Fake news has been a great problem in today’s media and AI has come to the rescue, by creating more fake news written by robot bots. As confusing as horrifying as this sounds, researchers of the University of Washington have decided that in order to detect fake news, the bots should be able to create fake news.
The fake news developing a neural network, which has been named GROVER, is all capable to create fake news with the perfect leads and headlines to make the readers believe, just like human writers. It will create a whole made up a story using just a headline.
GROVER creates a wide set of complex models, using unique styles, phrases, and ideas to create fake news. This uses millions of sets of data which is lifted from loads of fake news. This intelligence gives it a brain to identify fake news from any website.
Fake news detecting AI has already been in the market before. But the makers have been convinced that it was too dangerous to be released publicly. But the researchers claim that this tool can be more accurate than any other tools in the market today.
“We observed that best current discriminators can detect neural fake news from real, human-written, news with an accuracy of 73%, assuming access to a moderate level of training data,” wrote the researchers in the paper. “Hence, the best defense against Grover is always Grover itself, with 92% accuracy.”
Most people on social media now keep sharing a piece of news, without even checking if it is true or not, and these researchers claim to have developed this algorithm to put an end to this vast connections. However, GROVER is extremely easy to use and it could be easily put into bad use by the Villians of the media.