Training Day OpenAI’s finances are all over the place, but one thing is clear — not much of the money it spends or earns is going to the low-wage contract workers hired to train its AI. Two OpenAI contractors revealed to NBC News that the kind of work that goes into training large language models (LLMs) like those behind ChatGPT is very much “grunt work” — though in the case of these two individuals, they’re pretty happy to be doing it. As 34-year-old Kansas City denizen and OpenAI contractor Alexej Savreux explained, the job requires a lot of boring-sounding labor: labeling photos, writing predictions for the AI, and other things that help the LLM “learn” how to generate responses better. “We are grunt workers,” said Savreux, “but there would be no AI language systems without it.” “You can design all the neural networks you want, you can get all the researchers involved you want, but without labelers, you have no ChatGPT,” he added. “You have nothing.” Wages of War This kind of low-wage, uninsured contract work is a staple of the machine learning industry, according to Sonam Jindal of the San Francisco-based nonprofit research firm Partnership on AI. “A lot of the discourse around AI is very congratulatory,” Jindal told NBC. “But we’re missing a big part of the story: that this is still hugely reliant on a large human workforce.” While the work and its pay rate don’t exactly seem sexy, Savreaux credits his work-from-home AI training gig that brought…ChatGPT’s Dirty Secret: It’s Powered by "Grunts" Making $15 Per Hour