Can Turnitin Detect ChatGPT?

Tl;Dr;No, not really. The only way it can really detect ChatGPT is if students submit an assignment verbatim from ChatGPT without any editing.

With so many students using ChatGPT on assignments, instructors are wondering whether commonly used plagiarism tools like Turnitin.com can detect ChatGPT. While Turnitin says that it has a ‘AI detector’, can this really be effective in detecting and preventing the use of ChatGPT on tests? While this AI detector is effective in detecting content that has been pulled straight from ChatGPT with no editing, just a small amount of editing from a student makes the use of ChatGPT completely undetectable. Ultimately, this will catch students that have not edited a single word from ChatGPT, but it is likely that any instructor will be able to tell this visually anyway, as ChatGPT is very formulaic in its replies. However, most students know this, and are editing essays written mostly by ChatGPT, so they are completely undetectable.

Before deciding what kind of tool to use, it’s important to review the origins, evolution, and current status of AI plagiarism. ChatGPT was released in November 20, 2020, and since that time has grown in adoption significantly. Despite the prior existence of several other AI plagiarism tools like “Jasper”, or ContentBot, the release and widespread adoption of ChatGPT has only now truly brought these tools to mainstream awareness, with usage spiking higher than ever before as students seek ever more advanced AI essay writing tools. Given the wide variety and refinement of AI plagiarism tools, ChatGPT is effectively the tip of the iceberg when it comes to what AI essay writing can do. Today’s AI content generators are so powerful that it completely upends and disrupts the Humanities. Some academics have brought up that if AI can pass English and Humanities courses, then perhaps they should not be taught at all – but the same logic would therefore extend to every subject currently taught in not only secondary but also tertiary education. Today’s AI systems are so advanced that they are not only capable, but demonstrate expert knowledge on many secondary school subjects, and show above average capabilities on tertiary school subjects and professional examinations like the LSAT.

This is only the current generation of AI systems that have been released to the public. In just a few month’s time, OpenAI’s GPT went from earning a barely passing grade on some AP exams to earning perfect 5s across the board, in addition to passing a large number of other difficult exams. The next generation is certain to be more powerful, and with concerns growing in the technology industry and in government about the ramifications of what an even more powerful AI system could be capable of achieving, the education industry must also take a close look about what’s being done to preserve the value of education when AI can help any student pass any exam or assignment with zero prior knowledge.

Philosophically, there emerges a debate on the topic of the value of education. While there are those academics who believe that the existence of a very capable AI tool should cause us to change our entire education system, it is worth considering that comparably, the existence of an Einstein-level student does not require the abandonment of an entire educational framework. Furthermore, if Humans abdicate our raw knowledge and critical thinking to artificial intelligence tools, we will create an entire generation of uneducated individuals to coast through their education without obtaining its benefit. For AI tools to benefit everyone, there must exist a skilled and intelligent operator, and without the proper education, such skill and intelligence will be in short supply. In the near-term, instructors are grappling with how to score classes where every student receives a passing grade on an assignment. To be sure, ChatGPT isn’t going to help an ‘F’ student produce an ‘A+’ work every time, it may help them turn that ‘F’ into at least a ‘B’ every time, and with a little preparation, earn that 'A'. With many instructors grading on a curve, this poses competitive challenges for students who are not using ChatGPT. Furthermore, it creates an environment where honest students are faced with a dilemma: either use AI essay writing tools or be at a disadvantage to their peers who are never going to be caught for using these AI tools. As discussed in an earlier blog post, while it is possible for professors to create prompts that are more difficult for AI systems to write, ultimately this is a futile task with the increasing capabilities of AI.

Authoriginal is a solution to this problem, as a simple LTI app and Chrome extension that completely prevents the use of ChatGPT on assignments. Authoriginal is being released with a free and pro tier, in an effort to help instructors everywhere prevent the use of AI systems on assignments. Authoriginal does everything that Turnitin does, and also prevents all other forms of plagiarism, including purchased essays, AI systems, and more. Reach out to us at sales@authoriginal.com to see a demo. We look forward to hearing from you.