Can AI Combat Fake News?

The phenomenon of “fake news” may have captured the American public’s imagination during the 2016 presidential campaign and the subsequent investigation into Russia’s attempts to sway the election in Donald Trump’s favor via Facebook and other schemes.

Prior to the 2016 presidential election, fake or fabricated news has been utilized by many to spread propaganda and conspiracy theories for many years. Many websites, including InfoWars and Brietbart, have disseminated false news that supports their agendas.

Since the election, however, it has become a political and social issue, and Facebook has become the poster child for websites that fell for the scheme.

Recently, the social media company acknowledged its errors and attempted to make amends with their subscribers. It is now flagging fake news articles that appear in the news feeds of Facebook users. It employs AI to accomplish this.

The company uses artificial intelligence to identify words or phrases that may indicate an article is fake. This task’s data is based on the articles that Facebook users have flagged as containing false information.

Currently, the technology employs four methods to detect fake news. They consist of:

Ranking Websites.

Google was the first to use this technique. It uses facts to determine a website’s score. Clearly, website evaluation is a work in progress. However, as a result of Google’s efforts, the technology has grown significantly.

Consider Facts.

This technique employs engines for natural language processing to analyze the subject matter of stories. Using other models, AI determines if other websites report the same facts.

Predict Reputation.

This technique is based on artificial intelligence and uses predictive analytics and machine learning to predict the reputation of websites based on a variety of factors, including the domain name and Alexa web rank.

Discover Words of Sensation.

Proponents of fake news have utilized sensationalized headlines to attract the attention of a potential audience. Using keyword analytics, this method identifies and flags fabricated news headlines.

The actual detection of these types of articles by artificial intelligence is a challenging task. Obviously, big data analysis is involved, but data veracity is also a factor. Identifying it is essential to determining the veracity of the data. This can be accomplished by weighing the evidence. What happens if a fake news article simultaneously appears on hundreds of websites? In this situation, using the technique of weighing facts could lead AI to conclude that the story is credible. Using the predict reputation method in conjunction with weighing the facts may be helpful, but there may still be complications. For instance, reputable news source websites that do not take the time to verify a news story may publish it in good faith.

Clearly, the use of AI to identify these articles requires further development. A number of organizations are enhancing the capabilities of artificial intelligence. West Virginia University is one of these institutions that is involved.

Reed College of Media, in collaboration with West Virginia University’s Benjamin M. Statler College of Engineering and Mineral Resources, has developed a course that focuses on using AI to identify fake news articles.

Seniors enrolled in an elective computer science course are working in teams to develop and implement their own AI programs for the project.

Fake News Challenge is also attempting to find a way for artificial intelligence to effectively combat fake news. It is a grassroots organization comprised of over one hundred volunteers and seventy-one teams from academia and business to combat fake news. It is developing tools to help people verify the veracity of news stories and identify hoaxes.

As organizations work to improve AI in order to discover these stories, a variety of tools are available to combat them. These consist of:

• Spike, which identifies and predicts viral stories and uses big data to predict what will drive engagement, identifies and predicts viral stories.
• Hoaxy is a tool that helps users identify websites that publish fake news.
• Snoopey is a website that assists in identifying fake news articles.
• CrowdTangle, a monitoring tool for social media content.
• Meedan is a verification tool for breaking online news.
• Google Trends monitors search queries.
• La Decodes From Le Monde is a database of fake news websites and legitimate news websites.
• Pheme is a verification tool for user-generated and online content.