Want to get a job in the Bay Area? The fastest-growing listings are for operations engineers, applied scientists, senior machine-learning engineers, principal cloud architects, forward-deployed engineers, and the like, according to Laura Ullrich, the director of economic research in North America at the Indeed Hiring Lab. “A year ago, people were very concerned about these types of jobs not coming to fruition,” she told me. Today, “it’s all AI. It’s all AI jobs.”
Or not quite all. Demand for estheticians, Pilates instructors, massage therapists, therapist therapists, and interior designers is also surging. And in a way, those are AI jobs too. It makes sense that someone earning seven figures a year at Anthropic would want a skilled person to untie the knots in their back and another to untie the knots in their psyche. Or that someone working a 72-hour-a-week, “996” schedule at Nvidia would not have time to shop for lamps for their Mission-style mansion on their own. Extraordinary wealth wants extraordinary luxuries.
If history is any guide, what is happening in San Francisco and Santa Clara will happen in Indianapolis, Baton Rouge, and elsewhere too. AI will destroy jobs. AI will create jobs. And AI will create wealth that will create jobs—an obscure and overlooked aspect of how technological change shapes the labor market, and one that, in the long term, might be the most transformative.
[Annie Lowrey: Time to consider becoming a bricklayer]
AI and other general-purpose technologies—powerful and adaptable inventions, such as the printing press and the internet, rather than narrowly applicable (if cool!) innovations like the walkie-talkie and the IV—have first-order, second-order, and third-order effects on employment. ChatGPT and Claude Code, for example, are eradicating jobs for interpreters and call-center employees, and generating jobs for electricians constructing data centers and machine-learning experts constructing algorithms. That’s the first-order effect: direct, immediate, obvious.
Then, AI is creating second-order roles for people applying, supporting, or iterating on the technology. These jobs are often “impossible” to imagine in advance, María Jesús Ramírez, a fellow at the McKinsey Global Institute, told me. To explain, she pointed to the development of the combustion engine in the late 1800s. The machine gave birth to the automobile industry and fueled the railroad boom. Eventually, “it creates aircraft,” she told me, “and now we have pilots.” But who would have looked at a furnace and anticipated the need for cabin stewards and air traffic controllers? More to the point, who would have played around with Gemini a year or two ago and predicted that companies would start hiring robot trainers, deepfake investigators, AI-risk underwriters, and AI-powered perfumers? In a new report, Ramírez and her colleagues forecast that AI could create 500,000 to 2 million such novel positions in the coming years.
But then there are the third-order jobs—let’s call them “wealth jobs”—that technologies create, which are even harder to fathom. Advances such as hydraulic pumps, GPS systems, and Wi-Fi tend to create a small coterie of capitalists with a fortune to spend on indulgences. They also allow businesses to do more with less, lowering prices, bumping up profits, increasing the number of goods and services on offer, and amplifying worker wages. This means the elite aren’t the only ones who get to indulge: Middle-class consumers have more money in their pockets and get a bigger bang for their buck at the store.
“We become richer, and we can spend money on things or have jobs that, to our ancestors, would have been ridiculous,” explained Martha Gimbel, the executive director of the Yale Budget Lab. These jobs tend to be “things where you call your parents and you say, ‘I just paid someone to do X, and they say, ‘Martha, that is such a waste of money.’ Then they pause and go, ‘Was it great?’ And you’re like, ‘It was great, actually.’”
Take wedding planners. “You could have done that job 200 years ago,” she said. “But who would have paid you to do it?” Nobody, more or less. Yet today, perhaps one in 10 American couples hires someone to handle the logistics on their big day. Or CrossFit trainers. “If you said to our ancestors, ‘People are going to pay huge amounts of money to throw tires across a room,’ they would have thought you were crazy,” Gimbel told me. She paused. “To be sure, people doing CrossFit are crazy.” Nevertheless, hundreds of thousands of them exist, forking over $200 a month to get fit. She noted that “water sommelier” is an actual position held by actual people, albeit just a few of them.
[Rogé Karma: Three ways to think about AI and jobs]
Wealth jobs tend to be service jobs: As incomes rise, families spend a higher portion of their paychecks on labor-intensive expenses such as higher education, health care, restaurant meals, and vacations, and a smaller portion on tradable goods, such as televisions and automobiles. Experts anticipate that AI will generate demand for in-person, face-to-face, personalized services in particular—the kind of things that LLMs and robots cannot and will never do. “People don’t want 10 cars,” Tanguy Catlin, also of the McKinsey Global Institute, told me. “They’re going to want to have something different, and I suspect there will be an emotional component to that.” Ullrich pointed to “wellness” services, meaning more jobs for cold-plunge installers, sauna technicians, and massage therapists. She also mentioned an uptick in hiring for house managers, people who will do your laundry and shopping and returns and keep your car registration up-to-date for you. (Swoon.)
Still, the wealth jobs of the future will exist only if AI really generates wealth. If the technology proves too expensive for businesses to use, the average American’s wages won’t go up and they won’t have the cash to hire a therapist to talk them through a mushroom trip. And if the wealth AI produces never trickles down to middle-income Americans—perhaps because AI hollows out white-collar employment—it won’t generate many jobs either. Pretty much everyone I know would love to hire a house manager. Pretty much nobody I know can afford to hire one, yet.