How Good Is Turnitin at Detecting ChatGPT?

ChatGPT is a powerful AI chatbot that can generate natural language texts based on user inputs. It can answer questions, write essays, create stories, and even mimic celebrities. But how good is Turnitin, the popular plagiarism detection software, at identifying ChatGPT-generated texts? In this article, we will explore the strengths and limitations of Turnitin’s new AI detector and provide some tips for educators and students to deal with the challenges posed by ChatGPT.

Quick Links

What is ChatGPT?

ChatGPT is a language model developed by OpenAI, a research organization dedicated to creating safe and beneficial artificial intelligence. ChatGPT is based on GPT-3.5, a large-scale neural network that can produce coherent and diverse texts on almost any topic. ChatGPT is fine-tuned for dialogue, which means it can interact with users conversationally. ChatGPT can also follow instructions, provide feedback, and learn from human demonstrations.

ChatGPT is available for free at chat.openai.com, where anyone can try it out and see what it can do. Some of the examples of ChatGPT’s capabilities include:

Answering trivia questions
Writing summaries of articles
Generating lyrics for songs
Creating jokes and puns
Writing code and debugging errors
Translating languages
Tutoring in various subjects
Simulating characters for games

ChatGPT is not only impressive but also fun to use. However, it also poses some ethical and academic challenges, especially for educators and students who value originality and integrity in their work.

How Does Turnitin Detect ChatGPT?

Turnitin is a widely used software that helps educators and students check the originality of their work and prevent plagiarism. Turnitin compares submitted texts with a large database of sources, including academic journals, books, websites, and student papers. Turnitin also provides feedback tools that help improve writing skills and learning outcomes.

In February 2023, Turnitin announced that it had successfully developed an AI detector that can identify 97% of ChatGPT-generated texts, with a very low false positive rate of less than 1%. The AI detector uses state-of-the-art machine learning techniques to analyze the linguistic patterns and features of ChatGPT texts, such as word choice, syntax, semantics, coherence, and style. The AI detector also leverages Turnitin’s extensive database of human-written texts to compare and contrast with ChatGPT texts.

Turnitin’s AI detector is integrated into its existing products, such as Originality, Feedback Studio, and Gradescope. When a student submits a text to Turnitin, the AI detector will automatically scan it for potential ChatGPT content. If any ChatGPT content is detected, Turnitin will flag it with a red icon and provide a link to the source where the content was generated. Turnitin will also highlight the specific parts of the text that are likely to be ChatGPT-generated.

Turnitin’s AI detector aims to help educators and students maintain academic integrity and uphold high standards of quality in their work. By detecting ChatGPT content, Turnitin hopes to deter students from using ChatGPT as a shortcut or a cheat tool and encourage them to do their best original work.

What are the Limitations of Turnitin’s AI Detector?

While Turnitin’s AI detector is impressive and reliable in identifying most ChatGPT-generated texts, it is not perfect or foolproof. There are some limitations and challenges that users should be aware of when using Turnitin’s AI detector.

First, Turnitin’s AI detector may not be able to detect all ChatGPT-generated texts. As mentioned earlier, Turnitin’s AI detector has an accuracy rate of 97%, which means that up to 3% of ChatGPT-generated texts may go undetected. This could happen if ChatGPT produces texts that are very similar or identical to human-written texts in Turnitin’s database. For example, suppose ChatGPT copies or paraphrases a sentence from an existing source without changing it significantly. In that case, Turnitin may not be able to distinguish it from the source.

Second, Turnitin’s AI detector may not be able to detect other types of AI-generated texts. ChatGPT is not the only AI chatbot or language model that can produce natural language texts. There are other models, such as GPT-4, DALL·E 3, Codex, and others, that have different capabilities and features. Turnitin’s AI detector may not be able to recognize these models or their outputs as easily as ChatGPT. Moreover, there may be new models or versions that are developed in the future that could pose new challenges for Turnitin’s AI detector.

