Summary of Detecting Machine-generated Texts: Not Just “ai Vs Humans” and Explainability Is Complicated, by Jiazhou Ji et al.
Detecting Machine-Generated Texts: Not Just “AI vs Humans” and Explainability is Complicated
by Jiazhou Ji, Ruizhe Li, Shujun Li, Jie Guo, Weidong Qiu, Zheng Huang, Chiyu Chen, Xiaoyu Jiang, Xinru Lu
First submitted to arxiv on: 26 Jun 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
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 The paper introduces a novel approach to detecting Large Language Model (LLM)-generated text by shifting from binary classification to ternary classification, which includes an “undecided” category for texts that could be attributed to either human or AI sources. The authors create four new datasets and perform tests using state-of-the-art (SOTA) detection methods to identify the most effective approaches. They also analyze the explainability of three top-performing detectors and propose guidelines for developing future systems with improved explanatory power. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper is about trying to figure out who wrote a text – was it a person or a computer? Right now, we just say if it’s human or AI, but that’s not enough. We need to know why the detector chose that answer and how sure they are. To do this, we created new datasets with texts from different machines and humans, tested some popular detectors on these texts, and saw which ones did best. The results show us why having an “undecided” category is important for understanding what’s going on. |
Keywords
» Artificial intelligence » Classification » Large language model
