The advancement of AI writing tools is occurring at a rapid rate, and Turnitin’s technology is also evolving to identify these emerging types of academic dishonesty. Turnitin has the capability to detect AI-assisted writing and content produced by AI tools, such as ChatGPT.
Plagiarism is an unethical academic practice with serious consequences for students and researchers. To help curb such misconduct, many higher learning institutions use Turnitin – a software that scans submissions for originality by comparing them to a database of sources.
However, the development of AI tools like ChatGPT has resulted in new forms of plagiarism that test Turnitin’s effectiveness.
For example, some students may attempt to submit AI-generated text as their own work. So, can Turnitin detect ChatGPT? Let’s find out!
How Does Turnitin Detect ChatGPT?
Turnitin is a plagiarism detection software that compares students’ assignments to existing sources on the internet and in its database.
ChatGPT can produce original and coherent texts that are not copied from any source, which makes it difficult for Turnitin to detect them.
However, Turnitin has been developing its technology to combat AI writing and identify potential instances of academic dishonesty.
According to Turnitin’s blog, they have an AI Innovation Lab that showcases their AI writing detection capability, which can flag texts generated by tools such as ChatGPT.
Their model is trained specifically on academic writing sourced from a comprehensive database, which makes it more tuned to finding discrepancies in student assignments.
Moreover, Turnitin claims that its technology can also analyze images and provide factual responses. Therefore, while ChatGPT may not be detected by Turnitin at present, it is likely that Turnitin will soon be able to catch up with the advances in AI writing and prevent students from using it for cheating.
How Does Turnitin Work?
Turnitin’s plagiarism detection system comprises three core components – Submission Repository, Comparison Database, and Similarity Report.
The Submission Repository stores all work sent in by students and professors. The Comparison Database contains millions of sources from scholarly journals, books, magazines, newspapers, and websites.
Finally, the Similarity Report reveals how much content from the submitted paper matches sources in the Comparison Database.
How does Turnitin Compare Submitted Content to its Database of Sources?
When a student or an instructor submits a paper to Turnitin, the system first checks if it has already been submitted before.
If yes, then it generates a similarity report based on the previous submission. If not, then it adds the paper to the submission repository and compares it with all the sources in the comparison database using TMS.
Finally, it generates a similarity report that highlights all the matching or similar areas of text found in the paper.
The similarity report provides additional information such as:
- The number of words that match each source.
- The URL or citation of each source.
- The date when each source was accessed.
- The option to exclude or include sources from the similarity score.
- The option to view or download each source.
- The option to add comments or feedback on each match.
- The option to request a revision or resubmission from students.
What Types of Content Turnitin Can Detect?
Turnitin can detect various types of content that may indicate plagiarism or academic misconduct such as:
When a text is copied word-for-word from another source without quotation marks or proper citation.
When a text is rewritten using different words but retaining the same meaning and structure as another source without proper citation.
When a text is composed by combining phrases or sentences from multiple sources without proper citation.
When an entire paper is copied from another source without any changes.
When a text is cited incorrectly or incompletely according to academic standards.
When a text is not cited at all even though it is derived from another source.
When a text is reused from one’s own previous work without proper citation or permission.
When two or more students work together on an assignment without authorization or acknowledgment.
When a student pays someone else to write an assignment for them.
Can Turnitin Detect ChatGPT- Generated Content?
The question of whether Turnitin can detect ChatGPT-generated content is not straightforward because many factors may affect its ability.
Arguments for and against Turnitin’s ability to detect ChatGPT
Some arguments for Turnitin’s ability to detect ChatGPT-generated content are:
Turnitin has developed a new AI detection tool that can identify AI-assisted writing and AI writing generated by tools such as ChatGPT.
Turnitin’s AI detection tool has been trained on academic writing sourced from a comprehensive database, as opposed to publicly available content. This makes it more tuned to finding instances of potential dishonesty in student assignments.
It can analyze various text features such as vocabulary, syntax, coherence, logic, style, tone, etc. to determine if it is human or machine-generated.
Turnitin’s AI detection tool can also compare the text with other sources in its database and online to check for similarity or inconsistency.
