The rise of ethical concerns about AI content creation is evident and can't be ignored. AI-assisted writing seems to be the common rule in today's universities, online spaces, and other scenarios, which makes detecting inauthentic content more important than ever.

Turnitin, an online detection service already being used by over 34 million students in more than 153 countries, shows itself as a solution for spotting instances of plagiarism and maintaining integrity in academic institutions.

But can Turnitin detect Chat GPT? How accurate is it? Let's see what the data says to find out the correct answer to this question.

How Does Turnitin Work?

First of all, it's important to understand how Turnitin works. The educational software follows different academic guidelines, patterns, and datasets to maintain academic integrity in students' texts.

There are three main points that can help us understand the functionality of this comprehensive tool:

• Text Comparison: Turnitin uses an extensive database containing academic content, including papers, articles, and other written material. When a document is submitted, Turnitin compares the text against its database to identify any matches or similarities.
• Similarity Report: After comparing the submitted document, Turnitin generates a Similarity Report. This report highlights instances of matching text and provides a similarity percentage. This helps educators and students gauge the originality of the work and identify potential areas of concern.
• Citation and Referencing Analysis: Turnitin not only detects direct matches but also analyzes citations and references. It can identify whether sources are properly cited and if the content adheres to academic integrity standards. This feature helps institutions ensure that students give credit to the original authors and follow proper citation practices.

Turnitin Similarity Report vs. AI Writing Detection

Turnitin's Similarity Report checks students' work's originality based on the similarity percentage with other papers, resources, and writings.

However, it is worth noting that the Similarity Report is one of the tool's core features, and it's older than Turnitin's AI detection capabilities.

Turnitin launched its AI detection model in April 2023, which is now integrated into the Similarity Report, but they are two distinct features:

Can Turnitin Detect Chat GPT?

Yes, Turnitin can detect ChatGPT-generated content with 99% accuracy, according to the tool's website.

The tool analyzes individual sentences to determine the originality of student work, showing an AI detection score that ranges from 0% to 100%.

Nevertheless, the AI writing report is not always accurate, and, in practice, varies depending on many different factors.

The Washington Post conducted a study where they checked 16 different text samples of AI-generated content, mixed-source essay, and human-written text to Turnitin.

The tool accurately identified six out of 16 AI-generated texts but failed to identify three more.

Additionally, the detection tool marked as "AI-generated writing" some original parts of academic texts that were entirely generated by a human. This is known as "false positives," and according to Turnitin, it happens less than 1% of the time.

Can Turnitin Detect AI-Written Content from Other Tools?

Yes, Turnitin can detect AI content generated by other language models and tools in the market.

The company claims that its model is capable of detecting text generated by advanced AI-writing tools, and although it doesn't specify the name of the models or AI writing service providers, it is safe to say that Turnitin's AI detection capabilities extend across a broad spectrum of AI writing technologies.

How Accurate is Turnitin in Detecting AI-Written Text? Limitations & Bottlenecks

Despite showing itself as a powerful tool to differentiate between human writing and ChatGPT-written text, Turnitin has some limitations in its AI-writing detection feature:

• The tool has a false positive rate of 1%, meaning that out of 100 human-written texts, it can wrongly flag one as AI-written.
• Turnitin might miss 15% of AI-written text in a document, meaning that if the tool identifies that 50% of a document is written by an AI tool, it could contain as much as 65% AI writing.
• Turnitin focuses on long-form writing, indicating that the tool needs at least 300 words to correctly identify potential AI assistance in the document.
• The tool fails to identify ChatGPT-written text in bullet points, lists, tables, and other non-sentence structures within a text.
• Since large language models continue evolving over time, Turnitin depends on the submission of real-world, updated documents to optimize its detection model as time passes.

Key Takeaways: Can Turnitin Detect Chat GPT Writing?

In summary, Turnitin can indeed detect Chat GPT writing, but it could also flag human-written text as Chat GPT content erroneously.

Some educators say that AI poses a potential threat to our education system, and praise the rise of AI and plagiarism detection tools to preserve originality.

Others say that trying to prevent the use of AI tools in academic writing is the equivalent of trying to ban calculators in educational settings a few decades ago. Either way, there are our three main key takeaways from this topic:

1. Turnitin is not only a robust plagiarism checker, but also a powerful AI detection tool that is being widely used in different educational communities around the world.
2. Although Turnitin can indeed detect ChatGPT content, it might return false positives. Likewise, the tool aims to detect AI generation in long-form content and straight sentences, limiting its usage in some scenarios.
3. The AI writing indicator report isn't shared with students due to worries that they might attempt to use AI tools to bypass Turnitin's AI detection using AI tools. However, educators and instructors can download the complete report and share it with students in PDF format.