March 17, 2023
ChatGPT and The Battle Against Plagiarism
Garrett Burnett
ChatGPT (Chat Generative Pre-Trained Transformer) has taken the world by storm. This online, AI program has a wide range of functions that include, but are not limited to answering questions, generating and analyzing text, translating between languages, and chatting. Its easy-to-use interface allows users of all ages to interact with the platform and find solutions to their varied problems. However, ChatGPT has raised issues in educational communities as it can write entire essays for students based on any given prompt. If given a prompt and a word count, ChatGPT can output an essay in less than a minute. The biggest concern voiced by educators relates to plagiarism. I asked ChatGPT for its advice on how to avoid plagiarism and on the current AI versus plagiarism debate.
ChatGPT Response
In the world of online submissions, essays at all levels are often submitted into online plagiarism checkers that are able to quickly scan the papers for commonly used phrases and plagiarized material. ChatGPT itself says that it can guide users around these plagiarism checkers. I asked ChatGPT how it can bypass plagiarism checkers, and it quickly spit out four tips to avoiding them:
1. Use unique language: ChatGPT can avoid using common phrases or sentences that may be flagged as plagiarized. By using unique language, ChatGPT can create content that is less likely to be flagged as plagiarized.
2. Paraphrase content: ChatGPT can paraphrase content by rephrasing the input it receives from the user in its own words. This can help ensure that the content generated by ChatGPT is unique and original.
3. Cite sources: If ChatGPT is using information from a specific source, it can cite the source to avoid any plagiarism issues. By acknowledging the source of the information, ChatGPT can avoid any potential plagiarism concerns.
4. Use a plagiarism checker: ChatGPT can use a plagiarism checker to check its responses for any potential plagiarism issues. This can help ensure that the content generated by ChatGPT is unique and original.
ChatGPT Discusses Plagiarism
When I asked ChatGPT to discuss the current, tense situation between ChatGPT and the education system, it suggested several possible solutions as it relates to plagiarism. First, it suggests, as previously discussed, the utilization of a plagiarism checker to check work for plagiarized material. However, ChatGPT also noted that it can use a plagiarism checker itself to review its own responses when I asked it how to avoid these plagiarism checkers. ChatGPT also did not pass up this opportunity to boast about the sophistication of its AI. It said that “the model can generate text so similar to human-written text that it can pass undetected by even the most sophisticated plagiarism detection tools.” However, this does not seem to be accurate.
At the beginning of February, Turnitin, one of the most widely used plagiarism detectors, announced that their system could determine whether a submission was written by AI or not. According to the February 13, 2023 press release, Turnitin can identify “97 percent of ChatGPT and GPT3 authored writing, with a very low less than 1/100 false positive rate.” It seems that ChatGPT may be giving itself more credit than it deserves.
Overall, there are an increasing number of AI powered plagiarism checkers becoming available for public use. Additional examples include: GPTRadar, Originally.AI, GPTZero, CopyLeaks, ZeroGPT, and Winston AI. Educators are not only able to implement all of these checkers to review the work of students for plagiarism, but they are also able to compare the papers of their students. If students are inputting the exact same prompt into ChatGPT then it is very likely that they will receive strikingly similar outputs. ChatGPT may have outsmarted the education system from its release in November 2022 until February 2023, but I think that its days of providing material that can flawlessly and consistently bypass plagiarism checkers are gone.
Ethical Concerns of AI
ChatGPT continued on to discuss the incorporation of “ethical guidelines and principles into the development and use of AI language models” as a way to combat the use of ChatGPT and other similar models to produce plagiarized material. This suggestion has opened the door to a new series of questions. The first of which regards the party or parties that are responsible for creating these new guidelines. ChatGPT did, unsurprisingly, have a response to this question when prompted. ChatGPT suggested that “the development of guidelines for the use of AI in schools should involve a broad range of stakeholders, including educators, administrators, policymakers, and experts in AI and education.” However, it described the role of educators and administrators as facilitators of identifying the beneficial role that AI can have in the classroom rather than using them for their expertise in the overreach of AI in the lives of students.
Overall, the use of AI in schools is an incredibly slippery slope. Educators will continue to battle against these chat bots as AI becomes smarter and more capable of bypassing the most advanced plagiarism checkers. Although there is a wide potential of positive use of AI in schools, and it would be beneficial for students to learn how to utilize AI, there is a real need for students to learn how to develop their own written communication skills. In the end, AI like ChatGPT has the potential to strip students of these learning opportunities if educators, but it could also greatly enhance their learning if used correctly.