AI Authenticity Project

In recent years, the rise of AI models capable of generating human-like text has increasingly blurred the distinction between human-authored and AI-generated writing. While existing studies have compared AI-generated texts with native writing, the similarities and differences between AI-generated content and second language (L2) writing remain largely unexplored. Early research indicates that AI-generated texts may share certain features with L2 writing, potentially leading to misinterpretations where human-authored work is mistaken for AI-generated content. AI-generated texts are typically perceived as less credible, authentic, and trustworthy: this negative perception poses a risk for L2 writers, whose work may unfairly viewed as AI-generated.

This project adopts a mixed-methods approach to investigate the human perceptions of authenticity in L2 writing. Specifically, it will examine how teachers perceive the authenticity of L2 texts and compare these perceptions with the actual linguistic patterns present in both AI-generated and L2 human-authored texts. By doing so, the research seeks to uncover whether these perceptions align with the linguistic realities of the texts and to identify any biases that may arise.