ChatGPT Versus Human: Whose Text Comes with Higher Entropy? Assessing Shannon’s Equitability in Distinguishing Between ChatGPT and Human-Generated Texts: An Analytical Approach

This chapter investigates the applicability of Shannon’s equitability as a statistical measure to discern differences between texts generated by ChatGPT and human authors. By analyzing token frequency and Shannon’s equitability values, the research identifies significant disparities in the distribution of token usage between the two sources. The findings indicate that Shannon’s equitability can potentially be a differentiating factor between texts produced by humans and those generated by ChatGPT. In addition, we uncover substantial distinctions when studying the most frequent tokens. The chapter also presents a web application for quantitative text analysis. This application includes concepts from the study methodology.