If you have been copying and pasting directly from ChatGPT, you may have questions about Turnitin Detect. Many institutions are concerned about the risks of AI, especially when it comes to critical thinking. Students who use generative AI tools like ChatGPT may need to catch up in critical thinking. To ensure that you have completed assignments and other work on your own, Turnitin is a valuable tool for identifying plagiarism.
Turnitin is a plagiarism checker that uses OpenAI’s GPT model, which has been trained on vast amounts of data, primarily sourced from text-based data from sources like Wikipedia, Reddit, Twitter, books, and more. It works similarly to other plagiarism checkers like Edubird, which compares word-by-word content with web content. If it finds a match, it analyses the pattern and shows how much has been copied. However, this tool is not a plagiarism detector but provides reports on similarity.
What is Turnitin?
It has been in development for more than two years, and it is now more critical than ever for educators to be able to determine if the content is generated by AI or not. It uses the same technology that ChatGPT uses, which means it is trained to detect models including GPT-3, GPT-3.5, and even ChatGPT-4 (ChatGPT) partially, but it will be available with future updates.
After the launch of AGL models from OpenAI and Google, it’s hard to restrict students from copying content from ChatGPT or Bard’s generated responses to reduce workload. However, seeking assistance with your assignment is completely different from borrowing content. To help institutions maintain integrity, Turnitin Feedback Studio (TFS) has launched AI Writing Detection tools such as:
- Turnitin Originality
- Turnitin Similarity
- Originality Check
- Originality Check+
More than 2.1 million educators and 10,000+ institutions have used this tool, reaching more than 62 million students. Specifically for GPT, the company has introduced ChatGPT Screening, which helps determine if the content is generated by ChatGPT or not. It is too difficult for humans to identify, but machine learning is good at recognising patterns and specialises in doing one task well. Turnitin is a pretty capable tool.
Does Turnitin Detect Chat GPT?
Yes, Turnitin detects Chat GPT written content, which is accurate with AI-generated content, with an accuracy of ~98%. Academic institutions have also used it to ensure academic integrity and promote original work. It is capable of analysing the content in-depth and finding patterns across other works available on the internet, which means its underlying technology can determine if the content is generated by AI or not.
- The results also share a similar index and the original work you copied from the internet.
- It is also relatively easy to use and doesn’t require a lot of action from the user’s end to check assignments.
How Turnitin Works?
The company has not shared how Turnitin works, but it should be similar to the AI systems used to generate content. If AI can generate content, it should also be capable of identifying it. OpenAI, the developer of ChatGPT, has also released their version of Turnitin called OpenAI Classifier that can identify AI-written text. We want to note that it can only check content in English and may expand in the future, but it is unlikely.
Universities or institutes ask students to submit their work on Turnitin via their LMS environments or the Turnitin website. After submitting the work, AI starts analyzing the work word-by-word with the internet, a database of academic and student research papers. Each word and sentence also analyzes the pattern that helps to find if the content has been copied directly from ChatGPT or a website. Although ChatGPT doesn’t generate its content, it uses a dataset on which it has been trained: online sources.
After processing your work, it shares a report that shows the exact match and the percentage of original work. To provide more information, it includes reference articles, blogs, websites, and more which the content matches.
It also has ChatGPT screening that checks the content’s tone, language, and style to confirm if ChatGPT generates the content or if it is copied from websites. For this, it uses the same OpenAI underlying AI-Model GPT-3.5, but since GPT-4 is now generally available, it will be updated to also find content generated by GPT-4, and the company will improve concerning the company released the new AI-Model.
Since it focuses more on academic institutes and similar ones, it excludes the quote or any area within the work having a quote. You can outsmart Turnitin using AI-Detecting tools like Turnitin or simply by submitting original content. The company also ensures that it follows academic guidelines to resolve the challenges posed by advanced AI-Writing tools.
It’s a challenging task to identify AI-generated content, as AGL models can generate human-like content in a very conversational way. Geoffrey A. Fowler has a great report via The Washington Post on how well Turnitin can detect ChatGPT responses at the same time it flagged students. It is also expected that OpenAI will start any Digital watermark that will help educators spot the copied work or used by Text-generation, such as ChatGPT.