The battle between fact checkers and fake news is one of the more prominent of recent years, with social networks, governments and charities all looking to launch some form of fact checking service.  New research from the University of Waterloo suggests that AI could help in the fight against fake news.

The team developed a tool that utilizes AI algorithms to examine claims made in stories or on social media posts by testing whether they’re supported by other posts and stories on the same topic.

“If they are, great, it’s probably a real story,” the researchers say. “But if most of the other material isn’t supportive, it’s a strong indication you’re dealing with fake news.”

Stemming the tide

The researchers developed the tool in the wake of a growing wave of social media posts and news stories that were designed to deceive and mislead the reader, with the aim being to sow political doubt or even economic gain.

The tool marks the latest effort to fight back against this barrage of falsehood, and in tests in stance detection, was able to spot fake news with around 90% accuracy.

This kind of accuracy level was the benchmark established via the Fake News Challenge, which provided researchers with a dataset of fake news to play with.  The researchers hope that while their tool won’t automate fact checking entirely, it will act as an additional aid to human fact checkers by helping them screen out blatantly fake stories.

“It augments their capabilities and flags information that doesn’t look quite right for verification,” the researchers say. “It isn’t designed to replace people, but to help them fact-check faster and more reliably.”

It’s far from the only effort aiming to help ensure that society is well informed and the spread of misinformation is reduced, but it’s a nice example of the work being done.