New Jersey Tech, Business, & Creative Resources
AI Is Writing More Code, but Software Developer Jobs Are Rising
Remember when AI was supposed to wipe out software developers? It looks like nobody told the companies that they were hiring.
Despite the explosion of AI coding assistants, software developer job postings in the United States have climbed nearly 15% since February 2025, according to Indeed Hiring Lab data cited by CNBC. During the same period, overall job postings fell by approximately 7%.
For a profession that was supposed to disappear, that’s a pretty interesting plot twist. As AI tools like Claude Code and GitHub Copilot become increasingly capable of writing, editing, and debugging software, this trend will only continue.
Still, don’t begin celebrating just yet. Despite the recovery, not everyone is benefiting. In contrast, younger programmers are having a harder time breaking into the industry.
In other words, AI isn’t eliminating software development jobs. It’s changing who gets hired, what they’re expected to do, and how they build a career.
Software Developer Job Postings Are Up Nearly 15%
Let’s begin with the numbers.
According to Indeed Hiring Lab, software development job postings have increased nearly 15% since late February 2025. At the time, Anthropic introduced Claude Code, an AI coding tool that accelerated fears of programmers’ extinction.
As WhatJobs points out, the timing doesn’t mean Claude Code caused the hiring rebound. By comparing AI coding tools with each other, it provides a useful benchmark for measuring progress.
The AI-doomsday crowd, however, did not quite get their predictions right.
There’s a catch, though.
After a slight rebound, software development jobs are still approximately 27.5% lower than they were before the pandemic. Even though hiring is improving, the industry isn’t back to its former glory.
Moreover, the recovery is not evenly distributed. Between May 2025 and May 2026,
- 71% of the increase in software development postings came from senior-level positions.
- 37% of the increase came from job titles explicitly mentioning AI.
In other words, that tells us what employers are really looking for.
It’s not always necessary for companies to hire armies of programmers. More and more, they’re looking for experienced engineers with the skills to design systems, evaluate AI-generated work, solve complex problems, and keep everything together.
In recent years, writing code has become easier. The ability to understand what code should do, how it fits into a larger system, and whether it works remains valuable.
AI Is Already Writing a Ridiculous Amount of Code
The hype over AI coding tools is, of course, unfounded. In fact, quite the opposite is true.
Back in 2023, GitHub reported that AI was already generating close to half the code in projects where its coding assistance was available.
In the years since then, AI coders have become significantly more sophisticated. It’s estimated that AI-generated code currently accounts for more than half of the code written across GitHub, although the company hasn’t confirmed it.
Perhaps even more revealing is what GitHub has reported. In October 2026, approximately one in three pull requests involved an AI agent, compared with fewer than one in ten in October 2025.
In terms of how software is built, that’s a huge shift. By the end of two years, most code pushed to GitHub could be written by AI agents if current trends continue.
Unsettlingly, much of that code may never be fully readable by humans. Think about that for a second.
Our world may eventually be dominated by machines writing most of our software, while other machines test it, and humans increasingly supervise it. That sounds very efficient. In addition, it raises some uncomfortable questions about security, accountability, and what happens in the event of a malfunction.
Why Are Companies Hiring Developers If AI Does the Coding?
Doesn’t less programming need to be done if AI can handle more coding tasks?
Not necessarily.
According to Eric Kutcher, McKinsey’s North America chair, predictions about the profession disappearing have not materialized. In the meantime, companies continue to hire developers who understand how to work with AI, and they continue to produce more code.
It’s a familiar technology story. Often, businesses do not respond to cheaper or faster tasks by doing fewer of them. In some cases, they do a great deal more.
An AI system might allow a developer to complete a feature in hours rather than days. However, that does not mean the company will eliminate the position. Besides launching new products, it could improve existing systems or take on projects that weren’t feasible before.
There’s also the issue of timing.
Research from the McKinsey Global Institute suggests the earlier decline in software development postings wasn’t exclusively caused by AI.
When the pandemic hit, technology companies were on a hiring spree. In the aftermath of the boom’s end, many pulled back, especially after 2022. In addition to interest rates and corporate budgets, cloud investments and product development cycles all influenced hiring decisions.
In other words, blaming AI for every lost programming job misses the bigger picture. A second explanation was provided by Indeed Hiring Lab economist Sneha Puri to CNBC.
While AI writes the code, companies still require humans to oversee results, improve products, and integrate technology into daily business activities. As organizations move beyond experimenting with AI to determining how to use it, that’s particularly important.
Building an AI-powered tool is one thing. Making it reliable, secure, and compatible with existing business systems is another. And that second part still requires skilled humans.
There’s One Big Problem: Junior Developers Are Getting Left Behind
At this point, the encouraging employment numbers begin to look a lot less encouraging.
According to the Stanford 2026 AI Index, employment among workers aged 22 to 25 in occupations heavily exposed to AI has declined 16% since generative AI became widespread.
Among early-career software developers, the decline approaches 20%. For those just entering the field, that would be a substantial blow. And the reason isn’t particularly mysterious.
