Honesty's the best policy... So after regretfully letting Russia influence the US presidential election, Facebook took steps to be more open about threats it faces and how Zuck's army deals with them. In its 3rd Transparency Report (they're published every 6 months) Facebook revealed it deleted 2.2B fake accounts — just in the past 3 months.
Whoever wrote that code... deserves a real-life medal. Facebook's artificial intelligence is trained by humans to automatically detect and destroy fake, porn-y, hateful, and other bad content from Facebook. Here are some disturbing but also impressive stats from the report:
- 2.2B: That's the shocking number of fake accounts that were attempted by bad people in Q1 (99.8% of those were caught by AI very quickly).
- 4M: The number of hate-posts (65% were detected by AI).
- 670K: the number of posts about gun sales (70% were taken down by AI, before any users had to complain).
- ~20K: That's how many human moderators Zuck's hired to review and obliterate the bad content that AI misses.
Verifying users' IDs would help... Beer, TSA, opening a bank account. All require proving who you are. But not Facebook (or most social media). Its number of fake account attempts quadrupled from last year — The problem is getting worse. If accounts were linked to real people's names and reputations, lameness could be held accountable.