Regional elections of Spain It’s still almost four months away, but Irene Larraz and her team at Newtral are ready for impact. Every morning, half of Larraz’s team at the Madrid-based media company schedules political speeches and debates, preparing to fact-check politicians’ claims. The other half, weeding out misinformation, scouring the web for viral misinformation, and working to infiltrate disinformation groups. With the May elections approaching, a national election must be held before the end of the year, which will likely lead to a flurry of online fake news. “It will be quite difficult,” said Larraz. “We are prepared.”
The rise of misinformation and online propaganda means an uphill battle for fact-checkers worldwide who have to sift through and verify vast amounts of information in various situations. complex or fast-paced, such as Russia Invades Ukrainethe Pandemic caused by covid-19, or election campaigns. That task has become even more difficult with the advent of chatbots that use large language models, such as OpenAI’s ChatGPT, which can produce natural-sounding text at the touch of a button. , which essentially automates the generation of misinformation.
Faced with this asymmetry, fact-checking organizations are having to build their own AI-based tools to help automate and accelerate their work. It’s not a complete solution, but fact-checkers hope these new tools will at least keep the gap between them and their competitors from widening too quickly, at a time when the companies are out. Social media companies are scaling back their own censorship practices.
Tim Gordon, co-founder of Best Practice AI, an artificial intelligence strategy and governance consulting firm, and trustee of a UK fact-checking charity, for knows: “The race between the fact-checkers and the people they’re checking is an unequal one.
“The fact-checkers are often small organizations compared to those that create disinformation,” says Gordon. “And the scale of what innovative AI can create, and the speed at which it can do so, means this race is only going to get tougher.”
Newtral started developing its multilingual AI language model, ClaimHunter, in 2020, funded by profits from their TV wing, which produces a show reality check politiciansand documentaries for HBO and Netflix.
Use of Microsoft Language model BERT, the developers of ClaimHunter used 10,000 statements to train the system to recognize sentences that appear to include statements of truth, such as data, numbers, or comparisons. “We are teaching the machine to play the role of a fact checker,” said Rubén Míguez, chief technology officer of Newtral.
Just identifying the statements of political figures and social media accounts that need to be checked is a difficult task. ClaimHunter automatically detects political statements made on Twitter, while another app transcribes video and audio coverage of politicians into text. Both identify and mark statements that contain statements concerning public life that can be proved or disproved—such as unambiguous statements, questions, or opinions—and flag those claims are for review by Newtral’s fact-checkers.
The system isn’t perfect and sometimes marks comments as true, but its flaws help users to continually retrain the algorithm. Míguez says it has cut the time it takes to identify sentences worth checking by 70 to 80 percent.