Each weekday, Matt, a software program engineer, seems ahead to his four-hour prepare commute to Pawling, New York. It’s time he makes use of to work on his personal undertaking: a browser-based online game for which he writes each line of code himself.
“I’m actively making an attempt to maintain my axe sharp,” mentioned Matt, who didn’t wish to use his precise identify, to guard his employment. Within the final six months, Matt’s job has more and more shifted away from coding, downside fixing and software program structure in the direction of reviewing code generated by synthetic intelligence. Satisfied that the shift will weaken his expertise, he’s doing what he can to maintain them intact. “I’m making an attempt to not leverage AI the place I can.”
His profession as a software program engineer, which usually has paid greater than $200,000 yearly, used to really feel like a positive factor. However after a layoff final summer season and a warning from his present boss to make use of AI extra, he mentioned his future feels darkish.
For a era of staff like Matt, software program engineering promised stability, safety and upward mobility. However as AI adjustments how software program is developed – Google says 75% of its code is now written by AI – it’s altering the occupation a lot quicker than the rank-and-file anticipated. Software program engineers are annoyed, anxious and making an attempt to adapt to a startling new actuality by which the worth of their expertise is unclear. Because of this, they’re doubling down on fundamentals, chasing new expertise to remain related, searching for collective motion to push for higher protections and even considering exiting the trade completely, in accordance with greater than a dozen software program engineers who spoke to the Guardian.
Software program engineering was one of many largest and best-paying professions within the US in 2022, with 1.5 million practitioners incomes twice the nationwide median wage, in accordance with the US Bureau of Labor Statistics. Pay had risen amid escalating talent wars, throughout which firms offered bonuses of as much as a whole lot of 1000’s of {dollars} to snag and retain prime programmers. Final 12 months, almost 50 million folks worldwide labored as builders, in accordance with market analysis agency SlashData.
However for the reason that launch of OpenAI’s ChatGPT in 2022, more than 600,000 US tech workers have misplaced their jobs, in accordance with tech layoff tracker Layoff.fyi. In the meantime, the unemployment fee for pc science graduates rose to 7% in 2024, up from 6.1% the earlier 12 months, and their underemployment fee was greater than 19%, information from the New York Fed reveals. US tech job postings on Certainly additionally dropped 36% from 2020 to 2025.
Specialists are not sure what’s subsequent for software program engineers, however they agree on one factor: coding expertise could also be dropping worth, however the capacity to guage AI-written code is changing into extra vital.
“It’s onerous to say what precisely the occupation will appear to be in two years, however it’s clear that the talent of writing code is over,” Bouke Klein Teeselink, assistant professor of economics at King’s School London, instructed the Guardian. “AI is vastly augmenting what it means to be a software program engineer, and the affordable yardstick” for fulfillment relies on how effectively engineers use the tech.
Software program engineers nonetheless have a job to do, AI has simply modified it, mentioned Ethan Mollick, an affiliate professor of administration on the College of Pennsylvania’s Wharton Faculty and writer of the upcoming book Co-Existence.
“Now it’s not about who can write probably the most code,” he mentioned. As an alternative, focus is on defining issues, designing methods and directing AI instruments successfully, he mentioned. “It shifts the talents round, so instantly that’s the place the worth is.”
Earlier than AI, Matt mentioned he was a “main voice” in an answer’s execution, however now “the road between what I’m deciding and what AI gives has definitely blurred.”
Adapting to the AI-era
Software program engineers are grappling with a frightening selection: soldier on in an more and more unpredictable trade or search another path.
George Dover, a software program engineer of six years in Portland, Oregon, grew to become a substitute kindergarten instructor whereas he searched for brand new roles after a layoff at Inuit Mailchimp in late 2024.
“It’s very tough to let go of one thing that was a big a part of your character for a very good variety of years,” he mentioned. “What else is there for me?”
However Dover didn’t surrender. Realizing the necessity to perceive AI, he used AI to generate code to construct web sites after which evaluated it to study its strengths and limitations. He checked the code for errors, redundancies, uncommon AI choices, bugs and visible glitches.
“The standard needs to be rigorously examined,” he mentioned. “Generally that tradeoff is an efficient one, different instances it leads you down rabbit holes that take longer than coding it your self.”
It paid off. Virtually two years after being laid off, 400 purposes and several other interviews later, Dover landed a software program engineering job – one oriented in the direction of AI.
Dover isn’t the one one experimenting. Extra non-technical staff are writing code, increasing general output, which may drive demand for software program engineers, mentioned Teeselink. Validating AI-written code requires somebody to “discover vulnerabilities, perceive errors, test safety, and so forth., which non-coders will be unable to do”, he mentioned.
