Fake news and AI
For almost a month now, I have been working with my team on developing machine learning algorithms for busting fake news on social media. We tried to develop an app for this purpose and presented it at the innovation contest held at our workplace. We made it to the top 20 finalist teams, and the finale is yet to happen.
In the process, I learnt many things about fake news and how to identify them, to an extent (the solution and approach is beyond the scope of this article and yes, confidential. :) ). Many AI models were proposed, and algorithms were written. And the teams made a lot of generalisations on how to identify the features of fake news and tackle such content.
Fake news were prevalent in old times as well. The first one, as far as I know, was an April fool’s prank on Usenet in 1984. It claimed that the Soviet Union was joining the network. In recent times, it has been a huge problem as the users spread large amounts of fake news across social media networks like Facebook.
But, do you know why we fall victims to fake news? It’s because of the pressure, anxiety, stress and the uncertainty in the world. They force us to believe all the content on social media. Sensational news gets more attention, and we cannot blame ourselves for falling prey.
According to statista.com, 58% of news articles on Facebook, and almost 46% of those on Twitter are suspected to be fake. It seems 42% of traffic for fake news originated through the social media sites.
User engagement creates traffic for websites and bots used there tend to have a potential to spread the humongous amount of fake news articles. It’s because bots use algorithms to decode what type of information their users are glued to. People mass produce irrelevant but eye catchy content and circulate it through these bots. There is no way of checking the authenticity or credibility of the sources.
A survey by forbes.com suggests that almost 54.2% of user engagement was possible on social media through fake news articles on Facebook. But the circulation of fake news stories would definitely not happen if the user got to know that the news article was fake.
How can we not fall victim to fake news? Well, there are several apps and extensions for chrome (such as BS detector) available to detect if a news is fake. Then, there are fact checking websites (such as www.snopes.com and www.politifact.com) which give real-time notifications when you try to share fake news. Cyber security firms and software companies are taking up the responsibility to develop algorithms to bust fake news on social media. The accuracy level, however, isn’t 100 percent yet.
The bots use algorithms to spread fake news across social media platforms. And recently, there have been memes too that spread fake information through creative humour. Ironically, we need to code similar ML algorithms to get rid of these AI bots. It’s like, AI versus AI!
A recent Gartner study predicted that “the majority of people in advanced economies will see more false than true information by 2022”.
We just have to wait and watch!
