Meta’s New Sphere AI Tool Filters Web Content for Accuracy

Meta Platforms has unveiled Sphere, an AI-powered tool designed to verify open web content. “Building on Meta AI’s research and advancements, we’ve developed the first model capable of automatically scanning hundreds of thousands of citations at once to check whether they truly support the corresponding claims,” Meta says, noting that Sphere has “a dataset of 134 million public webpages — an order of magnitude larger and significantly more intricate than ever used for this sort of research.” Sphere is open sourced, which means third parties may be able to tailor its fact-checking algorithms for specialized use, such as legal, medical and architectural.

While Meta tested Sphere by authenticating Wikipedia content, the company says that while the non-profit Wikimedia Foundation, which runs Wikipedia, is interested in Sphere, the technology is “still in the research phase and not being used to automatically update any content on Wikipedia.”

Meta says it has no plans to use Sphere on its own platforms — including Facebook, Instagram and Messenger — “which themselves have long grappled with misinformation and toxicity from bad actors,” writes TechCrunch.

In June, Meta “began to crack down on fake reviews on Facebook,” notes CNET, which also writes that “Meta also has plans to scrap CrowdTangle, the tool it uses to monitor misinformation on its platforms, after the midterm elections in November.”

Meta says it has developed custom tools to moderate and manage its own content. As a third-party app, Sphere could be the CrowdTangle replacement Meta spokeswoman Erin McPike hinted at last month in response to reports about the demise of CrowdTangle, a “relevance tool” that is likewise platform agnostic, embraced by researchers, activists and scholars.

As detailed on CNET, a Sphere demo using Wikipedia content showed in a video how “Sphere scans citations and then displays whether the information failed or passed verification.” Sphere can also “call attention to questionable citations, allowing human editors to evaluate the cases most likely to be flawed without having to sift through thousands of properly cited statements.”

Sphere identifies irrelevant citations and suggests more appropriate sourcing. Meta says its eventual goal is to build a platform that will help editors like those managing Wikipedia — which adds more than 17,000 new articles added each month in 300 different languages — to “spot citation issues and quickly fix the citation or correct the content of the corresponding article at scale.”

“Meta notes that to train the Sphere model, it created ‘a new data set (WAFER) of 4 million Wikipedia citations”’ and “just five days ago, Meta announced that Wikipedia editors were also using a new AI-based language translation tool that it had built, so clearly there is a deeper relationship there,” TechCrunch writes of the Wikimedia association.

No Comments Yet

You can be the first to comment!

Sorry, comments for this entry are closed at this time.