Why AI won’t eliminate software engineering jobs

The past 12 months have seen remarkable advances in AI coding capabilities. To offer some perspective, there’s a benchmark that measures performance on 500 real-world engineering problems derived from real world software projects. Early last year, state-of-the-art large language models (LLMs) solved only about 4% of these issues. With this momentum in AI coding agents now drawing significant media attention, I’m asked more frequently: “Will this affect software engineering jobs?”

The tipping point in writing this article was a conversation with a prospective investor who wondered whether his child should continue pursuing an interest in computer science. In true engineering style, I’ll begin with a succinct “TL;DR” (too long; didn’t read) summary and then elaborate in detail below:

TL;DR

- “Indeed says engineering jobs are 70% down.” This is misleading.

- “AI will never fix a bug for me.” Time to wake up.

- “AI will change software jobs.” Absolutely.

- “AI will dramatically reduce the number of software jobs in the near future.” Almost certainly not.

- “Is it still worth learning computer science?” More than ever.

If I’ve still got your attention, let’s dive in.

Misreading the Indeed data

It’s true that software development postings on Indeed have dropped below their pre-pandemic baseline by 29%, and many people are blaming AI. However, here’s an inside secret for anyone outside the recruiting and software engineering circles: hiring for highly paid professionals (and many software developers fall into that category) is simply moving away from Indeed. I’ve hired hundreds of engineers over the course of my career, and I can’t recall the last time I posted a software engineering job on Indeed. It’s been years.

For context, consider that “Education & Instruction” job postings are up 46.8% on Indeed over the same period. Do we attribute that to AI magically creating 50% more demand for educators? Hardly. Until we see data from the Bureau of Labor Statistics or another official source, I’m inclined to classify any direct correlation between Indeed’s dip in engineering listings and AI as a “spurious correlation.”

The tech sector experienced a massive spike during the pandemic, followed by a significant correction that resulted in layoffs and hiring freezes. That had little to do with AI; if anything, AI might be reinvigorating the tech market by attracting new capital and creating specialized jobs.

The evolving landscape of AI coding tools

On the flip side, some individuals remain stuck in a world where “coding assistant” equals GitHub Copilot circa 2022. They’ve grown disenchanted with the market, and I can’t blame them. From overly optimistic claims of “25% productivity boosts” to confusing benchmarks focused on programming Olympiad tasks, AI capabilities can be both overhyped and misunderstood. AI is nuanced; it can perform astonishingly well on one task while failing spectacularly on a similar one.

Keep in mind that Olympiad programming tasks are not reflective of day-to-day software engineering. That’s akin to comparing a Certified Public Accountant’s responsibilities with the role of a company’s founding entrepreneur—they’re wholly different. So the next time you hear that a model “beat a human champion” in a programming competition, don’t assume you can hand it your “Instagram-killer” app and watch it code effortlessly. That said, we’re also not in Kansas anymore. It’s no longer 2022, and GitHub Copilot isn’t the leading edge of coding assistance.

Let me use a simile: have you tried doing legal research with the original ChatGPT? Now compare that experience to using the latest ChatGPT Professional. I have, and my personal conclusion—not legal advice, your mileage may vary—is that while I wouldn’t have used the original model for serious legal research, the newer version impressed me enough to recommend it to my attorney.

The same progression applies to coding assistants. They’re evolving from neat-but-limited tools into powerful agents that can shoulder simple tasks while you stay in the driver’s seat. And as rapid as the last 12 months of progress have been, the next 12 promise to be even more transformative. There are still multiple levers in AI software agents that the industry has yet to pull—so stay tuned.

Unlocking value for businesses and developers

This brings me to how AI coding agents can unlock tremendous value for both software engineers and their companies. Recent breakthroughs in agentic AI are driving a generational shift in coding assistance. These tools now tap into a much deeper contextual awareness, scanning entire codebases to suggest, test, and fix solutions aligned with bigger project goals.

For businesses, this translates to delivering sophisticated applications more quickly, adapting to market shifts in real-time, and expanding the boundaries of what’s possible. For developers, AI coding assistants eliminate the drudgery of repetitive, time-consuming tasks, freeing them to focus on creativity, innovation, and strategic problem-solving.

Will this reduce the number of software jobs?

So, if AI handles more routine work, does that mean we’ll need fewer engineers? I don’t think so. Over my career, I’ve seen more than a tenfold jump in engineering productivity, thanks to modern programming languages, open-source libraries, and cloud infrastructure.

And that’s on top of another tenfold improvement before I even entered the field. Innovation is the hallmark of technology jobs. The more powerful our tools become, the more room there is for creativity and value creation. That value translates into greater demand in the job market, not less. Our collective drive toward progress is our best job security.

The future of Computer Science education

Finally, let’s talk about learning computer science. The programming languages, libraries, and tools I use today are entirely different from those I studied in school. Yet that education remains the bedrock of my career. The next generation of computer scientists will be more resourceful, more collaborative, and more powerful than ever before.

Universities will adapt their curricula to meet industry needs, and hands-on experiences—like internships and open-source capstone projects—will help students build real-world skills that position them for success. AI will enhance their capabilities, not render them obsolete.

Conclusion

AI coding agents are evolving at breakneck speed, but they’re far from making software engineers obsolete. Instead, they promise to enhance developer productivity and creativity, leading to new opportunities for innovation. And for anyone wondering whether to invest time and effort in a computer science education: there has never been a better moment to do so.

We've featured the best monitors for programming.

This article was produced as part of TechRadarPro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro