(ANSA) – ROME, JUL 12 – Meta, the parent company of Facebook, launches Sphere, a tool based on Artificial Intelligence that verifies web content and potential fake news. He is currently ‘training’ on Wikipedia quotes. Sphere’s research team – explains the specialized site TechCrunch – currently is based on 134 million public web pages. Wikipedia is experimenting with this to automatically scan its entries and identify when citations are strongly or weakly supported. The idea behind using Sphere for the encyclopedia is simple: the online platform has 6.5 ;; million entries, every month around 17,000 crowdsourced articles are added on average, a huge amount of material that is not easy to verify for the team of editors in charge of overseeing it. Meta – explains TechCrunch – has no financial agreements with Wikipedia and to train the Sphere model, it created “a new data set of 4 million Wikipedia citations, the most complex ever used for this type of research”. “Our goal is to create a platform to help Wikipedia editors systematically identify citation problems and quickly correct article content,” explains Meta.
As Sphere is an open, open source tool, it can form a basis for AI training models. “Our next step is to train models to assess the quality of the recovered documents, detect potential contradictions, prioritize more reliable sources,” noted Meta which is currently not using this model for its platforms such as Facebook, Instagram and Messenger which in the past they have been at the center of controversy precisely for the spread of fake news and inaccurate content.
(HANDLE).
Source: Ansa
I have been working in the news industry for over 10 years now and I have worked for some of the biggest news websites in the world. My focus has always been on entertainment news, but I also cover a range of other topics. I am currently an author at Global happenings and I love writing about all things pop-culture related.
