Study Evaluates Effectiveness of Human and AI Detectors in Identifying AI-Generated Academic Writing

A recent study discussed in University World News by Wagdy Sawahel examines the effectiveness of six different AI content detectors alongside human reviewers in identifying AI-generated and AI-rephrased academic articles. The authors fabricated 50 rehabilitation-related articles using ChatGPT and rephrased them with Wordtune. AI detectors like Originality.ai perfectly identified AI-generated and rephrased texts, while others like Turnitin struggled, especially with rephrased content. Human reviewers, particularly professors, were effective at detecting AI-rephrased content but also misclassified some human-written articles. The study, led by researchers from Hong Kong Polytechnic University, suggests a combination of tools and human evaluation is crucial for maintaining academic integrity. The findings emphasize the need for ongoing improvements in AI detection tools and the training of human reviewers to adapt to advancements in AI text generation.

This is a summary of an article first published byUniversity World News