Ever been duped by an AI-generated academic paper?
Well, a new machine-learning tool is here to save the day! This tool, designed to spot when chemistry papers are written using the chatbot ChatGPT, is making waves in the academic publishing world.
Developed by Heather Desaire, a chemist at the University of Kansas, and her team, this tool outperforms existing AI detectors. It focuses on a specific type of paper to enhance accuracy, a strategy that could revolutionise AI detection efforts.
The tool uses machine learning to examine 20 features of writing style, including sentence length variations and the frequency of certain words and punctuation marks. It then determines whether a human or ChatGPT wrote the text.
In a recent study, the tool was trained on the introductory sections of papers from ten chemistry journals. It was then tested on introductions written by both people and AI from the same journals. The results were impressive: it identified AI-generated sections with 100% accuracy!
However, the tool is highly specialised for scientific journal articles. When presented with real articles from university newspapers, it failed to recognise them as human-written.
While this tool is a significant step forward, it doesn't address the wider issues driving the use of AI in academia. Many researchers are under pressure to quickly produce papers, and AI-detection tools won't solve these social problems. But it's a start, and it's a fascinating one at that!
Made with TRUST_AI - see the Charter: https://www.modelprop.co.uk/trust-ai
Comments