GLTR: Statistical Detection and Visualization of Generated Text

Sebastian Gehrmann, Hendrik Strobelt, Alexander M. Rush
6/10/2019
cs.CLcs.AIcs.HC

Official code: HendrikStrobelt/detecting-fake-text (501 stars, Apache-2.0, last commit 2024-01-18).

Abstract

The rapid improvement of language models has raised the specter of abuse of text generation systems. This progress motivates the development of simple methods for detecting generated text that can be used by and explained to non-experts. We develop GLTR, a tool to support humans in detecting whether a text was generated by a model. GLTR applies a suite of baseline statistical methods that can detect generation artifacts across common sampling schemes. In a human-subjects study, we show that the annotation scheme provided by GLTR improves the human detection-rate of fake text from 54% to 72% without any prior training. GLTR is open-source and publicly deployed, and has already been widely used to detect generated outputs

View on arXivView PDF

Code Implementations(1)

HendrikStrobelt/detecting-fake-textOfficial100%
Apache-2.0

Giant Language Model Test Room

501118CSS, HTMLApr 11, 20192 years agoApache-2.0
ainlpvisualization

Similar Papers

MAGE: Machine-generated Text Detection in the Wild

89% match2 repos2322023

Towards Possibilities & Impossibilities of AI-generated Text Detection: A Survey

88% match2023

Smaller Language Models are Better Black-box Machine-Generated Text Detectors

88% match2023

Ghostbuster: Detecting Text Ghostwritten by Large Language Models

88% match2 repos1872023

Fighting Fire with Fire: Can ChatGPT Detect AI-generated Text?

88% match1 repo72023

Who Said That? Benchmarking Social Media AI Detection

88% match2023

Cite this paper

@article{gehrmann2019gltr,
  title  = {GLTR: Statistical Detection and Visualization of Generated Text},
  author = {Sebastian Gehrmann and Hendrik Strobelt and Alexander M. Rush},
  year   = {2019},
  eprint = {1906.04043},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1906.04043}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1906.04043.svg)](https://codeofpaper.com/paper/1906.04043)

Discussion