It’s Easier for AI to Detect Nipples Than Hate Speech
Some things are just easier for robots to spot, he said.
2 min read
This story originally appeared on Engadget
Today, Facebook released its Q1 earnings, which showed the company can still make more money and attract more users year-on-year despite staggering controversy. During a Q&A session after with Mark Zuckerberg and other executives, the CEO fielded a question on artificial intelligence’s role in automatically detecting harmful content on the platform. Its AI isn’t so good at catching hate speech, but has done a great job intercepting terrorist content. Some things are just easier for robots to spot, Zuckerberg elaborated:
“It’s much easier to build an AI system to detect a nipple than it is to detect hate speech,” he said.
It shouldn’t be a surprise that Facebook’s AI can spot nipples given the platform’s history of banning, then begrudgingly allowing, breastfeeding and nudity in iconic photos. (Though it still doesn’t let folks #freethenipple on Instagram.) But Facebook has outlined practical reasons for why its AI should be combing the platform for nipples. Its adult and nudity policy lists all the content the platform removes by default — not just for users more sensitive to sexual imagery, but to prevent the sharing of content depicting non-consensual acts or underage people.