Chat GPT and Plagiarism: Separating Fact from Fiction


I was recently on Reddit and there seems to be a lot of confusion about Chat GPT and plagiarism. So let’s start with the simplest question, “Can Chat GPT plagiarize?”.

Chat GPT generates text based on patterns it has learned from training data. It produces non-original responses when asked to fit a specific pattern, such as completing the phrase “Roses are Red” with “Violets are Blue.” However, GPT generally produces responses that are not plagiarism.

Chat GPT and plagiarism

So in general the responses that you get to a text prompt from Chat GPT are not detectable as plagiarism. However, systems at schools like “Turnitin” will quickly be able to detect some Chat GPT texts. Why? Let’s take a look.

Table of Contents

How does GPT generate text? ( The Secret Sauce )

Chat GPT and text prompts

  • Text prompts are a way to ask Chat GPT a question or give it a specific topic to write about.
  • Chat GPT uses artificial intelligence to analyze the prompt and generate an answer or text based on its programming and previous inputs.

So here is the problem, if I ask Chat GPT please write me a 1000-word essay on George Washington and you ask it the same thing or something similar do we get the same text?

Photo By George
Text #1 For George Washington Essay

Let’s try that again asking the same question a little different way.

Text #2 for George Washington Essay

Take a look at the first sentence of the second paragraph, they are identical!!

If you use a similar text prompt to someone else you will get a very similar result from Chat GPT. That means that you will get caught if you try to turn in the George Washington essay above because someone probably already did.

Why does that happen? Let’s take a look at the outlines of these essays, (you aren’t seeing double they are almost identical).

Outline For Essay #1

Outline for Essay #2

If you use a similar text prompt as everyone else you will get a very similar result from Chat GPT. In effect you are doing the same this as asking it to finish, “Roses are Red”, it’s going to always answer with “Violets are Blue”.

CSC

Can GPT generate original content?

Can you get unique content from Chat GPT? Yes, you can! ( See my article here )

You can take a look at my article but the heart of it is this.

You need to feed Chat GPT a unique essay outline to get out a unique essay

Yep, you are going to have to do some work.

You can edit the George Washington essay for instance in Word, add your own points, remove or ad bullet points, put in more facts, etc.. Or you can just make your own outline, to begin with.

Let’s take a look at how that would work with the George Washington essay.

Modified George Washington Outline

Chat GPT Output

Keep in mind that

  • Chat GPT does not give you any references
  • Chat GPT will make facts up upon occasion
  • Double-check everything that you do not personally know

For instance, I am not 100% sure that George Washington was offered a commission in A British regiment. I would need to check on that. Note: That was not part of the outline that I gave Chat GPT it made it up.

  • I would also need to note a reliable source for that and any other facts in the text
  • Chat GPT is NOT a reliable source
  • Any source Chat GPT suggests to you is also suspect, CHAT GPT MAKES UP SOURCES
  • Check everything using Google, books, articles, etc.

What is Plagiarism?

Plagiarism is the act of taking someone else’s work or ideas and passing them off as your own, without giving proper credit to the original source. This includes:

  • Copying and pasting text from another source without citing it
  • Paraphrasing someone else’s work without giving credit
  • Using someone else’s ideas or research without acknowledging where you got it from

Plagiarism is a serious ethical offense because it involves stealing someone else’s intellectual property and pretending it’s yours. In academic settings, it’s especially important to produce original work and properly attribute any sources you use in your research or writing. If you’re caught plagiarizing, the consequences can be severe, including:

  • Academic penalties, like failing a class or getting kicked out of school
  • Damage to your reputation
  • Legal repercussions

To avoid plagiarism, make sure to properly cite any sources you use in your work. This can be done through in-text citations and a list of references or works cited at the end of your document. It’s also important to understand and follow the citation guidelines of your field or discipline since they can vary.

This section on Plagiarism was written by Chat GPT.

Is the Text You Are Creating with GPT Plagiarism or Cheating?

If you are simply using simple text prompts with no input or original ideas from you, you are committing plagiarism by using Chat GPT’s responses as your own work.

However, if you are using Chat GPT to help you write and contribute your own ideas and original content, you are likely not committing plagiarism.

Why?

If a student uses Chat GPT or any other tool to generate content based on a unique, detailed prompt, the resulting material should be considered the student’s own work if it is properly attributed to the tool and to any other sources that were used.

That of course is my own personal opinion, your teacher or audience may not agree.

It is ethically important to properly credit and acknowledge any assistance you receive from Chat GPT or other sources in your writing. In a classroom setting, it may also be necessary to turn in your detailed, unique outline to show the originality and effort put into your work.

Keep in mind the limits of Chat GPT and other natural language processing tools. These tools are designed to generate text based on a given prompt, but they may not produce completely original content on their own even with an original text prompt.

I would suggest turning in the detailed unique outline that you used as a text prompt to your teacher to make it clear that you are doing your own work. I view Chat GPT as a calculator for writing. It is very useful to do some operations but if you don’t know what you’re doing you’ll get nonsense. In programming, we call that, Garbage In —> Garbage Out.

Keep in mind you need to learn how to write for yourself as well. Essay tests happen in the classroom too without a computer to help you out. Learning your own voice and how to create a flow between paragraphs will only help you.

ChatGPT Plagiarism - A Lawyer Explains (w/ChatGPT)

So what about Copyright and using GPT?

Some interesting things have happened with AI recently. An artist was given a copyright for a graphic novel that was largely drawn by AI. She described her use of the tool as assisting her to create the novel.

Now here’s the problem, after this event made the news the copyright office decided to as the “First AI-Generated Work to Receive a Copyright”, the copyright office sent Kristina a letter potentially revoking the copyright.

I was asked to provide details of my process to show that there was substantial human involvement in the process of creation of this graphic novel,” Kashtanova explained by email.

Kristina Kashtanova

So how much of a work has to be “created” by a human being to get a copyright? That seems to be the standard that the copyright office is starting to use.

How much of your essay needs to be organized and created by you for it not to be plagiarism? That’s the question that teachers will have to answer in the future.

My suggestion is that you submit detailed unique outlines that you have created to help teachers see that you are doing the work and also let them know that you are using Chat GPT to help you do the writing.

For something like this blog post I would need to retain notes about each section and what Chat GPT wrote and how if it came to defending the copyright.

Note: The section on plagiarism was written by Chat GPT as noted, (based on an outline provided by the author). most of the rest of the text was written by the author directly 90+ percent.

Chris

Chris Chenault trained as a physicist at NMSU and did his doctoral work in biophysics at Emory. After studying medicine but deciding not to pursue an MD at Emory medical school Chris started a successful online business. In the past 10 years Chris's interests and studies have been focused on AI as applied to search engines, and LLM models. He has spent more than a thousand hours studying ChatGPT, GPT 3.5, and GPT4. He is currently working on a research paper on AI hallucinations and reducing their effects in large language models.

Recent Posts