Facebook Buying AI to Help Fight Fake News
Source: William Vogeler
Facebook is buying a startup that uses AI to spot fake news, and the company is going to need it.
After the disaster named Cambridge Analytics, Facebook has been fighting a losing public relations battle about fake news. The founder was dragged before Congress to testify about how Facebook users were targeted with propaganda to influence the 2016 presidential election.
If that weren't enough, Facebook algorithms recently branded passages of the Declaration of Independence as hate speech. It may take a lot more than AI to change public opinion in the future.
According to reports -- not fake ones -- Facebook will pay $23 to $30 million to acquire Bloomsbury AI. The startup has developed a natural language software that helps computers read documents and answer questions about them.
In an announcement, Facebook said Bloomsbury had a "built a leading expertise in machine reading and understanding unstructured documents."
"We are incredibly proud that one of our cohorts has attracted the attention of Facebook, a company that will no doubt help it reach its full potential," said Jonathan Brayne of Allen & Overy, which incubated Bloomsbury AI.
Facebook employs people to help flag undesirable content. Even with AI, however, the system is still in development.
Declaration of Independence
The new AI will go to work in London, but it could have helped in the United States recently. A Facebook misstep in Texas could have caused a second revolution.
A newspaper, the Liberty County Vindicator, celebrated the Fourth of July by posting excerpts from the Declaration of Independence. Facebook removed one post, notifying the paper that it violated hate speech standards.
Later, Facebook acknowledged the mistake and restored the post. Guy Rosen, VP of product management, said it was an algorithm problem.