Distinguish Human Generated Text From ChatGPT Generated Text Using Machine Learning

Abstract

The big improvement in language models like ChatGPT has made it really hard to tell if a piece of text was written by a human or a computer. This is a problem for things like making sure students are not cheating controlling what people can post online and being honest when we talk to each other on the internet. This paper is about a system that uses machine learning to figure out if a piece of text was written by a human or a computer. The system looks at how the words and sentencesre put together what kind of words are used how well the text makes sense and if the meaning is consistent. It checks the language patterns and statistics of the text to see if it was written by a human or a language model, like ChatGPT. A study is done to see how well eleven machine learning and deep learning algorithms work. These algorithms are used to classify things. The study looks at how they do, in different areas. The study uses a dataset from Kaggle that has 10,000 pieces of text. Some of this text, 5,204 pieces was written by people. Comes from news articles and social media. The rest of the text was made by machines. The text is cleaned up. Changed into something that the machine learning models can use to learn from. This helps the models get better at classifying things. The machine learning models are trained with this information. Additionally, the system incorporates visualization techniques to present real-time predictions, model accuracy, and performance metrics through graphical representations. The results demonstrate the effectiveness of the proposed approach in supporting reliable AI text detection and promoting responsible and transparent use of generative AI technologies.