Earlier this month I discussed a class action lawsuit, originated in the United States, from people who believe their privacy is being violated by the use of Kenyan data labelers to view their video output.
And the data labelers themselves are not happy, according to a 404 Media article “AI is African Intelligence.”
Before I get to the Kenyans, let’s talk about the reality of AI. No, AI output is not 100% generated by computers alone. There is often human review.
In some cases human review is understandable. There was a recent brouhaha when it was publicly highlighted that when a Waymo vehicle runs into a problematic situation, Waymo calls upon a human reviewer to intervene. People’s anger about this is pointless: would they prefer that Waymo NOT call upon a human reviewer, and just let the car do whatever?
Back to Kenya and the Data Labelers Association (DLA) reports of what data labelers actually do.
“Every day, Michael Geoffrey Asia spent eight consecutive hours at his laptop in Kenya staring at porn, annotating what was happening in every frame for an AI data labeling company. When he was done with his shift, he started his second job as the human labor behind AI sex bots, sexting with real lonely people he suspected were in the United States. His boss was an algorithm that told him to flit in and out of different personas.”
I’ve previously seen reports about people in the U.S. reviewing shocking material for social media companies, but it’s a heck of a lot cheaper to outsource the work abroad.
Unless the U.S. Government insists on bringing data labeling work to the United States, in the same way that it wants to bring call center jobs back here.
I do offer one caution: there is a lot of data labeling work that is NOT pornographic. In the identity verification industry, data labelers review real and fake faces, real and fake documents, and the like to train AI models. Such work does not have the emotional stress that you get from watching certain videos.
But it’s still hard work.
Bredemarket 