I think about this so goddamn often. Even the good uses are trained on slave labor.
Wambalo and other digital workers spent eight hours a day in front of a screen studying photos and videos, drawing boxes around objects and labeling them, teaching AI algorithms to recognize them.
Human labelers tag cars and pedestrians to teach autonomous vehicles not to hit them. Humans circle abnormalities in CTs, MRIs and X-rays to teach AI to recognize diseases. Even as AI gets smarter, humans in the loop will always be needed because there will always be new devices and inventions that'll need labeling.
Humans in the loop are found not only in Kenya, but also in India, the Philippines and Venezuela. They're often countries with low wages but large populations — well educated, but unemployed.
The pay for humans in the loop is $1.50-2 an hour.
"And that is gross, before tax," Wambalo said.
Wambalo, Nathan Nkunzimana and Fasica Berhane Gebrekidan were employed by SAMA, an American outsourcing company that hired for Meta and OpenAI. SAMA, based in the California Bay Area, employed over 3,000 workers in Kenya. Documents reviewed by 60 Minutes show OpenAI agreed to pay SAMA $12.50 an hour per worker, much more than the $2 the workers actually got, though SAMA says what it paid is a fair wage for the region.
It's destroying the environment. It's taking advantage of people who're desperate. It's traumatizing them for dollars an hour--if they're lucky and they aren't denied their pay for no reason. I think about this a lot, that these people were made to look at awful and disgusting and illegal things for the sake of training these stupid AI.
"I looked at people being slaughtered," Wambalo said. "People engaging in sexual activity with animals. People abusing children physically, sexually. People committing suicide."
Berhane Gebrekidan thought she'd been hired for a translation job, but she said what she ended up doing was reviewing content featuring dismembered bodies and drone attack victims.
SAMA says mental health counseling was provided by "fully-licensed professionals." Workers say it was woefully inadequate.
It's just absurd and disgusting and infuriating. Yes the good applications are worth humans working on. It's not a bad thing--if the people employed to do the work are compensated appropriately and cared for. But so many of the uses are just unnecessary.
It just. Sucks. And all they'd have to do to make it suck just a litte bit less would be to pay people appropriately, give them access to the counseling needs they have, treat them like human beings worthy of respect and care on a basic fucking level. It wouldn't resolve the environmental issues or the fact that people are thinking less and less for themselves in the name of getting all of their answers from gen AI but at least they could do one thing to make it a little less The Worst Thing Ever.