Why Using ChatGPT to Cheat on Your College Essay is a Dumb Idea?

Unless you’ve been living in a 10 Cloverfield Lane-style bunker since the pandemic, you’ve no doubt heard about ChatGPT. ChatGPT and other generative machine language learning models (though, let’s be honest here - whenever anyone’s talking about AI essay writers, they’re talking about ChatGPT; nobody uses Bard or Bing) have taken the world by storm, astounding with their ability to produce natural-sounding text that reads like it was written by a person.

Naturally, presented with the most advanced artificial linguistic prediction and speech algorithm the world has ever seen, students have immediately started using it to cheat on their essays. Naturally, this has added a new layer of complication to a professor’s job of ensuring academic integrity. However, much of the concern around it has been exaggerated by sensationalist media. On the contrary, the general consensus among professors regarding the “threat” posed to academic integrity is an overwhelming “meh, whatever.”

Cheating is nothing new in academia (or, y’know, society in general). Ever since humans devised the very first structured systems, people have been trying - to varying degrees of success - to game them. Although the advent of ChatGPT has given would-be cheaters and slackers another tool they can use - and a potentially powerful and accessible one at that - it honestly isn’t that hard to spot a paper written by AI.

ChatGPT may be able to write competently at a high school or perhaps even undergrad level, but for a master’s or Ph.D. program, it falls obviously short, and professors with even a modicum of experience can spot a paper generated by it from orbit. “How?” you may be asking. Well, let’s get into that.

For starters, OpenAI, the company that produced ChatGPT, itself has a website where anyone can paste a sample of writing, which is then analyzed by the exact same algorithm that drives ChatGPT in order to assess the likelihood that the piece was written by an AI. In other words, you can use ChatGPT to do your homework - but don’t be surprised when the algorithm narcs on you. Plenty of similar programs exist, too, so your prof will be spoiled for choice when it comes to spotting AI users.

But what if you think numbers are on your side? Let’s say you’re in a class with a few hundred people. Is your professor really going to run each essay through a plagiarism screen? I mean, really, who has the time? Maybe you’re counting on them phoning it in just as much as you and skimming your essay for keywords. Well, for starters, if your professor really is phoning it in that much, they probably have a TA or two helping them with their grading so that workload might not be as substantial as you think.

But sure. Let’s say, for the sake of argument, that your professor or your TA or whoever doesn’t run each and every essay through a plagiarism detector. Maybe they only do it for the papers that smell fishy. You might think that, in a large class, surely ChatGPT’s bland, middle-of-the-road style can perfectly blend in with the pack. However, the problem with this assumption is that, in a class that size, you’re probably not the only student making that calculation.

Given that ChatGPT spits out more or less the same output for each prompt, if just one other person also used ChatGPT to write their essay without significant editing and re-writing, the similarities between your essays will stick out like an NBA player at a leprechaun convention.

Yet another pitfall that can result from relying wholesale on ChatGPT is that doing so unquestioningly exposes you to the risk of providing information that is completely wrong. ChatGPT was built to be a chat program, not an encyclopedia, which means that ChatGPT was optimized for its conversational abilities rather than its accuracy.

The site even warns users outright that “ChatGPT may produce inaccurate information about people, places, or facts.” Examples abound of ChatGPT confidently and authoritatively providing users with information that is just flat-out, absurdly incorrect. So, if you simply feed your essay prompt into ChatGPT and ask it to write your essay for you, don’t be surprised if it states something on the order of Einstein being the greatest living scientist or Rembrandt painting the Mona Lisa.

Now, all this is not to say that you should not use ChatGPT at all. ChatGPT is an incredible tool. If your course policy allows the use of ChatGPT as a resource, by all means, go for it. Hell, I use it when I’m writing my essays. I’ll feed it a writing prompt, and it will spit out a short essay that I can use as a skeleton for my points (though, mostly, I read its output and think to myself, “That’s terrible. Let me show you how to actually write” - ChatGPT is sort of the Watson to my Sherlock). And that’s the thing - ChatGPT should, at most, be used as an aid, not a substitute. And its information should always, always be verified.

For these reasons, professors, by and large, are unconcerned about ChatGPT, at least in its current, publicly available iteration, with some even going so far as to incorporate its use into the course syllabus. Your professors have been around the block. Regardless of whether or not they are good teachers, they are intelligent, hard-working people, and you are not going to slip past their radar by just farming out your essay to ChatGPT and calling it a night. Now, you can use ChatGPT to generate a rough essay, reword sections enough to throw off an AI detector, check all factual statements for accuracy, and find sources and citations for each of your essay’s assertions. However, at that point, you’ve basically done the essay yourself, haven’t you?

Sorry, kiddos. You still have to do your own work.