ChatGPT is powered by these contractors to earn $15 an hour

Aleksej Savreux, 34, who lives in Kansas City, says he’s done all kinds of jobs over the years. He made fast-food sandwiches. He has been a mentor and a junk-holler. And they’ve done a good job technically for live theatre.

These days, though, her job is less practical: She’s an artificial intelligence trainer.

Savrex is part of a hidden army of contract workers doing the behind-the-scenes labor of teaching AI systems how to analyze data to generate the kind of text and images that have fueled newly popular apps like ChatGPT. Have impressed people by using the products. , To improve the AI’s accuracy, it has labeled photos and made predictions about what text the apps should generate next.

Pay: $15 per hour and up, with no benefits.

Out of the limelight, Severex and other contractors have spent countless hours over the years teaching OpenAI’s systems to respond better to ChatGPT. Their response fills an immediate and endless need for the company and its AI competitors: providing streams of sentences, labels and other information that serve as training data.

“We are hard workers, but without it there would be no AI language system,” said Savreux, who has worked for tech startups including OpenAI, a San Francisco company that released ChatGPT in November and set off a wave of hype around generative AI. Did.

“You can design all the neural networks you want, you can involve all the researchers you want, but without labelers, you have no ChatGPT. You have nothing,” Savreux said.

It’s not a job that will bring fame or money to Severex, but it is an essential and often overlooked job in the field of AI, where the seemingly magic of a new technological frontier can take a toll on the labor of contract workers. .

“A lot of the discourse around AI is very congratulatory,” said Sonam Jindal, program lead for the Partnership on AI, Labor and the Economy, a San Francisco-based nonprofit promoting research and education around artificial intelligence.

“But we’re missing a big part of the story: It’s still dependent on a large human workforce,” she said.

The tech industry has for decades relied on the labor of thousands of low-skilled, low-wage workers to build its computer empire: from punch-card operators in the 1950s to recent Google contractors who worked their way up to second-class status. complained about, including the yellow badge that separates them from full-time employees. Online gig work through sites like Amazon Mechanical Turk became even more popular at the start of the pandemic.

Now, the booming AI industry is following a similar playbook.

Work is defined by its volatile, on-demand nature, with people employed by written contracts either directly by a company or through a third-party vendor that specializes in temporary work or outsourcing. Benefits like health insurance are rare or non-existent – ​​which translates to lower costs for tech companies – and the work is usually anonymous, with all credit going to tech startup executives and researchers.

The Partnership on AI warned in a 2021 report that an increase in demand for “data enrichment work” was coming. It recommended that the industry commit to fair compensation and other better practices, and last year it published voluntary guidelines for companies to follow.

Google’s AI subsidiary DeepMind is the only tech company to have publicly committed to those guidelines so far.

Jindal said, “Many people have recognized that it is important to do this. The challenge now is to persuade companies to do it.”

“This is a new job that is being created by AI,” she said. “We have the potential for this to be a high-quality job and value the respect for the workers who do this work and the contribution they make to enabling this advancement.”

Demand has accelerated, and some AI contract workers are asking for more. Time magazine reported that in Nairobi, Kenya, more than 150 people who work on AI for Facebook, TikTok and ChatGipt voted to form a union on Monday, citing low pay and the mental toll of the work. Facebook and TikTok did not immediately respond to requests for comment on the vote. OpenAI declined to comment.

So far, AI contract work hasn’t inspired a similar movement among Americans quietly building AI systems word-for-word in the US.

Savrex, who works from home on a laptop, got into AI contracting after seeing an online job posting. He credits the AI ​​gig work—along with a previous job at sandwich chain Jimmy John’s—with helping him out of homelessness.

“People sometimes downplay these essential, painstaking tasks,” he said. “This is the essential, entry-level area of ​​machine learning.” $15 an hour is more than the minimum wage in Kansas City.

The job posting for AI contractors refers to both the allure of working in a cutting edge industry as well as the sometimes grinding nature of the work. An ad from Invisible Technologies, a temp agency for an “advanced AI data trainer,” notes that the job will be entry-level with pay starting at $15 an hour, but also that it be “beneficial to humanity.” Can

The job posting says, “Think of it like a language arts teacher or a personal tutor for some of the most influential tech in the world.” It doesn’t name Invisible’s client, but says the new hire will work “within protocols developed by world-leading AI researchers.” Invisible did not immediately respond to a request for more information on its listing.

There is no fixed number of how many contractors work for AI companies, but it is a common form of work around the world. Time magazine reported in January that OpenAI relied on low-wage Kenyan laborers to label text that contained hate speech or sexually offensive language so that its apps could better identify toxic content on their own.

OpenAI has hired about 1,000 remote contractors in places such as Eastern Europe and Latin America to label data or train company software on computer engineering tasks, online news outlet Semafor reported in January.

OpenAI is still a small company, with about 375 employees as of January, CEO Sam Altman said on Twitter, but that number doesn’t include contractors and doesn’t reflect the full scale of the operation or its ambitions. A spokeswoman for OpenAI said no one was available to answer questions about its use of AI contractors.

The task of creating data to train AI models isn’t always easy, and sometimes it’s complex enough to attract AI entrepreneurs.

Jatin Kumar, a 22-year-old in Austin, Texas, said he’s been doing AI work on a contract for a year since graduating college with a degree in computer science, and he said it gave him a glimpse of how AI technology is generating gives. leadership in the near term.

“What this allows you to do is start thinking about ways to use this technology before it’s even in the public markets,” Kumar said. He is also working on his tech startup Bonsai, which is building software to help with hospital billing.

Kumar, a conversational trainer, said his main job has been generating signals: participating in back-and-forth conversations with chatbot technology that is part of a longer process of training an AI system. The tasks become more complex with experience, he said, but he started very simply.

“Every 45 or 30 minutes, you’ll get a new task, new signals generated,” he said. Hints can be as simple as, “What is the capital of France?” They said.

Kumar said he worked with about 100 other contractors on the tasks of generating the training data, correcting the answers, and fine-tuning the model by providing feedback on the answers.

He said other employees handled “flagged” conversations: reading examples submitted by ChatGPT users who, for one reason or another, reported the chatbot’s answers to the company for review. When a flagged conversation arrives, he said, it is sorted based on the type of error involved and then used in further training of the AI ​​model.

“Initially, it started as a way for me to help OpenAI and learn about existing technologies,” Kumar said. “But now, I can’t see myself walking away from this role.”