Matt Davis had been working as an electrician for 10 years when he decided to make a career change. Davis, who lives in Squamish, British Columbia, was earning around $65,000 CAD per year ($49,000 USD)—not a lot for someone supporting a wife and young daughter. Plus, the long hours were getting to him. “The two major factors were financial and work-life balance,” he says.

A few of Davis’s friends turned him on to the idea of coding boot camps—intensive programs that teach essential skills to people seeking jobs in computer science. Some of Davis’s friends had themselves graduated from boot camps—and they were earning twice his salary, despite having fewer years of experience in their fields. Davis, 35, had taken computer science classes in college, so he had at least some idea of what the curriculum might look like. “I already knew I had this interest and capacity for it,” he says.

In July 2022, Davis enrolled in a boot camp through Lighthouse Labs, a Canadian for-profit tech education company that offers online training in areas such as language programming, data analytics, and web development. He quit his job as an electrician and took out an $18,000 CAD loan ($13,600 USD) to cover the tuition plus living expenses. Six months later, armed with a newfound virtuosity in things like JavaScript, Express, and React, he graduated.

Davis wasn’t exactly in rarefied air. Since their emergence a little over a decade ago, coding academies for aspiring programmers have become an estimated $1.3 billion industry. More than 600 programs offer courses around the globe. In North America, there are more than 100 academies offering full-time classes, either online or in person, with leading companies like Flatiron School and General Assembly charging around $15,000 on average for courses in everything from front-end web development to cybersecurity engineering. Together, these North American academies offer classes—and the promise of gaining a foothold in the tech industry—to some 25,000 people a year.

But in late May—nearly six months after graduating—Davis hadn’t yet found a job. In fact, he’d landed only two interviews. “Before I took the course, a lot of the people that I spoke to said they were getting jobs before they’d even finished the course,” he says with a hint of dejection. “My experience has not been like that at all.”

The economic environment is one reason. Thanks in part to rising interest rates and fears of a looming recession, the tech sector has been shedding jobs at an alarming rate: More than 800 tech companies have undergone layoffs this year, amounting to about 208,000 job cuts, according to industry tracker Layoffs.fyi.

But arguably more concerning than even a faltering economy, both for Davis and his broader cohort of boot-camp veterans, is the rise of generative artificial intelligence, prompted by the November 2022 debut of ChatGPT. There’s still no consensus as to how exactly AI will transform the economy. Some economists worry it will widen the pool of unemployment; others foresee job losses being offset by new opportunities. One thing is clear: The tech sector will be inexorably altered.

The outplacement firm Challenger, Gray & Christmas found in a recent study that all layoffs caused by AI in May occurred within the tech sector. And a working paper published in March by researchers at the University of Pennsylvania, OpenAI, and OpenResearch determined that a sizable share of computer programmer workloads have been affected by large language models (the machine learning technology that powers generative AI).

In their current form, AI tools mostly act as assistants to help coders perform routine tasks and identify bugs. But they’re growing more sophisticated. GitHub, the Microsoft-owned code repository, integrated OpenAI’s GPT-4 model in March to bring users an AI tool capable of suggesting and helping to fix code through natural language prompts. It can even write unit tests to isolate and assess the functionality of different parts of source code. In a survey of 500 programmers released in June by GitHub, 92% of respondents reported using AI tools in their day-to-day work to increase productivity. But they expressed optimism that these assistants would remain just that.

It might not be long, however, before the assistant outperforms its corporeal overlord: Research published last year in the journal Science found that when the Alphabet-owned lab DeepMind pitted its AlphaCode AI model in competition against human coders, AlphaCode’s performance roughly corresponded to “a novice programmer with a few months to a year of training.”

The exact sort of person, in other words, who might have recently graduated from a coding boot camp.

Though software development has long been a lucrative field, jobs were relatively scarce until the app economy emerged in the early 2010s. Suddenly, established companies and startups alike were looking for workers with coveted programming skills, and everyone from VC billionaire Marc Andreessen to President Barack Obama was espousing the virtues of coding. “Learn to code” soon became a panacea delivered to coal miners, factory workers, and anyone facing job displacement, as well as a smug directive aimed at critics of Silicon Valley’s excesses. And with this new mantra came education companies hoping to train people in—and charge them money for—those skills.

Seemingly overnight, for-profit programs began sprouting up, offering crash courses to prepare enrollees for careers in computer science. Programs varied between full- and part-time, but the premise was generally the same: For a few months’ time and many thousands of dollars (though less than tuition at most universities), aspiring coders—some of whom were coming in with absolutely no professional experience in the computer realm—would emerge bleary-eyed and viable for a high-paying job in an ironclad industry.

Over time, these academies began offering a wider variety of classes, in ancillary fields like marketing, design, and leadership. They also moved from in-person classroom settings to hybrid and even fully virtual instruction—a shift that accelerated during the pandemic.

That growth was fueled by venture capital and a number of high-profile acquisitions. General Assembly, for example, was snapped up by the recruitment firm Adecco Group in 2018 for $412 million. The online program management company 2U acquired Trilogy Education a year later for $750 million. Meanwhile, dozens of smaller, more regional boot camps (some associated with universities) took root throughout North America and beyond, offering their own promises of expedited expertise and vocational changes. By 2020, boot camps were collectively grossing around $350 million in tuition, according to Course Report, an online directory of boot camps.

