Computer scientists and tech giants are using algorithms and online data to spot misinformation faster than traditional fact-checkers can.
LONDON — In the battle against fake news, Andreas Vlachos — a Greek computer scientist living in a northern English town — is on the front lines.
Armed with a decade of machine learning expertise, he is part of a British startup that will soon release an automated fact-checking tool before the country’s election in early June. He also is advising a global competition that pits computer wizards from the United States to China against each other to use artificial intelligence to combat fake news.
“I’m trying to channel my research into something that is useful for everyone who’s reading the news,” said Vlachos, who is also an academic at the University of Sheffield. “It’s a positive way of moving artificial intelligence forward while improving the political debate.”
As Europe readies for several elections this year after President Donald Trump’s U.S. victory, Vlachos, 36, is one of a growing number of technology experts worldwide who are harnessing their skills to tackle misinformation online.
Most Read Nation & World Stories
- Couple finds their luggage is overweight because of stowaway dog
- Moderna vs. Pfizer: Both knockouts, but one seems to have the edge
- Sports on TV & radio: Local listings for Seattle games and events
- When a fisherman pulled in his line, he knew he had 'something weird': A 40-pound alligator gar
- One of the world's toughest COVID quarantine regimes is finally ending
The French electorate heads to the polls in the second round of presidential elections Sunday, followed by votes in Britain and Germany in the coming months. Computer scientists, tech giants and startups are using sophisticated algorithms and reams of online data to quickly — and automatically — spot fake news faster than traditional fact-checking groups can.
The goal, experts say, is to expand these digital tools across Europe, so the region can counter the fake news that caused so much confusion and anger during the U.S. presidential election in November, when outright false reports routinely spread like wildfire on Facebook and Twitter.
“Algorithms will have to do a lot of the heavy lifting when it comes to fighting misinformation,” said Claire Wardle, head of strategy and research at First Draft News, a nonprofit organization that has teamed with tech companies and newsrooms to debunk fake reports about elections in the United States and Europe. “It’s impossible to do all of this by hand.”
Researchers have tried to learn from the United States’ run-in with fake news, but the problem in Europe has mutated, experts say, making it impossible to merely replicate U.S. responses to the issue.
European countries have different languages, and their media markets are smaller than those in the United States. That means groups that set up fake news sites in the United States, seeking to profit from online advertising when false claims were shared on social media, are less prevalent in Europe.
With fake news already swirling around Europe’s forthcoming elections, analysts also worry that technology on its own may not be enough to combat the threat.
“There’s an increased amount of misinformation out there,” said Janis Sarts, director of the NATO Strategic Communications Center of Excellence, a think tank in Riga, Latvia, that will hold a hackathon with local coders this month to find potential tech solutions to this trend. “State-based actors have been trying to amplify specific views to bring them into the mainstream.”
Calls for combating fake news have focused on some of the biggest online players, including U.S. giants like Facebook and Google.
After criticism of its role in spreading false reports during the U.S. elections, Facebook introduced a fact-checking tool before the Dutch elections in March and the first round of the French presidential election April 23. It also removed 30,000 accounts in France that had shared fake news, a small fraction of the roughly 33 million Facebook users in the country.
Not everyone, though, has embraced Facebook’s response.
Most German publishers, for instance, have balked at participating in the company’s fact-checking efforts, saying it is the responsibility of the social network, not them, to debunk such claims. German lawmakers are mulling potential hefty fines against tech companies if they do not clamp down on fake news and online hate speech.
Since last year, Google also has funded almost 20 European projects aimed at fact-checking potentially false reports. That includes its support for two British groups looking to use artificial intelligence to automatically fact-check online claims before the country’s June 8 parliamentary election.
It similarly has teamed with French newsrooms to create digital tools, including ways to track trending topics during that country’s election.
David Dieudonné, head of the company’s news lab in France, said the project had debunked 43 reports since February (arguably a relatively small figure), including claims that Saudi Arabia was funding the campaign of Emmanuel Macron, the leading candidate.
“We’re trying something new,” Dieudonné said. “There’s no easy answer for this complicated issue.”
Not all potential solutions, though, are being driven by Silicon Valley’s big beasts.
David Chavalarias, a French academic, has created a digital tool that has analyzed more than 80 million Twitter messages about the French election, helping journalists and fact-checkers to quickly review claims that are spread on the social network.
After the U.S. presidential election last year, Dean Pomerleau, a computer scientist at Carnegie Mellon University in Pittsburgh, also challenged his followers on Twitter to come up with an algorithm that could distinguish fake claims from real news.
Working with Delip Rao, a former Google executive, he offered a $2,000 prize to anyone who could meet his requirements. By early this year, more than 100 teams from around the world had signed on to Pomerleau’s Fake News Challenge.
Using a database of verified articles and their artificial intelligence expertise, rival groups — a combination of college teams, independent programmers and groups from existing tech companies — have been able to accurately predict the veracity of certain claims almost 90 percent of the time, Pomerleau said. He hopes that figure will rise to the mid-90s before his challenge ends in June.
“This is just Round 1 of what we want to do,” said Pomerleau, who expects the teams to share their work with fact-checking groups worldwide. “Next, we want to move toward multimedia content like videos.”
In the rush to find solutions to fake news, some within the industry are taking a decidedly more low-tech approach.
Jimmy Wales, the founder of Wikipedia, recently started a crowdfunding campaign to create a news organization that would combine professional journalists with digital volunteers, who would contribute to reports in a similar way to how articles are created on Wikipedia.
Part fact-checking site, part traditional newsroom, the project — called Wikitribune — was inspired by the effect of misinformation on the U.S. presidential election. Wales said his project would choose subject areas based on the interests of the community of volunteers and paying subscribers to the service, relying more on traditional reporting techniques than high-tech wizardry.