Can Teachers Tell When You Use ChatGPT

If you are a student, whether in school or university, you are bound to have asked yourself – can teachers tell when you use ChatGPT? It’s a valid question, since any assignments you get are for you, and your tutors would be disappointed (putting it mildly) to learn that you didn’t finish them yourself. Still, everyone has a finite amount of hours in a day, so if you can be done with your homework in 5 minutes instead of 2 hours, and have no repercussions, why not do it?

Can Teachers Tell When You Use ChatGPT – Includes 5 Simple But Effective Ways To Bypass Detection

What is ChatGPT and Why Are Students Interested In it

ChatGPT is a tool that was made widely available to the public on November 30, 2022. It is developed by OpenAI, and it is a chatbot that answers questions. It is a generalized chatbot, so you can basically ask it any questions, and if the answer to it was available in it’s training data, it will most likely give you a very human-like answer. It was trained on books, articles and web pages. It can code, write poetry and yes, finish written assignments for students such as yourself. So if it can do that, why not use it? Because your teachers may not be happy if they find out, that’s why.

Can Teachers Tell When You Use ChatGPT?

The answer is… it depends. 

Can they 100% for sure know if the information you provide is AI generated? No, not really.

Are there tools out there that they may use to have reasonable suspicions whether or not the text was written by an actual human being? Yes. 

We are going to discuss how to bypass those tools, and look at some very real cases where students were accused of cheating by their teacher.

How Good Are AI Detectors and Can Schools Tell If You Use ChatGPT Generated Content?

The honest answer is that AI detectors are not that good these days. There are many cases out there where a student, or a group of students were accused of using ChatGPT content for their assignments, when in fact they wrote it themselves. It is an interesting situation, below you can see how an AI detector, GPT Zero, flags parts of the US Constitution as written by AI:

This goes to show that AI can cause a real big disruption in every field, even education, which by all accounts should really be the area that gets most enhanced by this technology.  On the one hand, you have teachers that are trying to preserve academic integrity, and make sure students learn something. On the other hand, you have students who are just writing out their thoughts, and can be accused of using AI without even having heard of it (example situation here).

As a student, you may be even in a situation where you’re trying to figure out how to make your written assignment not show as AI generated, even if you have written it all yourself! As if being a student wasn’t stressful enough!

So can teachers tell when you use ChatGPT? It is safe to assume that they can – when even a human written text is considered written by AI, I would advise to err on the side of caution.

How To Bypass AI Detection Of ChatGPT Content

The best way to make sure that your teacher can’t tell you use ChatGPT is of course not to use it at all – write the content yourself. Make sure you use a word processor that has history, and if accused of cheating – show them the edit history.

But if you’re reading this article, that may not be the best option for you, right? So how do you bypass the detection?

1. Edit the text excessively and use your own words in most parts

Simple and effective if you’re just looking to submit a few assignments this way, and save a bit of time if you’re in a hurry. Make sure you run your text through an AI text checker to ensure good results.

2. Rewrite the whole text you get from ChatGPT 

This is the most surefire way, but it takes the most time overall. You have the content and you reword it, and in the meantime you get to learn what you have written as well. It’s a win-win. You will most likely not save as much time as the other methods but you are almost guaranteed to pass the AI detectors. 

3. Prompt ChatGPT in such a way that it generates text that is not detected by AI detectors.

This is riskier in the sense that you may need a lot of time to come up with a good prompt. AI detectors use the fact that AI generated content is predictable on the level of words that come one after another, and has a predictable / common sentence structure (we will discuss how they work in a future article). Having that in mind, you can prompt ChatGPT with something like:

‘“[your request]. Write it using uncommon wording.” 

‘“[your request]. Write it using uncommon sentence structure.” 

This is a very crude way of formulating it, and the text generated this way sounds extremely unnatural. Unless you’re a person that may happen to speak that way, you will need to tweak the prompt.

4. Prompt ChatGPT multiple times, requesting amendments to the text.

You can use the following flow:

1. Request the text.
2. Test if it is detected as AI generated.
3. Tell ChatGPT to rewrite the text, with subtle changes, so it is not detected by AI.
4. Test if it is detected as AI generated once again.

5. Feed the AI text with your writing, and ask to write the assignment in that style.

This in my opinion is the best method if you already have some long form text that you have written academically. Just paste the text to the AI, or multiple such texts, and ask the AI to extract your writing style, and use it to write your new assignment. This way you have an assignment that is generated quickly, and your teacher can’t tell that you used ChatGPT, even if they are intimately familiar with your writing style.

How To Tell If Your Text Is Detected as ChatGPT Generated:

You can use a free tool that OpenAI made available to the public: https://platform.openai.com/ai-text-classifier

Note that teachers may use other tools that have different algorithms, so the above link saying your text is AI generated or that it is written by a human, doesn’t necessarily mean that your teacher will get the same outcome.

Conclusion

While using help from an AI with homework, you might have wondered, “Can teachers tell when you use ChatGPT to write your assignment?”. Hopefully this article has given you a bit of peace of mind when it comes to understanding what tools your tutors have at their disposal, and how to bypass them.

We are at a turning point in history, and AI is getting more and more powerful. I am at a point where I think that written homework will eventually not be used in the upcoming years, at least until there is a surefire way of verifying that the content is not AI generated. If you are a student and wrongfully accused of academic dishonesty, your education, and by proxy your future might be done with. If you are a teacher, is it worth risking this for an entire generation, and leave it at the hands of questionable AI detection software? And if you are a student, is it worth jeopardizing your grades for a little bit more free time? In the end, it is up to us each individually to decide.