The British fact-checking organization Full Fact published on Wednesday a road map for automated fact-checking. The document argues that fully automated fact-checking is not a far-fetched fantasy but an attainable goal.
We are months — and relatively small amounts of money — away from putting practical automated tools in fact-checkers’ and journalists hands. This is not the horizon of artificial intelligence; it is simply the application of existing technology to fact-checking.
In the report, Full Fact proposes tools that would equip journalists with real-time information to challenge inaccurate claims, even in press conferences or live on TV.
Efforts to automate fact-checking can be grouped into two general categories: those attempting an all-encompassing conceptual framework and those focusing on a specific step in the fact-checking process.
Academic projects have generally fallen in the first category: computer scientists at Indiana University led by Giovanni Luca Ciampaglia, for instance, showed how knowledge networks like Wikipedia could be used to solve simple fact-checking questions. On the other hand, an instrument like Claimbuster focuses specifically on spotting factual claims that could be object of scrutiny.
Full Fact, like others before it, argues for combining several tools into a single automated fact-checking workflow.
The organization is at varying stages of designing or developing some of these tools. With “Hawk” and “Trends,” Full Fact tracks popular factual claims. “Stats,” the most intriguing of the lot, automatically checks claims against potentially relevant data it surfaces from the database of the British Office for National Statistics.
Ciampaglia notes that the report is “an ambitious plan” but that “there are indeed huge automation opportunities that can be reaped from just what is currently available off the shelf.”
Beyond providing details on the Full Fact’s own work, the report summarizes other initiatives while offering categorizations that can help clarify the scope of this field. By posing open questions to researchers and calling for a collaborative approach, it also aims to spark a wider conversation on automated fact-checking.
Bill Adair, Knight Professor of the Practice of Journalism at Duke University, says “the report makes a strong case for international cooperation and common standards,” adding that “automated fact-checking will succeed if it’s a collaborative effort.”
Some of this cooperation is already happening. Automation will be the subject of the (Poynter-awarded) visiting fellowship by Pablo Martin Fernández, Editorial Innovation Director of the Argentinian fact-checking site Chequeado, to Full Fact.
Will Moy, Full Fact’s director, believes the only element missing for automated fact-checking to reach the finish line is funding: “Some ‘shovel-ready’ tools are just waiting to be built.”