This tool is designed to accurately determine whether a given text sample was generated by the GPT-2 language model. It achieves this by conducting a comprehensive analysis of the text's structure, vocabulary, and syntax, which is then compared to established samples known to have been generated by GPT-2. Educators can leverage the power of this tool on Canvas to ascertain whether a student's text submission is the result of the GPT-2 language model. By doing so, it supports efforts in identifying potential instances of plagiarism and detecting the dissemination of fake news generated by language models.
This tool is designed to detect whether a given text sample was generated by the GPT-2 language model. By analyzing the structure, vocabulary, and syntax of the text, it can compare it to known samples generated by GPT-2. Teachers can use this tool on Canvas to determine if a student's text submission was generated by GPT-2. This tool is useful for identifying potential cases of plagiarism and detecting fake news created by language models.
GPTZeroX by Edward Tian (Princeton University)
GPTZeroX, developed by Edward Tian from Princeton University, is a tool that can determine whether a language model, like GPT-2, generated a given text sample. The tool analyzes the structure, vocabulary, and syntax of the text, and provides a perplexity score and burstiness score. Although it does not directly integrate with Canvas, this tool is beneficial for teachers to detect potential plagiarism cases or to assess the output of AI-generated text, such as ChatGPT, when completing assignments.
While it may be tempting for students to use ChatGPT during a Canvas quiz or exam, the likelihood of detection is high. Canvas provides instructors with various tools to monitor student activity and identify instances of cheating, including using external sources like ChatGPT. To maintain academic integrity, students should always rely on their own knowledge.