Each weekday, Matt, a software engineer, eagerly anticipates his four-hour train journey to Pawling, New York. He utilizes this time to focus on a personal project: a browser-based video game for which he personally codes every element.
“I’m actively working to maintain my skills,” stated Matt, who chose to remain anonymous to safeguard his job. Over the past six months, his professional responsibilities have increasingly shifted from hands-on coding and software architecture to reviewing code produced by artificial intelligence. Concerned that this transition could diminish his technical expertise, he is making a conscious effort to code without relying on AI. “I’m trying to avoid using AI whenever possible,” he added.
Once a stable and lucrative career path, Matt’s role as a software engineer, typically earning over $200,000 per year, now feels uncertain. Following a layoff last summer and an admonition from his current employer to integrate AI into his workflow, he described his outlook as bleak.
For many in his generation, software engineering was synonymous with stability, security, and opportunities for advancement. However, as AI transforms software development, with Google reporting that 75% of its code is now AI-generated, the industry is evolving at a pace that many did not foresee. Software engineers are expressing feelings of frustration and anxiety as they confront a rapidly changing landscape where the value of their skills is increasingly ambiguous. Consequently, they are focusing on fundamental skills, seeking new competencies to remain relevant, advocating for better protections, or even considering leaving the profession altogether, according to insights from multiple engineers interviewed by the Guardian.
In 2022, software engineering was one of the most lucrative and sought-after careers in the United States, with approximately 1.5 million professionals earning double the national median wage, as reported by the US Bureau of Labor Statistics. Competitive compensation had surged due to a talent war, with companies offering substantial bonuses to attract and retain top talent. By last year, nearly 50 million individuals globally were employed as developers, according to market research company SlashData.
However, since the emergence of OpenAI’s ChatGPT in 2022, more than 600,000 technology professionals in the United States have lost their jobs, based on data from Layoff.fyi, a tech layoff tracker. Additionally, the unemployment rate among computer science graduates rose to 7% in 2024, up from 6.1% the previous year, with an underemployment rate exceeding 19%, according to the New York Federal Reserve. Job postings for tech positions on Indeed have also seen a 36% decline from 2020 to 2025.
Experts are uncertain about the future of software engineering, but there is a consensus that while traditional coding skills may be diminishing in value, the ability to assess AI-generated code is becoming increasingly vital.
“It’s difficult to predict what the profession will look like in two years, but it’s evident that the skill of writing code is evolving,” remarked Bouke Klein Teeselink, an assistant professor of economics at King’s College London. “AI is significantly enhancing what it means to be a software engineer, and success is now measured by how effectively engineers can leverage this technology.”
According to Ethan Mollick, an associate professor of management at the Wharton School of the University of Pennsylvania and author of the upcoming book “Co-Existence,” software engineers still have crucial roles, but the nature of those roles has changed. “The focus has shifted from who can write the most code to defining problems, designing systems, and effectively directing AI tools,” he explained. “This reorientation of skills determines where value now lies.”
Matt noted that he was once a key contributor to executing solutions, but the division between his decisions and AI-generated outputs has now become blurred.
Engineers like Matt are faced with a challenging choice: continue in a profession that is becoming increasingly unpredictable or pursue a different career path.
George Dover, a software engineer with six years of experience in Portland, Oregon, took up a position as a substitute kindergarten teacher while searching for new opportunities after being laid off from Inuit Mailchimp in late 2024. “It’s challenging to let go of a role that has been a significant part of my identity for many years,” he reflected. “What else is out there for me?”
Nonetheless, Dover persevered. Recognizing the importance of understanding AI, he utilized it to generate code for website development and subsequently evaluated the output to grasp its strengths and weaknesses. He diligently checked for errors, redundancies, unusual AI decisions, bugs, and visual issues. “The quality needs to be rigorously tested,” he emphasized. “Sometimes the tradeoff is worthwhile; other times, it leads down paths that take longer than coding it manually.”
His efforts were rewarded. Nearly two years after his layoff, following 400 applications and numerous interviews, Dover secured a software engineering position focused on AI.
Dover’s experience is not unique. An increasing number of non-technical professionals are beginning to write code, which may enhance overall productivity and potentially drive the demand for software engineers, according to Teeselink. Evaluating AI-generated code necessitates a skill set that non-coders typically lack, encompassing the need to identify vulnerabilities, understand errors, and ensure security.
However, it remains too early to make sweeping assertions about the profession, particularly since AI began producing high-quality code only recently, as noted by Shriram Krishnamurthi, a professor of computer science at Brown University. Nevertheless, he predicts that the rising demand for code reviews may lead to a culling of some professionals. “Some software engineers are well-prepared for this shift, while many are not,” he pointed out. “Those who are prepared will succeed; those who aren’t will need to adapt.”
Human software engineers will still be essential, primarily due to the costs associated with AI, according to David Malan, a computer science professor at Harvard University. Reports indicate that OpenAI spent around $8 billion in its efforts to develop and deploy models, while Anthropic is projected to have incurred losses of $3 billion last year, as per Reuters. It is anticipated that these expenses will eventually be passed on to consumers.
