Since ChatGPT broke the internet last fall, everyone had something to say about AI, but none as much as teachers. In the first month since its arrival, educators bemoaned the end of writing assignments and complained that students could now plagiarize without being discovered. Almost immediately, detection tools started appearing designed to uphold academic integrity and catch cheaters. This spring, Turnitin developed its own AI detector, making it available to over two million teachers and claiming 98% accuracy.
However, many independent investigations, including student activist groups and high-profile journalists, have been showing mixed results for Turnitin’s product and free detector tools available online, such as GPTZero or OpenAI’s own Classifier.
Of course, it might bother hard-working students to learn that some fake papers generated with AI by their less scrupulous peers fly under the radar and get them good grades for nothing. However, more disturbingly, they might find themselves in a situation where their proper work receives a “false positive” from a detector, and their integrity is questioned.
How likely is your hand-written essay to be marked as AI-assisted, and how can honest paper writers prove they didn’t use AI assistance to create the work they submit? In this article, we will give you a few tips on demonstrating you’ve written your paper yourself when your instructor suspects you of cheating.
Why detectors aren’t reliable
First of all, none of the existing detectors claims to be 100% accurate. All they can show is statistical probability. Even a combination of several detectors cannot definitively answer whether a piece of text is human-written or AI-generated. Sometimes, different detectors provide widely different scores, which is very confusing. This, however, doesn’t prevent some instructors from perceiving this tentative probability assessment as hard fact and giving a student 0% for their work with an accusation of academic dishonesty to boot.
Despite claiming low inaccuracy rates, detectors are playing a catching-up game against the growing sophistication of the Large Language Models (LLMs) such as GPT-4 and Bard. One of the ways for detectors to stay relevant is by raising the bar for the text to be marked as human. However, this has an unfortunate side effect: more false positives. This means more students whose honest work is marked as cheating, which can have devastating consequences for young people, especially when submitting their college admission essays or personal statements for scholarship applications.
Another challenging aspect of AI detection is that it’s not as straightforward as plagiarism detection. There are no source documents to reference as proof. It means that the teacher is left with nothing more than statistical probability and gut feeling to decide whether the paper is original or not, leading to bias and unfairness.
Finally, the texts that detectors are supposed to sniff out are consistently average and predictable. Being based on large amounts of human-generated content, they produce the mean average, so to speak. However, some writers, especially new and unskilled writers, like many students, tend to be average. They, too, imitate texts they’ve read, especially in terminology-heavy subjects that leave little place for personal style, like economics, math, physics, etc.
What can you do to prove you’ve written your paper without AI?
As you can see, your essay being flagged as AI-generated isn’t all that improbable. This can happen to anyone, even to a hard-working student like you. When this happens, how can you prove your innocence?
First of all, keep your cool. As unpleasant as it is to be unfairly accused, making a scene won’t help your case. Take a deep breath, and count to ten. Get things into perspective: These accusations happen often and are bound to become even more frequent, so you are not the only one in this situation. Staying calm and rational will help you gather the evidence you need to prove your authorship.
1. Gather your sources
To defend yourself against false accusations of cheating, it’s vital to restore the timeline of your work. Collect all the sources and materials you used while working on your assignment: books, articles, videos, and other types of content that inspired you, such as samples from a free essay writer service or topical blogs. Don’t forget to document collaboration efforts, feedback from your peers that contributed to your final submission, emails, discussions in messengers, etc.
2. Provide draft history
Outlines and drafts are the most compelling evidence of your authorship. If you prepared your essay in Google Docs, that should be easy, as the app has the “Revision history” feature. It automatically takes snapshots of your drafts and stores them. In Word, if your file is stored in OneDrive or SharePoint, you can go back to previous copies too. If you use another word processor, it may depend on a particular version you have. Still, in some cases, you can retrieve past drafts.
If your processor doesn’t keep changes history, consider saving each iteration of your draft as a separate file for the future, or getting a browser extension like Draftback, if you use a browser version of a word processor.
3. Restore the timeline
When you have all this, organize your material chronologically, demonstrating how your work evolved. Document when you started working on your assignment, the milestones in its completion, and what changes you made after finding another relevant source or consulting your instructor. This way, you can prove that you built your essay step by step instead of copying it from an AI generator and pasting it into your word processor.
Of course, it seems frustrating and unfair that you should do all this legwork to prove you didn’t cheat, but this is the only way to deter the accusations.
4. Learn your rights
Understanding school policies on academic integrity and your rights within them is crucial for defending yourself. Consult official documents like course syllabus, student handbook, writing center memos, university code of honor, etc. Pay attention to requirements and deadlines for submitting appeals and requesting reviews.
Also, be aware of the consequences of academic dishonesty under your school’s policies to know where you stand and what is at stake.
5. State your case
Now you’ve gathered all the evidence, communicate with your professor or academic integrity advisor to defend yourself. Reach out via email, explain the situation, and present your evidence. Make sure to express your concern about the accusation politely and respectfully. Don’t accuse, don’t pose ultimatums, and don’t demand immediate decisions. Just describe the matter and clearly state that you didn’t use ChatGPT or other AI tools.
If email communication won’t be enough to thoroughly review the case, request a meeting or a video call. Be open to questions and encourage your professor to quiz you on the topic or ask about your writing process.
6. Be honest
In this case, more than ever, honesty is the best policy. Own up to your mistakes. If you asked someone to edit the final draft for you or if you’ve used AI for some portions of your draft, for example, to write a conclusion or to generate the initial outline, better say it as it is. If you are an ESL student and you’ve written the essay in your first language and used ChatGPT to translate it, fess up and point out that your ideas were genuine.
Tell about the challenges you faced while working on your assignment to provide the context and dimension to your case and demonstrate your commitment. Explain your thought process and the evolution of your ideas. However, don’t blow things out of proportion and stick to the facts.
7. Offer alternative verification methods
If all this wasn’t enough to clear your name, suggest alternative methods to test your knowledge and verify the originality of your work, such as an oral examination or re-writing the essay in class under observation to prove your knowledge of the topic and mastery of the course material.
Although AI detection tools aren’t perfect and got you into trouble in the first place, you might attempt to defend yourself by presenting scores from instruments other than the one used by your school, for example, Sapling, Content at Scale, or Copyleaks. Even if it doesn’t prove your innocence beyond a reasonable doubt, the mere discrepancy of scores should raise questions about the effectiveness of AI detection.
Finally, you can ask your instructor to compare the essay to your other written assignments to prove the consistency of your style.
Being wrongly accused of using AI tools like ChatGPT for academic assignments can be very frustrating. However, to defend yourself, you must stay composed: know your rights, gather evidence, calmly communicate with professors, and seek support.
Remember, AI detection isn’t definitive evidence of cheating. Assert that a statistical probability prediction by an AI tool isn’t enough to dismiss your work without giving you a chance to defend yourself. If you are innocent, maintain your work’s originality and offer ways to verify its authenticity. Uphold academic integrity, stay calm, and good luck.