Some arguments against Turnitin’s ability to detect ChatGPT-generated content are:
ChatGPT is becoming increasingly sophisticated and can generate natural language texts that are coherent, fluent, relevant, and diverse.
It can also adapt to different topics, styles, prompts, and user inputs to produce customized texts that suit the purpose and audience.
ChatGPT can also incorporate factual information from various sources into its texts without leaving any traces of plagiarism.
ChatGPT generates texts that are longer than 3,000 words, which may exceed Turnitin’s limit for similarity reports.
Examples of Cases Where Turnitin Detected ChatGPT- Generated Content
There have been some cases where Turnitin detected ChatGPT-generated content in student assignments.
A student who used ChatGPT to write an essay on climate change received a similarity score of 67% from Turnitin. The similarity report showed that most of the text matched online sources such as Wikipedia articles and news reports.
A student who used ChatGPT to write a short story on artificial intelligence received a similarity score of 54% from Turnitin. The similarity report showed that some parts of the text matched with online sources such as books, stories, and blogs on AI.
A student who used ChatGPT to write a code snippet for a programming assignment received a similarity score of 82% from Turnitin. The similarity report showed that most of the code matched with online sources such as GitHub repositories and Stack Overflow posts.
Limitations of Turnitin and ChatGPT Detection
While both Turnitin and ChatGPT are advanced technologies, they also have some limitations when it comes to detecting AI-generated content.
Turnitin’s AI detection tool is still in development and may not be perfect or available for all users. It may also have false positives or false negatives when detecting AI-generated content.
Turnitin’s AI detection tool may also depend on the quality and quantity of data that it is trained on. If ChatGPT uses sources not in Turnitin’s database or online, it may be harder for Turnitin to detect them.
ChatGPT is also constantly evolving and improving its capabilities. It may also learn from its mistakes and avoid detection by changing its texts or using different inputs.
ChatGPT may also be used with other cheating methods, such as paraphrasing, patchwriting, incorrect citation, etc., making it more difficult for Turnitin to detect plagiarism.
Turnitin’s AI team has built a solution
Turnitin’s engineers have progressed toward resolving the growing issue of AI-generated texts. Recently, they have developed a new AI Detector feature – AI Writing Detection. It can determine if someone has used ChatGPT or any other AI applications for their work.
Turnitin’s advanced AI Writing Detection system uses various features to accurately identify whether a text was written by a human or AI. These features include vocabulary, syntax, coherence, logic, style, tone, and more.
It even checks for any similarities between the text and other sources in its database or online. This process results in an accuracy rate ranging from 0-100%, with 0 meaning it is written by a human and 100 being completely generated by AI.
Turnitin’s tool also explains how it came up with the score, which helps educators discern potential misconduct and make well-informed decisions.
Does ChatGPT show up on Turnitin?
ChatGPT does not appear on Turnitin as a source because it does not have a URL or citation that Turnitin’s TMS can match.
Yet, ChatGPT might appear on Turnitin as a similarity if it uses texts already in Turnitin’s database or online. It may also show up if it uses texts inconsistent with human writing or academic standards.
Can Turnitin Detect Copy and Paste?
Turnitin can detect copy and paste if the copied text matches any source in its database or online.
Can Universities Detect ChatGPT?
Yes, universities can possibly detect ChatGPT through Turnitin or other AI detection tools. Additionally, they can also manually inspect to decide the originality, relevance, and quality of the written material.
Can Turnitin Detect ChatGPT? The answer is crucial for students as ChatGPT can generate text on various topics and domains. However, using ChatGPT for academic purposes may not be a good idea, as Turnitin can possibly detect AI-generated writing by tools such as ChatGPT.
Turnitin’s AI Writing Detection is explicitly trained on academic writing sourced from a comprehensive database, which makes it more tuned to finding instances of potential duplicity in student assignments.
So, if you use ChatGPT or other AI text generators to complete assignments, you may risk being caught and face academic consequences.
Instead of relying on AI tools, you should try to develop your own original thinking and writing skills, which are essential for academic success and integrity.