top of page
  • Let'sData

AI-enabled Facebook News “cure”​ or “booster”​ of Fake News?

On the Facebook yearly review meeting, one of the company’s executives announced the development of an AI system called TL;DR (Too Long; Don`t Read), which will summarize the news in short fragments [1].

It looks like FB is going to use summarized content and AI-optimized headlines to lure users to its new Facebook News service.

Facebook’s technology director, Mike Schrepfer, compared this summarization of AI to a “cure” for social media fake news disease. He also noted that technology will be a key tool in the fight against hate speech, misinformation, and the problem of news content, noting that Facebook now detects 95% of intolerant statements in the feed. [2] This statement was soon commented on by a former Facebook employee, who said that the social network would detect no more than 5% of the text with conflicting content even with such technology.[3]

The news agencies’ representatives did not like the news about such technology since Facebook today absorbs most of the online advertising and creates a platform where misinformation and the spread of contradictory information sources are common. This was also reported in a recent analysis by a group of researchers at Princeton University, noting that Facebook is a place of the overflow of unreliable sources in 15% of cases, which is many times higher than such statistics in Google (3.3%), Twitter (1%) and others. [4]

Essentially, the model for text reduction and speech processing should be a technological breakthrough, given the size of datasets operated by FB. However, the question remains open whether, with current problems with content moderation and fake news, additional processing of unmoderated content will not make it even more difficult to detect?

3 views0 comments
bottom of page