#ai-controlledbots

False information dates before the invention of the Internet, of course, but the advent of this disruptive technology has changed radically the way false information is presented and circulated in public space. The low cost to create and maintain fraudulent websites and the facilities to spread information on the web have taken falsehood to the next level, evolving to what became known as fake news.

To start from the beginning, fake news is news, stories or hoaxes created to intentionally misinform or deceive people. And these stories are usually fabricated to either influence people’s views, push political agenda or cause confusion.

The process of deceiving starts even before fake news contents reach its readers. In a study called “The spread of fake news by social bots”, researchers from Indiana University explored the role of software-controlled profiles and pages, known as social bots, in the process of making false information go viral. 

Having analysed 122 websites that, according to the established media, constantly publish false and/or misleading information, they found that social bots actively share links in the first few seconds after they are first posted. In many articles, one or two accounts are responsible for the whole activity. Some accounts share the same information up 100 times or more.

It was also found that one of the bots’ strategies is to expose influential people to a claim. By exposing journalists and politicians to an affirmation, social bots create the false impression that the subject is being widely discussed and shared, therefore increasing its probability of being more diffused.

“People tend to trust social contacts and can be manipulated into believing and spreading content produced in this way. To make matters worse, echo chambers [another way to refer to filter bubbles] make it easy to tailor misinformation and target those who are most to believe in it”. 

In another study, “The spread of true and false news online”, researchers from MIT investigated the diffusion of 126 thousand stories on Twitter from 2006 to 2017. Adopting a broader definition of the term news, as “any story or claim with an assertion in it”, they concluded that falsehood spreads significantly farther, faster, deeper and more broadly than the truth. Political fake news achieved the same results when compared to false information about terrorism, natural disasters, science, urban legends, or financial information.

It was noted that while truth seldom outstretched to 1000 people on Twitter, the top 1% of fake news routinely reached 1000 thousand to 100,000 people. Also, it took the truth around six times more to reach 1500 people.

Surprisingly or not, users who spread false information had fewer followers, followed significantly fewer people, had been on Twitter for amazingly less time, were significantly less active on the platform and also were verified less often.

According to the retweeting likelihood model used in the study, false information was 70% more likely to be shared in comparison to true statements. Besides, it was more novel than true stories and displayed higher unique information.

To give some real-life examples to the aforementioned data, analysis from BuzzFeed, back in November 2016, found that fake election news outperformed factual stories on Facebook in terms of engagement in the last three months before the US presidential election. In the social media, which is designed to engage people rather than displaying trustworthy content, the 20 top-performing fake election news had 8,711,000 shares, reactions and comments on the platform. On the other hand, the 20 top-performing elections stories from the 19 major media outlets websites generated 7,367,000, shares, reactions and comments. 

The most engaging post overall was one referring to a statement from Pope Francis endorsing Donald Trump for president. The post, published by Ending the Fed, had 960,000 engagements in total. Donald Trump himself shared the information on Twitter. Pope Francis had never made such an endorsement.

In “Filter Bubbles and Fake News”, researchers mentioned two other independent studies, one from the University of Southern California, and the other from Oxford University, the University of Washington, and Corvinus University of Budapest which were found that AI-controlled bots were spreading pro-Trump content in “overwhelming numbers”. 

Whether it is difficult to determine if false information had either influenced the result of the election or not, research from the Pew Research Center can give us some food for thought.  

In a survey conducted by the institute, in 2015, 61% of Millennials reported relying on Facebook as their main source for government and political news. The social media was also mentioned by 51% of Generation X as the principal media for political and government news; 39% was the percentage among Baby Boomers.  

In an article about the role of news on Facebook, in 2013, about half of US adults Facebook users (47%) “ever” got news there. This percentage accounts for 30% of the US population. 

Despite relying on Facebook for political news, most of US adults do not access Facebook for seeking news. In fact, 78% mostly see the news on the platform for other reasons, as the same article reveals.