L. D. M. S. Sai Teja, N. Siva Gopala Krishna, Ufaq Khan, Muhammad Haris Khan, Partha Pakray, Atul Mishra
The paper presents DAMASHA, a framework for detecting transitions between human and AI-generated text using stylometric cues and perplexity signals, and introduces a new benchmark for testing its robustness against adversarial text segments.
As AI-generated text becomes more common, distinguishing between human and AI writing is crucial for maintaining trust and authenticity. This study introduces a new method, DAMASHA, to identify where authorship changes in a text that contains both human and AI contributions. The method uses linguistic style and text complexity to pinpoint these transitions. Additionally, a new dataset has been created to test how well this method works, even when the text is deliberately altered to confuse detection systems. The research also explores how easy it is for humans to understand the detection process, aiming to improve trust in human-AI collaborative writing.
