To ChatGPT, or not to ChatGPT: That is the question!
Alessandro Pegoraro, Kavita Kumari, Hossein Fereidooni, Ahmad-Reza Sadeghi
TL;DR
The paper surveys contemporary approaches to detecting AI-generated text, focusing on ChatGPT and categorizing detectors into simple classifiers, zero-shot, fine-tuning-based, and other methods, alongside online tools. It evaluates these detectors using a large benchmark dataset comprising ChatGPT and human responses across multiple domains and social media content. Results show that none of the existing methods reliably distinguish ChatGPT-generated text, with the best detectors achieving less than 50% true-positive rates, highlighting significant vulnerabilities. The study emphasizes the need for more robust, multi-faceted detection strategies and standardized evaluation benchmarks to improve reliability in real-world settings.
Abstract
ChatGPT has become a global sensation. As ChatGPT and other Large Language Models (LLMs) emerge, concerns of misusing them in various ways increase, such as disseminating fake news, plagiarism, manipulating public opinion, cheating, and fraud. Hence, distinguishing AI-generated from human-generated becomes increasingly essential. Researchers have proposed various detection methodologies, ranging from basic binary classifiers to more complex deep-learning models. Some detection techniques rely on statistical characteristics or syntactic patterns, while others incorporate semantic or contextual information to improve accuracy. The primary objective of this study is to provide a comprehensive and contemporary assessment of the most recent techniques in ChatGPT detection. Additionally, we evaluated other AI-generated text detection tools that do not specifically claim to detect ChatGPT-generated content to assess their performance in detecting ChatGPT-generated content. For our evaluation, we have curated a benchmark dataset consisting of prompts from ChatGPT and humans, including diverse questions from medical, open Q&A, and finance domains and user-generated responses from popular social networking platforms. The dataset serves as a reference to assess the performance of various techniques in detecting ChatGPT-generated content. Our evaluation results demonstrate that none of the existing methods can effectively detect ChatGPT-generated content.