Researchers have found that classical machine learning algorithms can effectively identify text generated by large language models without relying on complex deep learning architectures. This traditional approach provides a highly efficient and lightweight alternative for systems designed to detect AI-generated content with minimal computational overhead.