DAMASHA: AI Text Detection with Human-Interpretable Attribution

This research explores a novel approach to detect AI-generated text, focusing on the challenging scenario of mixed adversarial texts. The human-interpretable attribution aspect is particularly promising for transparency and understanding the detection process.

• •Focuses on AI text detection in mixed adversarial text.
• •Employs segmentation techniques for improved accuracy.
• •Emphasizes human-interpretable attribution for understanding.