A team of US researchers has developed a program that eliminates false news, incorrect accounts and false comments through a GLTR artificial intelligence tool designed by Harvard and Massachusetts Institute of Technology (MIT) technology experts to try to counter the false news.
The device relies on predictive language models in its algorithms, which allow computers and robots to write text to distinguish text whether human or whole.
GLTR has helped improve the detection rate of fake text from 54% to 72%. It can also help identify fake profiles on social media accounts such as Facebook and Twitter, which can often be used to publish incorrect data.
The program works on the assumption that the text created by the computer uses the most common words in the algorithm, but in contrast human writing chooses unexpected words, which are logical and fully relevant even if they are unexpected in the rhythm of a sentence.
These results are color-coded for ease of use. The words that are statistically predictive, which are generated by the computer are highlighted in green, and the yellow, purple, and red word combinations are defined by the human machine.