Asking For Help
Low-paid AI workers in Africa who perform AI and social media content moderation work for major Silicon Valley companies like Meta and OpenAI have published an open letter imploring US President Joe Biden to help ensure fair working conditions in a system that the letter’s 97 signees say amounts to “modern day slavery,” Wired reports.
To build AI models, like OpenAI’s DALL-E image creator or Meta’s “Llama” language model, those companies require a massive — and endless — amount of data, which has generally been acquired by a process of scraping the web. But AI models are what they eat: if you feed them a ton of wretched, unsafe, or low-quality material, their outputs will reflect those undesirable qualities.
That’s where these workers come in. Their job is effectively to clean AI training datasets, removing bad or unsafe content and providing labels for text, images, and other inputs. But this job takes a toll; these folks are tasked with constantly consuming the worst aspects of humanity, all to ensure the safety of what those on the other side of the process experience as a human-free tool.
This work is done at an incredibly low wages. Largely based in Kenya, African content moderators are often paid a measly sum of less than two dollars an hour. In the letter, the moderators also say their work has left them with life-altering post-traumatic stress disorder (PTSD); the signees further claim that they “weren’t warned about the horrors of the work” before starting.
“US Big Tech companies are systemically abusing and exploiting African workers,” reads the letter, which was published Wednesday ahead of Kenyan President William Ruto’s visit to the US this week. “Our working conditions amount to modern day slavery.”
“Any trade-related discussions between the US and Kenya,” signees add, “must take into account these abuses and ensure that the rights of all workers are protected.”
Great Costs
The letter offers a harrowing picture of the African tech workers’ day-to-day lives. The job, the workers say, comes at “great cost to our health, our lives and our families.”
“US tech giants export their toughest and most dangerous jobs overseas,” the document continues. “Our work involves watching murder and beheadings, child abuse and rape, pornography and bestiality, often for more than 8 hours a day.”
Particularly in the case of AI, it’s easy to forget that there are people on the other side of slick, automated tools and apps who work to keep those products safe — especially when the people who do are located in the global south in what, to us, is a largely invisible factory. Whether this letter influences policy remains to be seen. In the meantime, though, it’s a reminder that these people are real — our experience on these apps is proof enough.
More on AI training: AI Companies Running out of Training Data after Burning Through Entire Internet