DetectLLM: Leveraging Log Rank Information for Zero-Shot Detection of Machine-Generated Text
By Jinyan Su, Terry Yue Zhuo, Di Wang, Preslav Nakov. Conference on Empirical Methods in Natural Language Processing, 2023
299 citations recorded by the source corpus.
With the rapid progress of large language models (LLMs) and the huge amount of text they generated, it becomes more and more impractical to manually distinguish whether a text is machine-generated. Given the growing use of LLMs in social media and education, it prompts us to develop methods to detect machine-generated text, preventing malicious usage such as plagiarism, misinformation, and propaganda. Previous work has studied several zero-shot methods, which require no training data. These methods achieve good performance, but there is still a lot of room for improvement. In this paper, we introduce two novel zero-shot methods for detecting machine-generated text by leveraging the log rank information. One is called DetectLLM-LRR, which is fast and efficient, and the other is called DetectLLM-NPR, which is more accurate, but slower due to the need for perturbations. Our experiments on three datasets and seven language models show that our proposed methods improve over the state of the art by 3.9 and 1.75 AUROC points absolute. Moreover, DetectLLM-NPR needs fewer perturbations than previous work to achieve the same level of performance, which makes it more practical for real-world use. We also investigate the efficiency--performance trade-off based on users preference on these two measures and we provide intuition for using them in practice effectively. We release the data and the code of both methods in https://github.com/mbzuai-nlp/DetectLLM
Subfields: Artificial Intelligence
View Paper: https://www.semanticscholar.org/paper/1e66b2d2a56842f07e5d6bf4ac0dbe5da829d052
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Avi Schwarzschild
- Micah Goldblum
- Valeriia Cherepanova
- Aniruddha Saha
- Linyi Yang
- Zhiyang Teng
- Jonas Geiping
Citations in the corpus
- Fast-DetectGPT: Efficient Zero-Shot Detection of Machine-Generated Text via Conditional Probability Curvature (2023, methodology): “, 2023) and zero-shot classifiers (Gehrmann et al., 2019; Mitchell et al., 2023; Su et al., 2023).”
- Spotting LLMs With Binoculars: Zero-Shot Detection of Machine-Generated Text (2024, methodology): “This two-model mechanism is the basis for our general and accurate detector, and we show that this mechanism is able to detect a number of LLMs, even when they are unrelated to the two models used by Binoculars .”
- MAGECODE: Machine-Generated Code Detection Method Using Large Language Models (2024, unknown): “This paper assesses six metric-based detection methods, including Log-Likelihood [13], Rank [14], Log-Rank [4], Entropy [14], GLTR [14], and LRR [15].”