Through the course of my reporting, I would come to conclude something even more startling. Not even in Silicon Valley did other companies and investors move until after ChatGPT to funnel unqualified sums into scaling. That included Google and DeepMind, OpenAI’s original rival. It was specifically OpenAI, with its billionaire origins, unique ideological bent, and Altman’s singular drive, network, and fundraising talent, that created a ripe combination for its particular vision to emerge and take over. “I get the sense that Sam is the most ambitious person on the planet,” a former employee says. In other words, everything OpenAI did was the opposite of inevitable; the explosive global costs of its massive deep learning models, and the perilous race it sparked across the industry to scale such models to planetary limits, could only have ever arisen from the one place it actually did.

Against this backdrop, OpenAI began the first phase of its project in late 2021. Under the code names PBJ1, PBJ2, PBJ3, and PBJ4, it thrust teams of Sama workers into more traumatic content-moderation work, for on average between $1.46 and $3.74 an hour. Workers had no idea for whom or why they were doing the project, kept in the dark under the nondisclosure terms of the contract, common in the data annotation industry. What they did know was what was in front of them: the hundreds of thousands of grotesque text-based descriptions that they needed to read and sort into categories of severity. Was it violence or extremely graphic violence, harassment or hate speech, child sexual abuse or bestiality?

Gradually, the work broke many of the workers, the impacts radiating beyond each individual to the people who depended on them in their communities. Only after the release of ChatGPT would they begin to grasp what exactly they had paid for with their peace of mind.

When I wrote the story of Okinyi and the other three Kenyan workers for The Wall Street Journal, OpenAI sought to distance itself from the responsibility of the toll its project exacted. It was Sama that had followed inadequate procedures to protect their workers, OpenAI leadership said; with Sama’s pristine reputation before then, OpenAI couldn’t have known that the workers were struggling.

But the consistency of workers’ experiences across space and time shows that the labor exploitation underpinning the AI industry is systemic. Labor rights scholars and advocates say that that exploitation begins with the AI companies at the top. They take advantage of the outsourcing model in part precisely to keep their dirtiest work out of their own sight and out of sight of customers, and to distance themselves from responsibility while incentivizing the middlemen to outbid one another for contracts by skimping on paying livable wages. Mercy Mutemi, a lawyer who represented Okinyi and his fellow workers in a fight to pass better digital labor protections in Kenya, told me the result is that workers are squeezed twice—once each to pad the profit margins of the middleman and the AI company.

Scale downgraded Kenya to a Group 5 designation: blacklisted. There was also another reason to exit Kenya. By then, Scale was moving on to a new focus, following the demands of the AI industry. OpenAI and its competitors were increasingly searching for highly educated workers to perform RLHF—doctors, coders, physicists, people with PhDs. So went the profit-chasing progression of chatbot development. Those willing to pay money for chatbots were not casual consumers but businesses that expected tools to perform complex tasks such as in science and software development. Kenya did not fulfill the new labor demand. Scale was now recruiting a fresh workforce primarily in the US with a new worker-facing platform called Outlier, offering as much as forty dollars an hour.

It was yet another stark illustration of the logic of AI empires. Behind promises of their technologies enhancing productivity, unlocking economic freedom, and creating new jobs that would ameliorate automation, the present-day reality has been the opposite. Companies pad their bottom lines, while the most economically vulnerable lose out and more and more highly educated people become ventriloquists for chatbots.

The empire’s devaluing of the human labor that serves it is also just a canary: It foretells how the technologies produced atop this logic will devalue the labor of everyone else. In fact, for the artists, writers, and coders whose labor the empires of AI turned into free training data, that is already happening.

Scale’s decision would send Winnie and her family spiraling. By then Millicent had lost her job and Remotasks had been the only thing keeping them afloat. Now they were struggling to feed their kids. Winnie was terrified they would soon be evicted. In her inbox, the email Scale sent to inform workers of the shutdown was cold and clinical: “We are discontinuing operations in your current location,” it read. “You have been off-boarded from your current project.”

To solve OpenAI’s data bottleneck, Brockman turned to a new source: YouTube. OpenAI had previously avoided this option—scraping YouTube to train OpenAI’s models, YouTube’s CEO would later confirm, violated the platform’s terms of service. But under the new existential pressure for more data, the question became whether YouTube, or its parent, Google, would enforce it. If Google cracked down, it could jeopardize its own ability to scrape other websites for its large language model development. Brockman was willing to take the risk.