November 20 - 09:11
AI spots almost 95% of the hate speech Facebook and Instagram remove
Facebook says AI is becoming increasingly effective at spotting hate speech on its social network. Of the 22.1m examples of hate speech removed on its platform in the third quarter of this year, 94.7% was found by its AI models. That’s up from just 24% in 2017, and 80% in this year’s second quarter. AI also accounted for 95% of the 6.5m example of hate speech on Facebook’s Instagram.
> Tweets