CopyLeaks (Trial): AI detects when student blends paraphrased and AI-generated text

CopyLeaks (Trial): AI detects when student blends paraphrased and AI-generated text

Wednesday, March 8, 2023

CIECT has recently used CopyLeaks, a leading plagiarism detection tool, to positively identify a blend of paraphrased and AI-generated text. CopyLeaks utilises advanced Artificial Intelligence (AI) and machine learning algorithms to detect and flag text that has been generated by various document formats: MS Word, Google Docs, etc.

Example 5: CopyLeaks can detect a mix between AI-generated and paraphrased text

During our trial period, CIECT has requested ChatGPT (AI text generator) to create a piece, using the following phrase: “3-page document on how remote learning is beneficial”. The CIECT facilitators proceeded to paraphrase some of the original text generated by ChatGPT, before uploading it to CopyLeaks (as a PDF file). CopyLeaks scanned the text and produced a detailed report that identified matches found online. The report also highlighted the percentage of similarity between the text and the sources detected as illustrated below.

CopyLeaks uses advanced algorithms to compare the text with an extensive database of online content, including websites, articles, books, and more.

 

Faculties must engage in discussion: Detect and embrace (balance?) – of ChatGPT

Example: Requested ChatGPT to generate a '3-page document on how remote learning is beneficial'.

ChatGPT provided a detailed script on how remote learning is beneficial which included: Access to Education, Flexibility, Cost Savings, Enhance Digital literacy, and Collaborative Learning. 

As the Internet of Things (IoT) continues to grow, and more content becomes available online, tools like CopyLeaks will become even more essential to maintain the integrity of written content. Furthermore, Faculties should engage in discussions regarding these innovative AI tools. How do you strike a balance? It is important to detect the originality of the work. However, are you going to dismiss all work generated by AI (such as ChatGPT)? Will you provide guidelines (for your specific discipline), in order for the students to use these specific AI tools, while maintaining academic honesty?