Summary of Deep Learning Detection Method For Large Language Models-generated Scientific Content, by Bushra Alhijawi et al.
Deep Learning Detection Method for Large Language Models-Generated Scientific Content
by Bushra Alhijawi, Rawan Jarrar, Aseel AbuAlRub, Arwa Bader
First submitted to arxiv on: 27 Feb 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This research paper introduces a novel method for detecting ChatGPT-generated scientific texts, called AI-Catcher. The model integrates two deep learning models: multilayer perceptron (MLP) and convolutional neural networks (CNN). The MLP learns feature representations of linguistic and statistical features, while the CNN extracts sequential patterns from textual content. The fused hidden patterns enable AI-Catcher to distinguish between human-written and ChatGPT-generated scientific texts with high accuracy. The authors also collect a new dataset, AIGTxt, containing 3000 records across ten domains, divided into three classes: Human-written, ChatGPT-generated, and Mixed text. Experimental results demonstrate the superiority of AI-Catcher over alternative methods, achieving an average improvement in accuracy of 37.4%. This research has significant implications for the scientific community, as it enables detection of potentially compromised publications generated by LLMs. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps us figure out whether a piece of scientific writing was written by a human or a computer program called ChatGPT. The researchers created a new tool called AI-Catcher that uses two different kinds of machine learning models to analyze the text. They also made a special dataset with lots of examples of texts, some written by humans and others written by ChatGPT. By testing their tool on this dataset, they showed that it can accurately identify which type of text is which. This is important because scientists rely on being able to trust the information they read in academic journals. |
Keywords
* Artificial intelligence * Cnn * Deep learning * Machine learning
