Matt, a software engineer, looks forward to his four-hour train commute to Pawling, New York, on weekdays. He now has time to work on his own project, a browser-based video game in which he writes every line of code himself.
“I actively try to keep the ax sharp,” said Matt, who declined to use his real name to protect his employment. Over the past six months, Matt’s work has increasingly shifted from coding, problem solving, and software architecture to reviewing code generated by artificial intelligence. He is convinced that this change will weaken his skills, and he does everything he can to keep them intact. “We try not to utilize AI as much as possible.”
His career as a software engineer, typically paid more than $200,000 a year, once felt like a sure thing. But after being laid off last summer and being warned by his current boss to do more with AI, he said he felt his future was bleak.
For a generation of workers like Matt, software engineering promised stability, security, and upward mobility. But as AI changes the way software is developed (according to Google, 75% of code is now written by AI), it is changing professions much faster than the public expected. Software engineers are frustrated, anxious, and trying to adapt to a surprising new reality where the value of their skills is unclear. As a result, they are doubling down on the basics, pursuing new skills to stay relevant, seeking collective action to promote better protection, or even considering exiting the industry altogether, said more than a dozen software engineers interviewed by the Guardian.
According to the U.S. Bureau of Labor Statistics, software engineering will be one of the largest and highest-paying occupations in the United States in 2022, with 1.5 million workers earning twice the national average salary. As the war for talent intensifies, salaries are rising as companies offer bonuses of up to hundreds of thousands of dollars to attract and retain top programmers. According to market research firm SlashData, about 50 million people worked as developers around the world last year.
But more than 600,000 U.S. technology workers have lost their jobs since OpenAI’s ChatGPT was released in 2022, according to Layoff.fyi, a tech layoff tracking company. Meanwhile, the unemployment rate for computer science graduates rose to 7% in 2024 from 6.1% the previous year, and the underemployment rate exceeded 19%, according to data from the New York Fed. U.S. technology jobs posted on Indeed also fell by 36% from 2020 to 2025.
While experts aren’t sure what software engineers will do next, they do agree on one thing. That means that while coding skills may be losing value, the ability to evaluate code created by AI is becoming more important.
“It’s hard to say exactly what the profession will look like in two years’ time, but it’s clear that the skill of writing code is dead,” Buk Klein Teesling, assistant professor of economics at King’s College London, told the Guardian. “AI is significantly enhancing what it means to be a software engineer, and a reasonable measure of success will depend on how well engineers can leverage the technology.”
Software engineers still have a job to do, and AI is just changing that, says Ethan Mollick, associate professor of business administration at the Wharton School at the University of Pennsylvania and author of the forthcoming book “Co-Existence.”
“Now it’s not about who can write the most code,” he said. Instead, he said, the focus is on defining the problem, designing the system, and effectively directing the AI tools. “It changes the skill, and suddenly there’s value in it.”
Before AI, Matt said he was the “leading voice” in implementing solutions, but now “the line between what I’m deciding and what the AI is delivering has definitely blurred.”
Adapting to the AI era
Software engineers are faced with a daunting choice: compete in an increasingly unpredictable industry or pursue a different path.
George Dover, who worked as a software engineer in Portland, Oregon, for six years, became a kindergarten substitute teacher while he looked for a new job after being laid off at Inuit Mailchimp in late 2024.
“It’s very difficult to let go of something that has been such a big part of your personality for so many years,” he says. “What else do I have?”
But Dover didn’t give up. Recognizing the need to understand AI, he generated code to build a website using AI and evaluated it to learn its strengths and limitations. He checked the code for errors, redundancies, unusual AI decisions, bugs, and visual glitches.
“Quality must be rigorously tested,” he said. “Sometimes that trade-off is good, and other times it can lead you down a rabbit hole that takes more time than coding it yourself.”
It worked. Almost two years after being laid off, 400 applications and several interviews later, Dover landed a software engineering job focused on AI.
Dover isn’t the only city experimenting. Teeslink said more non-technical workers are writing code, expanding overall production, which could drive demand for software engineers. Validating code written by AI requires someone to “find vulnerabilities, understand errors, check security, etc., and only programmers can do that,” he said.
