The flow of fake and misleading news on the web is nothing new, but the rise of this fact has recently alarmusing Scientists from the University of Michigan and the University of Amsterdam began to deal with the global problem of "Fake News" when they came across an interesting fact. According to the findings, fake news has different characters and linguistic features than real news. These are other types of sentence structure, random punctuation or a special choice of words. The discovered facts were used to create an algorithm that is supposed to help artificial intelligence in detecting fake news and false information.
You could be interested in
Artificial intelligence needs frequent training, thanks to which it can improve in a given function. Training the AI to detect fake news was very simple - throwing thousands of false information and misleading articles into the system, which the AI compared against each other. It is a very simple, albeit complex, programmable method that would certainly find use in practice. After the launch of the publication in which this report was published, almost immediately several news portals announced that they would like to use this system in the fight against Internet propaganda and disinformation.
If you are a frequent user of the social network Twitter, you have certainly not avoided fake news or "Fake News".
According to the creators of this system, the problem is not in the creation of the algorithm, but in its correct programming. Detecting fake news using conventional methods is very difficult and time-consuming. In most cases of detection of "Fake News", the administration can be eliminated, but at this stage it is already so widespread that the mere removal of a post, article or, in most cases, a tweet has almost no effect. However, artificial intelligence can quickly compare a huge number of Internet messages with attributes that, based on grammar, punctuation and sentence structure, can detect the spread of misinformation. When creating a kind of "intelligence truth detector", however, it is very difficult to use the algorithm in practice. Many authors write their articles in a misleading way, in which they also use a significant amount of irony or witty additions, which the system simply cannot detect.
However, the researchers working on this project are determined to continue working with the issue of fake news. The automatic system that will use the aforementioned algorithm should focus on the sources of global administrations and subsequently their offshoots. In this sphere, artificial intelligence should acquire information about the speed of gathering and subsequent publication of messages that, according to a percentage estimate, could appear as false messages. Likewise, the fact of exposing articles that are not found on the web in very small numbers and are simply too exaggerated or new to be true will also serve this purpose.
