Copyright © 2018 DataScience.US All Rights Reserved.
Can Big Data Fight Fake News?
The concept of fake news has received a lot of media attention over the past year. The internet is a fantastic tool for acquiring information, however not all the information you can find on the internet can be trusted. In that sense, fake news is really just the problem of misinformation gaining more relevance and being branded with the new title. Could big data and data science techniques tackle the problem of fake news?
The Problem of Fake News
The problem of the fake news is multifaceted. In fact, much of the problem comes from information silos and echo chambers. Now that many people are more free to choose what kind of sources they want to engage with, the news that gets around a community is often unbalanced and barely fact checked.
While echo chambers and information bubbles represent their own problem, researchers are also blaming the proliferation of fake news on how Twitter and Facebook enable the mass-distribution of misinformation. The ubiquity of these platforms is creating new threats, according to experts who study the problem, due to the fact that these forms of the media are enabling the deception of larger audiences by small groups of people.
Bots are often used on Twitter to not just disseminate false information, but to make it look like a large number of people are tweeting or retweeting a certain article, which will then bring it to the attention of others. The sheer amount of new stories being generated each day by increasingly beleaguered teams of reporters also contributes to fake news. Reporters are frequently urged to produce as much content as they can and thus do not have the time to fact check every piece of information.
How could an AI be employed to stop misinformation from showing up in people’s news-feeds?
The Role of AI in Fighting Fake News
Machine learning and artificial intelligence have already been tapped as possible solutions to the fake news conundrum. In 2015 Google published a paper which explored a new method of ranking webpages based on how accurate the facts on them were. Their algorithm utilizes a “trust score” which is then integrated into Google’s general page rank algorithms to determine their search rank. Facebook has also decided it will leverage the power of AI to detect various patterns and word groupings that are often found in fake news stories. Other companies and entities have also tried to grapple with the problem of fake news, including the French news media which is using a system called CrossCheck to identify news articles which are likely to be false.
However, attempting to label an entire article or news source as “fake” or “real” is a problem. An article may get a few details wrong here or there, yet still be largely correct. There’s also the fact that there are many different ways to interpret the information we come into contact with, often making “fake news” a matter of perspective, and seeing many different perspectives on a news item can be healthy as it allow us to make decisions which are better informed.
There could be utility for a browser plug-in that would automatically discern quotes and claims from an article and cross-reference them with other reports on the topic. It could display which facts have been independently verified and reported by different news sources, and let a reader see if those details are contested or widely agreed upon, to present a just-the-facts version of the news report. This approach to fighting fake news would duck the problem of labeling a news item as either “fake” or “real”, but instead just give the reader a tool to help them be better informed and more skeptical.
One of the best chances for AI to assist in fighting fake news is to develop technology that can actually understand the meaning of the text found in an article, capable of accurately differentiating between real and fake news. This would mean designing systems that can accurately determine clear indicators of each.
A Blended Approach to Dealing With Fake News
While the potential for an advanced AI to automatically determine which claims are true or false is an attractive idea, it isn’t possible with today’s current technology. The technology may get there one day, but until then a better approach might be combining AI tools with human intuition. An AI can scan through millions of articles much faster than a human ever could, using sophisticated pattern recognition algorithms to determine if a news story is likely to be full of misinformation, and which facts are common amongst multiple reports. Yet it would still be up to the reader to make a judgment about what they should believe.
Such a combined system could help cut down on the scourge of fake news yet still preserve the autonomy of the reader, and not merely dictate to the reader what news is real or fake. It could also prove more effective than pure AI systems, at least at this point in time, similar to how the most effective chess players in the world are teams of humans and AIs working together.
Whether or not AI works alongside humans to sniff out misinformation, people will still need to be instructed in how to think critically and skeptically about the information they come into contact with. The sooner we start educating people in these necessary skills, the better.