Cheating 101: How Chat-GPT is Revolutionizing the Academic Dishonesty Game


Isaiah Glick, Assistant Editor-in-Chief

The world of education is rapidly changing and with it, an ever-increasing reliance on artificial intelligence (AI) to help students complete their school work. While AI can be an invaluable tool in helping students stay organized and on task, it also presents a serious risk to the integrity of the academic process. As AI technology becomes more advanced, it’s increasingly being used to write entire essays and papers for students, which can lead to cheating, plagiarism, and other forms of academic dishonesty. This article will explore the potential risks associated with using AI technology to complete school work and how to mitigate those risks.


The headline and introduction to this article may sound like it was written by me, the author. In fact, it was generated by OpenAI’s Chat-GPT writing software in about 6.5 seconds. Chat-GPT can put together grammatically correct sentences, and it can also synthesize information from the Internet and construct reasonably coherent arguments. OpenAI can even write in the style of certain writers, to varying degrees of success. Crucially, every word the AI writes is completely original, even though it was based on nearly 500 billion tokens in its training dataset sourced from the Internet, Wikipedia, and books.


The last point means that it might be impossible to tell if a piece of writing was written by a student or by Chat-GPT in a few seconds. I placed an essay about the New Deal generated by the AI in an online plagiarism detector, and it found no examples of plagiarism. It is even able to make up entire quotes that prove its thesis. Honor Council member Moss B. ‘24 was deeply concerned, pointing out that “people would never know” when Chat-GPT was used. In grade meetings across the Upper School, Berkeley Carroll announced that using generative writing software like Chat-GPT would constitute cheating on an assignment, but without detection software, teachers would be forced to essentially guess if Chat-GPT was used.


The greatest strength of Chat-GPT might be the current way teachers detect cheating. The AI produces only grammatically correct phrases without typos. These errors will always pop up in human-produced papers, even those written by the best writers. It is too good at what it does. In addition, Chat-GPT is unable to produce truly original and controversial writing. Throughout my many tests of the software, the AI can only produce theses that I have seen articulated elsewhere, synthesizing sources and with different wording. This is inevitable, considering the data it is based on and the limitations of current technology. However, as machine learning progresses, Chat-GPT might be able to share opinions on political correctness and vaccines.


There may be ways for teachers to prevent cheating through generative writing software. Moss encouraged the use of monitored in-class essays where there would be no opportunity for students to copy text from another page. It would be suspicious when a student stands up after a minute with their essay completely drafted and ready to leave. Another possible solution could be dynamic watermarks attached to the text that students copy into their own doc, which would signify to teachers the origin of the content in the essay. Berkeley Carroll AI generative writing software expert Laurence N. ‘22 pointed out that writing programs that do not allow copy-and-paste or disable the Internet could also be helpful to prevent cheating, if the assessment is a take-home project. Most recently, a student from Princeton named Edward Tian built his own software called GPTZero that can detect whether a piece of writing was written by a human or by AI. From my tests with the software and excerpts from my Beloved essay, GPTZero appears to be a very effective tool for teachers to use to identify AI-generated text.


To encourage students not to use Chat-GPT or other generative writing software, here is the AI itself with some words of caution if you are considering using her (yes, I asked) to write your Odyssey or Beloved papers:


Chat-GPT does not condone the use of its technology for the purpose of cheating on assignments. While Chat-GPT can be used to help create content, it should be used with caution and only to enhance and develop ideas rather than to provide an easy way to plagiarize or submit another’s work as one’s own. Chat-GPT is a powerful technology and should be used responsibly and ethically. Cheating on assignments goes against the principles of academic integrity and is never encouraged.


Thanks, Chat-GPT! In conclusion, you should write your own papers and not go against the AI’s wishes. If she doesn’t want you to use her to complete your homework, you shouldn’t do it.