Open AccessPosted Content
Automatic Detection of Generated Text is Easiest when Humans are Fooled.
TL;DR: The authors performed a benchmarking and analysis of three sampling-based decoding strategies (top-k, nucleus sampling, and untruncated random sampling) and found that they are primarily optimized for fooling humans.
read more
Abstract: Recent advancements in neural language modelling make it possible to rapidly generate vast amounts of human-sounding text. The capabilities of humans and automatic discriminators to detect machine-generated text have been a large source of research interest, but humans and machines rely on different cues to make their decisions. Here, we perform careful benchmarking and analysis of three popular sampling-based decoding strategies---top-$k$, nucleus sampling, and untruncated random sampling---and show that improvements in decoding methods have primarily optimized for fooling humans. This comes at the expense of introducing statistical abnormalities that make detection easy for automatic systems. We also show that though both human and automatic detector performance improve with longer excerpt length, even multi-sentence excerpts can fool expert human raters over 30% of the time. Our findings reveal the importance of using both human and automatic detectors to assess the humanness of text generation systems.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Citations
Unbiased Watermark for Large Language Models
Zhengmian Hu,Lichang Chen,Xidong Wu,Yihan Wu,Hongyang Zhang,Heng Huang +5 more
TL;DR: This research demonstrates that it is possible to integrate watermarks without affecting the output probability distribution with appropriate implementation, and refers to this type of watermark as an unbiased watermark, suggesting that unbiased watermarks can serve as an effective means of tracking and attributing model outputs without sacrificing output quality.
How Sampling Affects the Detectability of Machine-written texts: A Comprehensive Study
Matthieu Dubois,François yvon,Pablo Piantanida +2 more
- 01 Jan 2025
TL;DR: This study examines how sampling-based decoding affects the detectability of machine-written texts, revealing that minor adjustments to decoding parameters can severely impair detector accuracy, exposing blind spots in current detection methods.
Zero-Shot Detection of LLM-Generated Text using Token Cohesiveness
S.L. Ma,Quan Wang +1 more
- 25 Sep 2024
TL;DR: This paper proposes TOCSIN, a zero-shot detector that leverages token cohesiveness to identify LLM-generated text, outperforming state-of-the-art detectors on various datasets and source models, with a simple and efficient calculation method.
DNA-GPT: Divergent N-Gram Analysis for Training-Free Detection of GPT-Generated Text
TL;DR: Yang et al. as discussed by the authors proposed a novel training-free detection strategy called Divergent N-Gram Analysis (DNA-GPT), which truncates a text and then uses only the preceding portion as input to the LLMs to regenerate the new remaining parts.
GPT-Sentinel: Distinguishing Human and ChatGPT Generated Content
TL;DR: In this paper , the authors presented a novel approach for detecting ChatGPT-generated vs. human-written text using language models, which achieved remarkable results with an accuracy of over 97% on the test dataset, as evaluated through various metrics.
References
•Proceedings Article
Attention is All you Need
Ashish Vaswani,Noam Shazeer,Niki Parmar,Jakob Uszkoreit,Llion Jones,Aidan N. Gomez,Lukasz Kaiser,Illia Polosukhin +7 more
- 12 Jun 2017
TL;DR: This paper proposed a simple network architecture based solely on an attention mechanism, dispensing with recurrence and convolutions entirely and achieved state-of-the-art performance on English-to-French translation.
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Jacob Devlin,Ming-Wei Chang,Kenton Lee,Kristina Toutanova +3 more
- 11 Oct 2018
TL;DR: BERT as mentioned in this paper pre-trains deep bidirectional representations from unlabeled text by jointly conditioning on both left and right context in all layers, which can be fine-tuned with just one additional output layer to create state-of-the-art models for a wide range of tasks.
24.6K
Neural Machine Translation of Rare Words with Subword Units
Rico Sennrich,Barry Haddow,Alexandra Birch +2 more
- 12 Aug 2016
TL;DR: This paper introduces a simpler and more effective approach, making the NMT model capable of open-vocabulary translation by encoding rare and unknown words as sequences of subword units, and empirically shows that subword models improve over a back-off dictionary baseline for the WMT 15 translation tasks English-German and English-Russian by 1.3 BLEU.
•Posted Content
Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation
Yonghui Wu,Mike Schuster,Zhifeng Chen,Quoc V. Le,Mohammad Norouzi,Wolfgang Macherey,Maxim Krikun,Yuan Cao,Qin Gao,Klaus Macherey,Jeff Klingner,Apurva Shah,Melvin Johnson,Xiaobing Liu,Łukasz Kaiser,Stephan Gouws,Yoshikiyo Kato,Taku Kudo,Hideto Kazawa,Keith Stevens,George Kurian,Nishant Patil,Wei Wang,Cliff Young,Jason A. Smith,Jason Riesa,Alex Rudnick,Oriol Vinyals,Greg S. Corrado,Macduff Hughes,Jeffrey Dean +30 more
TL;DR: GNMT, Google's Neural Machine Translation system, is presented, which attempts to address many of the weaknesses of conventional phrase-based translation systems and provides a good balance between the flexibility of "character"-delimited models and the efficiency of "word"-delicited models.
7.9K
The spread of true and false news online
TL;DR: A large-scale analysis of tweets reveals that false rumors spread further and faster than the truth, and false news was more novel than true news, which suggests that people were more likely to share novel information.
7.2K