in your own words (e.g. the ideas of a really smart individual), and then it rewrites it using totally different language but preserving the same meaning. (old) Plagerism detection tools don't work and hell, it's not hard to fool the never ones. You can try tools for detecting if something is AI written by a particular model and weights (e.g. to prove if they used GPT2-Medium), but if I fine-tuned those same weights, than proving it was plagiarism will become exceedingly difficult.
Welcome to the brave new world of cheating. Also, techniques like this are coming to a CS department near you (in the form of source code generation powered by NLP models).
I feel like we're overestimating the value of generating text and selectively picking out things that sound good. This is up for debate, but you'd probably do better to read actual authors and pick things up from them and maybe making your own changes. The time spent parsing the output to see if it's good could be spent on thinking about something to write?
Would be neat to try and publish books with a percentage of AI generated text and see how well they do. Maybe there's a sweet spot for productivity.
GPT-3 would be a godsend for cheaters, but still requires a human to jump in and rewrite whole sections.
No, if you want to REALLY want to cheat using AI, you should most likely utilize either 1. Abstractive Summarizers (e.g. Pegasus) or 2. paraphrasing tools (e.g. like at https://quillbot.com/). I believe that Quilbot is primarily powered by MLMs like BERT rather than CLMs like GPT-2 (but someone who works there can enlighten me more).
Copy and paste a text that you want rewritten in your own words (e.g. the ideas of a really smart individual), and then it rewrites it using totally different language but preserving the same meaning. (old) Plagerism detection tools don't work and hell, it's not hard to fool the never ones. You can try tools for detecting if something is AI written by a particular model and weights (e.g. to prove if they used GPT2-Medium), but if I fine-tuned those same weights, than proving it was plagiarism will become exceedingly difficult.
Welcome to the brave new world of cheating. Also, techniques like this are coming to a CS department near you (in the form of source code generation powered by NLP models).