However it’s too early to make any definitive declarations concerning the occupation, particularly since AI solely began producing good high quality code final 12 months, mentioned Shriram Krishnamurthi, professor of pc science at Brown College. Nonetheless, the rising want for code evaluations will possible weed out some professionals, he mentioned.
“Some software program engineers educated effectively for this, and plenty of didn’t,” he mentioned. “Those that did will thrive; those that didn’t are going to must re-tool.”
Human software program engineers will nonetheless be wanted if for no different motive than the price of AI, mentioned David Malan, a Harvard College pc science professor. Of their race to construct and run fashions, OpenAI reportedly spent $8bn and Anthropic is anticipated to have burnt $3bn final 12 months, in accordance with Reuters. Prices are broadly anticipated to finally cross on to clients. So slightly than firms relying completely on the tech, Malan expects to see a “more healthy stability of software program engineers being supported by AI”.
The highs and lows of coding
Greater than a decade in the past, the push to code got here from each course.
In December 2013, President Barack Obama launched a $4bn initiative known as Computer Science for All to show pc science to all American college students, calling it the “new fundamental” talent for financial alternative. The non-public sector echoed the same message, with Mark Zuckerberg and Bill Gates collaborating in coding tutorial movies for academic non-profit CodeAI. The identical 12 months, coding bootcamps exploded, producing greater than 2,000 graduates, which rose greater than 1000% by 2020, in accordance with coding bootcamp monitoring group Course Report.
“Be taught to code” grew to become a mantra for constructing a profitable profession.
That’s why Los Angeles resident Sam, who requested for anonymity to guard his employment, pivoted to software program engineering a decade in the past. After dropping out of college for a music diploma, racking up $130,000 in scholar loans, and taking a stab at some odd jobs, he pursued a software program engineering diploma.
However in his first 5 years on the job, he’s gone from feeling professionally secure to fearing he may lose all of it. He hates that AI has taken over “the artistic, enjoyable a part of the job”, and lowered it right down to the worst half: “reviewing code I didn’t write”, mentioned Sam.
He’s frightened concerning the future, fearing a layoff would finish his profession by forcing him to compete with displaced expertise from Google, Amazon and Netflix, he mentioned.
“I’m pondering as I sit in my workplace, ‘What if I opened a meals truck? What if I acquired into forestry?,’” he added.
Sam isn’t alone in his malaise. Greater than a decade after the coding push, the thrill for the occupation is fading. Enrollment in pc and knowledge science packages at four-year universities fell 8.1% and graduate enrollment dropped 14% within the 2025-2026 faculty 12 months, in accordance with the Nationwide Scholar Clearinghouse, a non-profit that tracks academic establishments.
Tech staff unite of their plight
As AI threatens their livelihoods, software program engineers are more and more organizing and banding collectively.
For Kaitlin Cort, watching AI reshape her function led her to give up and begin a useful resource heart for tech staff caught within the trade’s disruption.
“I’m not a really senior-level engineer,” she mentioned. “I can see that the tempo at which AI was getting higher was quicker than the tempo at which I used to be getting higher. That positively made me actually anxious.”
Cort initially grew to become a software program engineer to repay scholar mortgage debt. She additionally taught at coding bootcamps, the place college students from low-income backgrounds and home violence shelters confirmed her firsthand what coding may do for folks’s lives. However as soon as her job shifted from writing code to reviewing what AI generated, she started to rethink her profession and the place she may make a higher affect.
In February, Cort began What We Will, which helps tech staff navigate layoffs and negotiations, entry fundamental earnings in unemployment, upskill and arrange unions. The middle’s inaugural marketing campaign, launched in its first month, aimed to assist Amazon staff arrange in opposition to speedy AI adoption and information laid off staff by means of their advantages. Its second marketing campaign focused staff minimize from Oracle, aiding them with severance negotiation and creating help teams. In Could, it labored with Meta staff to debate worker surveillance.
Worries about AI and mass layoffs are driving membership and curiosity in collective motion, Cort mentioned. She receives at the very least 10 new purposes every day, and “in the previous couple of months, so many extra folks [are] particularly reaching out searching for unionization”, she mentioned.
However she’s largely constructing with no roadmap since few within the trade have achieved this earlier than. Because of this, she’s turned to Alphabet Staff Union and the Washington Submit Tech Guild, amongst others, for steering.
“There’s a number of want proper now,” she mentioned. “We simply, as an trade, don’t have a guild – we don’t have rules or requirements which can be actually shared.”