AI tools are automating real parts of software development: boilerplate code, routine bug fixes, documentation, and repetitive create, read, update and delete (CRUD) work. That's not in dispute. What's less settled is what that means for the profession as a whole, and the data points toward augmentation, not replacement. Software developer employment reportedly reached a record high in early 2025, up sharply since the launch of mainstream generative AI tools — hardly the pattern you'd expect if AI were simply eliminating the job.
The more accurate picture is that risk isn't spread evenly across the profession. Entry-level roles built around repetitive, well-defined coding tasks are absorbing most of the disruption. Roles built around system architecture, distributed systems and security engineering remain comparatively insulated because they require judgment calls AI tools aren't positioned to make on their own.
What the job growth data actually shows
According to the U.S. Bureau of Labor Statistics, employment of software developers, quality assurance analysts and testers is projected to grow 15% from 2024 to 2034, much faster than the average for all occupations, with about 129,200 annual openings and a median annual wage of $133,080 as of May 2024. The BLS attributes that growth to continued expansion in AI, robotics and automation software, the very technologies raising the "will AI replace me" question in the first place.
Job posting data tells a similar story: Software engineering postings were up 28% year-over-year, according to one market research platform's analysis. While hiring slowed for a stretch as companies right-sized after a pandemic-era hiring surge and redirected budgets toward AI infrastructure, postings for developers have continued to grow even through that adjustment.
None of this means every software job is safe by default. But it does mean the "AI is ending software engineering" narrative doesn't hold up against the employment data.
According to McKinsey & Company, AI-powered tools are increasing the productivity of the average developer by 20% to 45%. Another study by Gartner shows that 80% of software engineers will need to upskill in AI by 2027.
The bottom line is that AI won’t replace the profession, but workers will need to learn these new tools to remain competitive in the workplace.
What software engineering skills still require human judgment
The skills gaining the most value aren't the ones AI tools replicate well. They're the ones that sit upstream and downstream of writing code:
- Requirements gathering, stakeholder alignment and translating a vague business need into a technical plan
- System architecture and technical strategy decisions
- Security engineering and threat modeling
- Evaluating and validating AI-generated code, beyond just producing it
- Communicating tradeoffs to non-technical teams and leadership
A CompTIA survey of more than 1,100 U.S. businesses found that nearly 4 in 5 companies that tried substituting AI for human work later backtracked, citing AI underperformance, scalability limits, cost and workflow integration problems. Separately, about a third of hiring managers who eliminated a role due to AI ended up rehiring for the same or a similar position. Companies are learning in real time that AI can accelerate a skilled engineer's output, but it doesn't replace the judgment that makes that output usable.
How the role of a software engineer is evolving
Every hiring cycle now includes AI literacy as a baseline expectation, not a specialty. But the deeper shift is in who gets hired. SignalFire State of Talent Report found that, as junior-level, repetitive coding work compresses, demand is consolidating around "super ICs” — individual contributors with deep technical skills and toward management roles, rather than entry-level generalist positions. Companies increasingly want engineers who can effectively direct AI tools.
Magdalena Balazinska, director of the University of Washington's Paul G. Allen School of Computer Science & Engineering, put it directly to more than 2,000 undergraduates concerned about their future: "AI is not killing your job options. It's expanding them." Companies are realizing that AI needs human drivers who can manage and understand the outputs of AI coders, thereby boosting the amount of software produced and the products made.
Her reasoning echoes a pattern economists have seen before. Boston University's James Bessen has pointed to 19th-century textile automation as a precedent, where technology that reduced costs and improved quality didn't shrink the workforce; it grew demand enough to expand it. Software development, he argues, may be following the same arc.
That doesn't mean the day-to-day job looks the same as it did five years ago. Engineers today spend less time on repetitive, boilerplate work and more time reviewing, directing and integrating AI-assisted output, shifting the profession toward architecture, oversight and judgment.
The degree that prepares you for both today's job and tomorrow's
If you're choosing between a computer science and a software engineering degree with this shift in mind, it helps to know how they differ.
ASU's Bachelor of Science in software engineering, offered both online and in person, is an ABET-accredited, applied program built around project-based learning. It concludes with a capstone experience that gives you hands-on practice in architecture and systems design skills that remain in high demand as AI absorbs more routine coding work. It's designed to build the kind of workforce-ready, senior-track skills employers are prioritizing right now.
ASU's Bachelor of Science in computer science, also offered both online and in person, is ABET-accredited but leans more theoretical, with dedicated coursework in AI and machine learning for students who want to build the underlying technology itself, not just apply it.
Neither path treats the human side of the job as optional. Across ASU's technology programs, building sound technical judgment, not just technical skill, is treated as core to the curriculum, because that's the part of the job AI still can't do for you.
