Towards Explainable AI-Generated Text Detection Using Ensemble and Combined Model Training

Our research tackles the challenge of distinguishing between human and AI-generated text, a crucial issue in the era of advanced language models. We propose a unique approach using an ensemble and mixed-model strategy, focusing on accuracy and explainability. This method involves a variety of advanced text classification algorithms, applied to both English and Dutch texts across multiple genres. Notably, our work integrates SHapley Additive exPlanations (SHAP) for clearer insights into model decisions, emphasizing the importance of explainable AI (XAI). Our study is significant in ensuring the authenticity and integrity of digital content in an increasingly AI-driven world.

Keywords: AI-generated Text Detection, Ensemble Model, Explainable AI (XAI), Natural Language Processing (NLP), Transformer-Based Models.