After the birth of her second child, Chelsea Becker took a one-year unpaid leave from her full-time job as a flight attendant. After watching a video on TikTok, she found a side business: training artificial intelligence models for a website called Data Annotation Tech.
For a few hours each day, Ms. Becker, 33, who lives in Schwenksville, Pennsylvania, sat at her laptop and interacted with an AI-powered chatbot. For every hour of work she was paid between 20 and 40 dollars. From December to March she earned over $10,000.
The boom in artificial intelligence technology has put a more sophisticated spin on a type of work that doesn't require you to leave home. The growth of large language models like the technology behind OpenAI's ChatGPT has fueled the need for trainers like Ms. Becker, who speak English fluently and can produce quality writing.
It's no secret that AI models learn from humans. For years, AI makers like Google and OpenAI have relied on low-wage workers, typically contractors employed by other companies, to help computers visually identify subjects. (The New York Times is suing OpenAI and its partner, Microsoft, over copyright infringement claims.) They could tag vehicles and pedestrians for self-driving cars or identify images on photos used to train artificial intelligence systems.
But as AI technology has become more sophisticated, so has the work of the people who must painstakingly teach it. Yesterday's photo tagger is today's essay writer.
There are usually two types of work for these trainers: supervised learning, in which the AI learns from human-generated writing, and reinforcement learning from human feedback, in which the chatbot learns from how humans evaluate their responses .
Companies specializing in data curation, including San Francisco-based startups Scale AI and Surge AI, hire contractors and sell their training data to larger developers. AI model developers, such as Toronto-based startup Cohere, also recruit in-house data annotators.
It is difficult to estimate the total number of these gig workers, the researchers said. But Scale AI, which hires contractors through its subsidiaries, Remotasks and Outlier, says it's common to see tens of thousands of people working on the platform at any given time.
But as with other types of temporary work, the ease of flexible hours comes with its challenges. Some workers said they had never interacted with recruitment site administrators, while others were cut from their jobs without explanation. Researchers have also expressed concern about the lack of standards, since workers typically do not receive training on what are considered appropriate responses from chatbots.
To become one of these contractors, workers must pass an assessment, which includes questions such as whether a social media post should be considered hate speech and why. Another takes a more creative approach, asking potential clients to write a fictional short story about a green dancing octopus, set in Sam Bankman-Fried's FTX offices on November 8, 2022. (That was the day Binance, a competitor of FTX, said he would buy Mr. Bankman-Fried's company before quickly backing out of the deal.)
Sometimes, companies look for subject matter experts. Scale AI has posted jobs for contract writers who have master's or doctoral degrees in Hindi and Japanese. Outlier has job ads that mention requirements such as academic qualifications in mathematics, chemistry and physics.
“What makes AI really useful for its users is the human layering of the data, and this needs to be done by intelligent, skilled, human beings with a particular degree of expertise and a creative bent,” Willow said Primack, vice president of AI-scale data operations. “As a result we focused on contractors, particularly in North America.”
Alynzia Fenske, a self-published fiction writer, had never interacted with an AI chatbot before hearing a lot from other writers who considered AI a threat. So when she came across a video on TikTok about data annotation technology, part of her motivation was simply to learn as much as she could about AI and see for herself if the fears surrounding AI were justified.
“It's giving me a completely different take on it now that I'm working on it,” said Ms. Fenske, 28, who lives in Oakley, Wisconsin. “It's comforting to know that there are human beings behind it.” Since February, she has been aiming for 15 hours of data annotation work each week so she can support herself while she pursues a career as a writer.
Ese Agboh, 28, a master's student in computer science at the University of Arkansas, was tasked with coding projects, for wages ranging from $40 to $45 an hour. He would ask the chatbot to design a motion-sensing program that helps gym-goers count reps and then evaluate computer codes written by the artificial intelligence. In another case, he would upload a dataset of grocery items into the program and ask the chatbot to design a monthly budget. He sometimes also evaluates the codes of other annotators, which experts say are used to ensure the quality of the data.
He earned $2,500. But her account was permanently suspended from the platform for violating its code of conduct. She received no explanation, but suspected this was because she worked in Nigeria, as the site only required workers based in certain countries.
This is the fundamental challenge of online gig work: it can disappear at any time. With no one available to help, frustrated contractors turned to social media, sharing their experiences on Reddit and TikTok. Jackie Mitchell, 26, has gained a large following on TikTok thanks to her content on side hustles, including data annotation work.
“I understand the appeal,” he said, referring to secondary scams as an “unfortunate necessity” in this economy and “a hallmark of my generation and the generation above me.”
Public records show that Surge AI has the data annotation technology. Neither the company nor its CEO, Edwin Chen, responded to requests for comment.
It is common for companies to hire contractors through subsidiaries. They do this to protect the identity of their clients and this helps them avoid the bad press associated with the working conditions of low-paid contract workers, said James Muldoon, a management professor at the University of Essex whose research focuses on working on AI data.
Most of today's data workers depend on wages from their work. Milagros Miceli, a sociologist and computer scientist who studies labor conditions in data work, said that while “many people do it for fun, because of the gamification that comes with it,” much of the work is still “done by workers ”. who actually really need the money and do it as their main income.
Researchers are also concerned about the lack of security standards in data labeling. Workers are sometimes asked to address sensitive issues, such as whether certain events or acts should be considered genocidal or what gender should appear in an AI-generated image of a soccer team, but are not trained in how to make that assessment .
“It's fundamentally not a good idea to outsource or crowdsource security and ethical concerns,” Professor Muldoon said. “You have to be guided by principles and values and what your company actually decides is the right thing to do on a particular issue.”