PRECISION IN SOURCE ATTRIBUTION: A PRECISE CLASSIFICATION STUDY OF HUMAN AND CHATGPT-GENERATED RESPONSES

Abstract

In the age of digital communication, identifying the origin of text-based responses is critical for a number of applications, such as trust evaluation and content regulation. This study explores the creation and assessment of a strong framework for source classification. The basis is a carefully selected dataset that includes a variety of conversations from online forums, social media, and messaging apps. Each response is carefully marked as "Human" or "ChatGPT." To maintain uniformity and standardization, data preprocessing techniques such as tokenization and text cleaning are utilized. Feature extraction looks at lexical, syntactic, and semantic features to find differences between ChatGPT responses and human ones. Our study is centered on a comprehensive analysis of several classification methods, such as Random Forest, Support Vector Machines, Naive Bayes, Logistic Regression, and Neural Networks (LSTM). The study shows a cutting-edge model that is more than 95% accurate, showing that it is possible to clearly tell the difference between information made by ChatGPT and content written by humans. These findings have important ramifications for strengthening fact-checking procedures, regulating material, and guaranteeing reliability in digital communication.

Keywords: AI Source Classification, Machine Learning, Logistic Regression, Naive Bayes, Support Vector Machines, Neural Networks, Human-AI Interaction, AI Textual Analysis