But Shriram Krishnamurti, a computer science professor at Brown University, said it’s too early to make definitive statements about the profession, especially since AI only started producing quality code last year. Still, some experts are likely to be weeded out as the need for code reviews increases, he said.
“Some software engineers were well trained for this, but many were not,” he said. “Those who are successful will be successful, and those who are not will need to readjust.”
David Malan, a computer science professor at Harvard University, said human software engineers will continue to be needed, unless for no other reason than the cost of AI. OpenAI spent $8 billion last year competing to build and run models, and Anthropic is expected to spend $3 billion, according to Reuters. It is widely expected that those costs will eventually be passed on to customers. So rather than companies relying entirely on technology, Malan expects to see “a healthier balance of software engineers supported by AI.”
Pros and cons of coding
Over a decade ago, code efforts were coming from all directions.
In December 2013, President Barack Obama launched a $4 billion initiative called “Computer Science for All” to teach computer science to every American student, calling it the “new foundational” skill for economic opportunity. The private sector expressed a similar message, with Mark Zuckerberg and Bill Gates participating in coding tutorial videos from education nonprofit CodeAI. That same year, coding bootcamps exploded, producing more than 2,000 graduates, and the number of graduates grew by more than 1000% by 2020, according to coding bootcamp tracking organization Course Report.
“Learn to code” has become a watchword for building a successful career.
That’s why Sam, who lives in Los Angeles and asked to remain anonymous to protect his employment, pivoted to software engineering 10 years ago. After dropping out of school to get a degree in music, racking up $130,000 in student loans and trying a few odd jobs, he earned a degree in software engineering.
But over the course of his first five years on the job, he went from feeling professionally secure to fearing he might lose everything. Sam said he doesn’t like it when AI takes over the “creative, fun part of the job” and reduces it to the worst part: “reviewing code I didn’t write.”
He is worried about his future, saying he fears the layoffs could end his career as he competes with talent displaced by Google, Amazon and Netflix.
“I’m sitting in my office thinking, ‘What if I opened a food truck? What if I got into forestry?'” he added.
Sam isn’t the only one feeling tired. After more than a decade of promoting coding, the buzz around this profession is fading. In the 2025-2026 academic year, enrollment in computer and information science programs at four-year colleges decreased by 8.1%, and graduate school enrollment decreased by 14%, according to the National Student Clearinghouse, a nonprofit organization that tracks educational institutions.
Tech workers unite in plight
As AI threatens people’s lives, software engineers are becoming increasingly organized and united.
When Caitlin Cote saw her role being reshaped by AI, she decided to quit her job and start a resource center for tech workers caught up in industry disruption.
“I’m not a very senior-level engineer,” she said. “I found that the AI was improving faster than I was improving. That definitely made me really nervous.”
Cote originally became a software engineer to pay off student loan debt. She also teaches at coding bootcamps, where students from low-income settings and domestic violence shelters have shown her first-hand what coding can do to people’s lives. But when her job shifted from writing code to reviewing what AI produced, she started rethinking her career and where she could make a bigger impact.
In February, Cote launched What We Will to help tech workers negotiate layoffs, get unemployed workers access basic income, upskill and unionize. The center’s first campaign, launched in its first month of existence, aimed to help Amazon employees organize against the rapid adoption of AI and mentor laid-off employees through benefits. The second campaign targeted employees who had been fired from Oracle, helping them negotiate severance and establishing support groups. In May, we worked with Meta employees to discuss employee monitoring.
Cote said concerns about AI and mass layoffs are driving membership growth and interest in collective action. She receives at least 10 new applications every day, and “more people have received applications in recent months.” [are] In particular, we are reaching out to seek unionization,” she said.
But she’s largely been built without a roadmap, as few in the industry have done this before. As a result, she turned to the Alphabet Workers Union, the Washington Post Tech Guild, and others for guidance.
“There’s a lot of need right now,” she said. “We just don’t have guilds as an industry. We don’t really have any shared regulations or standards.”
