In 2012, as a 33-year-old staff writer at The American Prospect in D.C., I had the opportunity to travel around the country, looking for stories that would show how real people were connected to the decisions made by politicians inside the Beltway. On a trip to Colorado, while interviewing voters in a suburban swing district just west of Denver, I met a young couple staying in a homeless shelter who mentioned that they had previously stayed at a slightly rundown hotel, paying weekly rent, until they could no longer afford it. I had been looking for a chance to write a long narrative feature on the rise of suburban poverty during the Great Recession, and when I checked out the hotel, I found many other families living in it and in other hotels nearby.
Back in D.C., I pitched my editor and later returned to the Denver area to live in the hotel for about five weeks. I got to know the people living there—how they lost their home and what life living in a hotel was like. My article, “The Weeklies,” ran in March 2013. I remain incredibly proud of it, because it highlighted the struggle ordinary Americans were facing as the economy slowly recovered from the housing crash.
I thought of all of this recently when I came across a link to a website called In the Weights. Designed to look like a 1980s computer game, it’s a database where you can “find out whether you live on” in large language models such as OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini, and assigns a score based on your prominence. That’s how In the Weights frames it, anyway—a naked appeal to vanity. In reality, the site shows you how much of your work has been used to train AI chatbots. I found that a lot of mine was: I was given a “strength” score of 735, which put me in the top 5 percent.
So I turned to the chatbots themselves for details. Some would not divulge which works of mine were used in their training, but Gemini cited my work at The American Prospect and TNR:
Your Reporting on Class and Poverty: I have access to the core concepts and reporting from your career, such as your pieces for The American Prospect covering the shredded social safety net, food insecurity among military families, and health disparities, as well as your recent coverage of working-class politics and economic anxiety for The New Republic.
There have been more than a few times lately when I wanted to throw my laptop across the room, drop everything, and go live in the woods. This was one of them.
“The Weeklies” wasn’t just a 7,000-word article; it was months of work. In fact, you could say I’d begun work on it even before I’d formed the idea. I had spent most of my twenties in low-paid, low-level journalism jobs, building up the reporting and writing expertise to even contemplate tackling such a story. Journalism jobs famously do not pay well: Into my mid-thirties, by the time I was writing long narratives, my salary remained $50,000. I struggled to pay rent in D.C. while also repaying my student loans from journalism graduate school. (I am still making $611.31 monthly payments and likely will continue to do so until I retire.)
I don’t want to throw too much of a pity party, but as someone who went to college thanks to financial aid and then borrowed to go to graduate school because I didn’t know how else to move into my desired career, I never had family money or connections to rely on. During these years, I struggled to keep up with my bills and to stay one step ahead of the layoffs devastating the entire industry. At various times, I put my student loans in forbearance, had a Honda Civic repossessed, and defaulted on credit cards. It took me years to dig out of that financial hole.
These are fairly common ups and downs for someone who grows up working-class in the U.S. These are also the kinds of trade-offs many people make early in their careers, as they hope to cash in on the experience later. Working hard at a low-paying job is supposed to allow you to step up a ladder to a more stable career. At least, that’s what we’re promised when we borrow money to get an education and then live on ramen, with multiple roommates, as we embark on a career.
AI is not the first technology to upend that promise, but it is the latest—and the one that has touched me most personally. It makes sense that I would be In the Weights. I wrote a lot of words during the early years of internet publishing, creating a body of work that is easy to find and offers free material for the LLMs to suck into their gaping maw. But it is a mistake to think of everything I published under my name as just internet writing. Each word came from years and years of labor.
A number of organizations are trying to tackle how AI is used in the workplace, and how workers might have say in its use and be protected from job loss. The AFL-CIO, in a platform released after its national convention last month, demanded that “working people have a say in how and whether AI and other advanced technologies are developed and deployed.” Former Commerce Secretary Gina Raimondo is working with states, policymakers, industries, and Big Tech to design policy, training, and payment programs that help transition workers who may be displaced by AI. New America released a report late last month about “centering workers in the AI economy,” using lessons learned from major labor disruptions of the past. Even the pope has weighed in.
We definitely need to talk about where we’re headed, but what about where we’ve been? Vermont Senator Bernie Sanders is the rare voice on Capitol Hill to acknowledge that AI “is based on the collective knowledge of humanity and the creative work of tens of millions of people.” Last month, he introduced a bill that would levy a one-time 50 percent tax on the stock of the largest AI firms to create a $7 trillion sovereign wealth fund, which every year would pay out more than $1,000 to everyone in the U.S. It would also give the federal government voting shares and positions on company boards to allow the public to have a say in its future use.
This legislation recognizes the real problem. I lived paycheck-to-paycheck to create the work that helps fuel LLMs today, yet I get zero compensation for it while global investment in AI is in the billions and it’s making Silicon Valley even richer. Elon Musk’s SpaceX absorbed his AI model, Grok, to increase the aerospace company’s value ahead of an IPO that briefly made Musk the world’s first trillionaire. AI may be a bubble, and I agree with the sociologist Zeynep Tufekci that AI can’t possibly steal all of our jobs, but it will make many people extremely rich no matter what. And they will have gotten rich because AI relied on the labor of working people without paying a cent for it.
Of course, this is part of a broader trend in which executives and investors mint millions while workers get an increasingly thin sliver of the pie. If you want one chart to explain why people are mad about the economy right now, look at the Federal Reserve Bank of St. Louis’s chart showing how labor’s share of gross domestic product has declined since the 1950s.
These trends are global, but in the U.S. the divide is more extreme than in other countries. Our productivity keeps going up, but the gains are rewarding capital and pumping up CEO salaries. If the past is any guide, the people who will be the most hurt by AI disruptions are the workers who are later on in their careers who have built up a specific expertise and don’t easily transition into new jobs at the same salary as their previous jobs—and this chart shows that these are the workers who’ve been underpaid this whole time.
While anger at Big Tech is growing, I don’t know if people are angry enough, and I wonder if that’s partly because a lot of the work that has fueled social media platforms and now AI is in creative fields like art and writing, which are misunderstood or esoteric to many Americans. The writing that many people do in their own working lives might be annoying paperwork, like a self-evaluation, or a cover letter in a job application—typing, basically. But writing, whether a quick take on the day’s news or months of reporting, is really the product of months or even years of labor. Writing also demands thinking, which is work in itself. Generative AI can write, albeit very poorly, but it can’t really think—not as humans do. The best it can do is scan all of its inputs—the result of human beings’ thoughts—in order to approximate the output a human would provide.
Big Tech wants us to believe they’re concerned about the effects their models might have on real people, like creating mass unemployment. Anthropic positions itself as an ethical AI company, and its CEO, Dario Amodei, has aligned with the Vatican on its concerns about the human costs of AI. Other tech CEOs make noises about policies like guaranteed income that would allow people to earn salaries even without jobs. But that kind of generosity is at odds with the way they built their models in the first place—by stealing others’ intellectual property.
Only politics can address the problems with AI, but for the past 50 years our leaders in Washington have largely abandoned labor in favor of companies’ capital growth. Wages, taxes, and regulations have undervalued the work and safety of real people, driving the K-shaped economic growth we see now. Political and economic systems won’t change course unless we force them to. To use an example from the past, the Luddite rebellion was not a wholesale rejection of modern advances. The Luddites were textile workers angry about how mechanization was being used to exploit working people, and they protested by destroying automated power looms. My desire to smash my laptop comes from an old tradition.