Because these organizations are often unaccredited, they can update curricula easily—a flexibility that makes them better positioned to help train people for the most in-demand skills at any given time. “They can pivot super fast,” says Liz Eggleston, cofounder of Course Report. Case in point: When the pandemic hit, many boot camps were operating fully online within a week or two.

Today, they’re pivoting quickly toward AI. General Assembly focuses on training people for all levels of jobs in such fields as software engineering, data analysis, and UX design. For years, it didn’t offer much AI instruction to students on other paths. That’s now changing, says Robert Jones, General Assembly’s vice president of product strategy. “We think it’s still fundamentally important that, to be a good software engineer, you need to know how code works,” he explains. “But we’re now introducing parts at the very beginning of the program that provide an overview of large language models and generative AI.”

General Assembly has begun training all incoming software developers on ChatGPT, and the company is now starting to layer in introductory lessons on how large language models work across all of its degree programs. For the higher-level classes, it will offer instruction on building an AI model. For novices, the focus is more foundational: how to write an effective prompt, for example.

Other leading academies, including BrainStation, Flatiron School, and Code Fellows, are also creating their own AI-centric curricula. Like General Assembly, they’re training would-be software developers in prompt engineering and teaching UX and UI designers how to cocreate with AI and spot its biases.

“These places keep up with change, specifically with what employers are looking for,” says Course Report’s Eggleston. With 18% of global venture funding going to AI companies in the second quarter of 2023, boot camps are touting their ability to give students the skills to get in on this new gold rush. Whether their graduates will end up with pans full of pyrite, however, is an open question.

In March 2023, four students who attended the Bloom Institute of Technology (BloomTech) filed a class-action lawsuit, accusing the school of advertising false information about job-placement rates. The coding boot camp, which was founded in 2017 under the name Lambda School, has raised more than $120 million from investors that included GGV Capital, Gigafund, and Stripe. At the time these students were enrolled, in 2020, the school touted an 86% job-placement rate. But an internal report from the same year obtained by co-counsel Student Defense revealed the actual rate was “50% placement for cohorts that are six months graduated.” Additional internal documents revealed that the rate had plummeted to just 30% in the first half of 2020.

BloomTech hit further headwinds last year as layoffs increased across the tech sector and it struggled to move beyond its reliance on questionable income-sharing arrangements with students to pay for their tuition. It ended 2022 by cutting half its staff.

BloomTech is hardly alone in courting controversy. Coding academies have been sources of contention almost since the schools first burst onto the scene. The state of New York settled with the Flatiron School in 2017 after the state’s attorney general alleged the organization “improperly marketed and promoted its job-placement rate and the average starting salary of its graduates.” Four years later, a group of 47 students filed suit against the coding academy Make School, which (like BloomTech) sold would-be developers on income-sharing arrangements to fund their tuition. They accused the school of forcing them into contracts that could cost them as much as $250,000 per graduate.

Critics are now concerned that coding academies’ penchant for hyperbole will cause even more damage, given that they might well be graduating students into a market that has little use for lower-level entrants. “I think everybody’s like, ‘Oh shit,’ a little bit,” says Daniel Chukwuemeka Arisa, a 26-year-old Flatiron alum who graduated last November and, months later, is still looking for work.

“The boot-camp industry, which is already [selling] snake oil, is going to get even worse,” says Aline Lerner, founder and CEO of the recruitment training platform Interviewing.io, which helps place both novice and experienced coders at jobs throughout the tech industry. The way she sees it, “Boot-camp grads are in for a world of hurt” when they realize that the skills their schools have been selling them are being made obsolete by AI. “There’s a world where AI replaces the bottom 10% of engineers and some work that junior engineers do,” she adds. “And junior engineers are boot camps’ sweet spot.”

The fate of junior engineers is being weighed across the industry. Earlier this year, Matt Welsh, a former top engineer at Google and Apple, turned heads when he penned an essay titled “The End of Programming” in the journal Communications of the ACM, published by the Association for Computing Machinery, the world’s largest computing society. “I believe the conventional idea of ‘writing a program’ is headed for extinction,” he wrote. “In situations where one needs a ‘simple’ program, those programs will, themselves, be generated by an A.I. rather than coded by hand.”

Welsh is now the cofounder and CEO of the AI startup Fixie.ai, so he has a vested interest in that forecast. But his ideas are reflected elsewhere, from the vaults of Reddit to the opinion pages of The New York Times.

Even General Assembly’s Jones can’t fully dismiss the cloud of existential uncertainty hanging over the coding sector. “Our pool of applicants may increase because we see [people] coming from other jobs that have been displaced,” he says. “On the other hand, if I were to look at it pessimistically, I would say there may be fewer entry-level [coding jobs].”

Ironically, some of the manual-labor professions that were once targets of “learn to code” solicitations are now seen as the most future-proof. A Goldman Sachs report released in March identified construction, maintenance, and repair jobs as the least affected by AI.

Of course, the tech industry is known for its hype cycles, and there’s always a chance that the one surrounding AI will die down. The economy could course correct, and tech firms could once again go on hiring sprees for developers. Maybe boot camps will become more popular than ever. Maybe.

Matt Davis is certainly rooting for that scenario. But he doesn’t have much time for speculation. He’s too busy trying to find work. Where once he dreamed of a larger salary, he says he’s now focused on “not stagnating, financially speaking.

“If I was to go back,” he says, reflecting on his decision to enroll full-time in a boot camp, “I would just continue working.”