Classification of Human- and AI-Generated Texts: Investigating Features for ChatGPT
This work addresses the challenge for educators in identifying AI-generated or rephrased student work, representing an incremental improvement with specific gains in detection accuracy.
The paper tackled the problem of detecting AI-generated and AI-rephrased texts, especially in advanced cases where AI mimics human writing, by exploring traditional and new features on a new corpus of 10 school topics. The best systems achieved F1-scores over 96% for basic and advanced AI-generated text detection and over 78% for AI-rephrased text detection, with one system outperforming GPTZero by 183.8% in relative F1-score.
Recently, generative AIs like ChatGPT have become available to the wide public. These tools can for instance be used by students to generate essays or whole theses. But how does a teacher know whether a text is written by a student or an AI? In our work, we explore traditional and new features to (1) detect text generated by AI from scratch and (2) text rephrased by AI. Since we found that classification is more difficult when the AI has been instructed to create the text in a way that a human would not recognize that it was generated by an AI, we also investigate this more advanced case. For our experiments, we produced a new text corpus covering 10 school topics. Our best systems to classify basic and advanced human-generated/AI-generated texts have F1-scores of over 96%. Our best systems for classifying basic and advanced human-generated/AI-rephrased texts have F1-scores of more than 78%. The systems use a combination of perplexity, semantic, list lookup, error-based, readability, AI feedback, and text vector features. Our results show that the new features substantially help to improve the performance of many classifiers. Our best basic text rephrasing detection system even outperforms GPTZero by 183.8% relative in F1-score.