GLTR

What is GLTR?

GLTR is a powerful tool developed by the MIT-IBM Watson AI lab and HarvardNLP, designed to detect and analyze automatically generated text. It operates by examining the likelihood of each word in a text, identifying those that appear "too likely" to be human-written, a common trait of computer-generated content. This makes GLTR an invaluable asset for forensic analysis and fake text detection, such as counterfeit reviews or news articles. It has been effectively used in diverse scenarios, from scrutinizing GRE tests and scientific papers to analyzing automated text systems like those used by The Washington Post. Recognized for its innovative approach, GLTR has been nominated for the best demo at the ACL 2019 demo track, and its source code is openly accessible on Github.