Third, Turnitin’s AI detector may not be able to detect human-AI collaboration or manipulation. Some students may try to use ChatGPT as a tool or a helper rather than a substitute for their work. For example, they may use ChatGPT to generate some ideas, sentences, or paragraphs and then edit, revise, or combine them with their writing. Alternatively, they may use ChatGPT to generate a text and then modify, rewrite, or paraphrase it to make it more original or coherent. These methods may make it harder for Turnitin’s AI detector to identify ChatGPT content, as the texts may not have the same patterns or features as pure ChatGPT texts.

How Can Educators and Students Deal with ChatGPT?

Given the limitations and challenges of Turnitin’s AI detector, educators and students should not rely solely on Turnitin’s AI detector to detect ChatGPT content. They should also use other strategies and tools to ensure the originality and quality of their work. Here are some suggestions for educators and students to deal with ChatGPT:

Educators should design assignments that are specific, relevant, and meaningful for their students. They should avoid generic or vague topics that ChatGPT or other sources could easily answer. They should also provide clear instructions, expectations, and rubrics for their assignments. They should encourage students to use their voice, perspective, and creativity in their work.
Educators should monitor and support their students throughout the writing process. They should check in with their students regularly, provide feedback and guidance, and ask questions that challenge their thinking and understanding. They should also use formative assessments, such as drafts, outlines, or peer reviews, to track their students’ progress and identify any potential issues or problems.
Educators should educate their students about the ethical and academic implications of using ChatGPT or other AI tools. They should explain the importance of academic integrity and honesty, and the consequences of plagiarism and cheating. They should also discuss the limitations and risks of using ChatGPT or other AI tools, such as accuracy, reliability, quality, and safety. They should help their students develop critical thinking and digital literacy skills that enable them to evaluate and use information sources appropriately.
Students should do their work and avoid using ChatGPT or other AI tools as a shortcut or a cheat tool. They should understand that using ChatGPT or other AI tools is not only dishonest but also detrimental to their learning and development. They should also realize that using ChatGPT or other AI tools may not guarantee good grades or outcomes, as they may produce inaccurate, irrelevant, or low-quality texts that Turnitin or other methods could detect.
Students should use ChatGPT or other AI tools responsibly and ethically as a tool or a helper rather than a substitute for their work. They should only use ChatGPT or other AI tools for legitimate purposes, such as generating ideas, finding inspiration, or learning something new. They should also acknowledge and cite any sources that they use from ChatGPT or other AI tools. They should always check, verify, and edit any texts that they use from ChatGPT or other AI tools.

Frequently Asked Questions – FAQs

Is Turnitin’s AI detector perfect at detecting ChatGPT content?

No, while it’s highly accurate, it’s not foolproof, with a 97% accuracy rate.

Can Turnitin detect other AI-generated texts besides ChatGPT?

It may have difficulties identifying texts from other AI models.

How should educators design assignments to deter ChatGPT use?

Assign specific and meaningful tasks that encourage creativity.

What responsibilities do educators have when using ChatGPT in education?

They should educate students about ethics, provide guidance, and monitor the process.

What should students keep in mind when using ChatGPT?

Use it responsibly, avoiding shortcuts, and always acknowledge sources.

Is Turnitin’s AI detector the ultimate solution to academic integrity?

No, it’s a tool to complement other strategies for maintaining integrity.

Conclusion

ChatGPT is a remarkable AI chatbot that can generate natural language texts on almost any topic. However, it also poses some ethical and academic challenges for educators and students who value originality and integrity in their work. Turnitin’s AI detector is a useful tool that can help detect most ChatGPT-generated texts, but it is not perfect or foolproof. Educators and students should use Turnitin’s AI detector along with other strategies and tools to ensure the originality and quality of their work.

In case you missed it: How to Use Canva Plugin with GPT-4 in ChatGPT