Think about the work traditionally assigned to junior developers. In addition to fixing minor bugs, they write straightforward functions, draft documentation, conduct research, and complete relatively routine programming tasks.
The thing is, AI coding assistants are getting better at performing precisely those kinds of tasks, which is bad news for new graduates. A senior engineer can now delegate some of that work to an AI agent instead of assigning it to a junior colleague.
From a company’s short-term perspective, that might look like an easy productivity win.
But there’s a problem with that strategy. Where exactly are tomorrow’s senior developers supposed to come from?
Puri warned that employers could create a talent shortage by not hiring entry-level employees. In the end, senior engineers don’t emerge from computer science programs fully formed.
By working on small projects, making production mistakes, and gradually taking on more responsibility, they develop their expertise.
It is possible that businesses that eliminate those learning opportunities will save money today, but will face a shortage of experienced developers in a decade’s time. Basically, it’s the software industry’s version of requiring five years of experience for entry-level positions.
In the future, entry-level positions might not exist at all.
The Long-Term Job Outlook Still Looks Surprisingly Healthy
According to government projections, software development isn’t on the verge of extinction despite the disruption. The Bureau of Labor Statistics projects that software developer employment will grow approximately 10% between 2025 and 2035.
In that case, employment would increase from 1.72 million to 1.89 million jobs, adding approximately 174,700 jobs. Not exactly the collapse AI evangelists predicted.
As shown in an analysis of BLS data by economist Steve Rattner, employment for software developers increased from May 2023 to May 2025, according to CNBC. Nevertheless, it’s worth keeping some perspective.
Compared to blue-collar data center building jobs, employment grew roughly in line with employment across all occupations. However, employment projections differ from actual job opportunities. In addition to adding jobs, an industry can become significantly more difficult to enter.
An optimistic national employment forecast won’t be of much comfort to a recent computer science graduate sending out hundreds of applications if senior engineers and AI specialists capture the majority of the growth.
AI-Written Code Creates Another Problem: Who’s Checking It?
Another reason experienced developers matter. It still takes someone to make sure all that AI-generated code works.
Researchers at Anthropic have found that human oversight and user input remain important parts of AI-driven workflows. There is, however, no guarantee that the arrangement will last forever.
In the future, companies may reduce the amount of direct human involvement as AI agents become more autonomous. This poses an interesting dilemma.
On one hand, AI can generate enormous amounts of code, potentially accelerating development and lowering costs. On the other, more code creates more opportunities for vulnerabilities, errors, and unexpected behavior.
As GitHub has raised similar concerns about security, they warn that organizations must prevent more problems up front and reduce the amount of human effort required to solve the rest.
Suppose an AI agent creates thousands of lines of code for a business-critical application. The code passes automated checks. It appears that everything is working. The output, however, contains a security vulnerability that nobody noticed because nobody thoroughly examined it.
Who takes responsibility when that vulnerability gets exploited? The AI agent? The engineer overseeing it? The company deploying the software?
That’s not just a programming question. It’s a business risk.
For now, these concerns help explain why employers still value experienced engineers capable of establishing security standards, designing testing procedures, reviewing architecture, and managing production systems.
But there’s a second possibility.
Eventually, companies may need fewer humans to do automated testing and AI-based review, as well. As a result, mid-level developers might feel under pressure after being hired for junior positions.
In other words, today’s hiring rebound doesn’t guarantee every type of software development job is safe.
What Developers and Employers Should Take Away
For anyone entering software development, the lesson isn’t to abandon programming. After all, it’s to recognize that the job is changing. To succeed in the workplace, Puri recommends students learn AI tools as well as communication, interpersonal, and leadership skills.
While that advice makes sense, developers should take it a step further.
Knowing how to prompt an AI coding assistant is useful. Understanding software architecture, debugging unfamiliar systems, evaluating security risks, and explaining technical decisions to non-technical colleagues is considerably more valuable.
The goal shouldn’t be to compete with AI over who can write a basic function faster. That’s probably a losing contest. Rather than relying on AI-generated solutions directly, developers need to understand the larger problems they’re solving.
Employers have responsibilities here, too.
It may be possible to improve this quarter’s productivity metrics by replacing every junior developer with an AI subscription. This could also undermine the company’s talent pipeline in the future. To develop practical judgment, businesses still have to provide mentorship, structured apprenticeships, supervised AI-assisted projects, or other training opportunities for inexperienced engineers.
It is also important that policymakers be careful about interpreting overall employment growth as evidence that the transition is proceeding smoothly.
In the end, the industry can grow, while a generation of prospective developers cannot find a job. Both things can be true.
AI Isn’t Killing Software Development. It’s Reshaping It.
While AI is predicted to replace programmers, employment data reveal a more complicated picture.
As AI handles more coding tasks, software developer job postings are increasing. As a result, engineers and AI specialists are benefiting, but those at the beginning of their careers have a significantly more challenging time finding jobs.
Meanwhile, companies are discovering that generating code and delivering dependable software aren’t the same thing.
AI is capable of writing functions. It can suggest a fix. You can even use it to build substantial portions of your application. However, taking responsibility when things go wrong, making sure code works, and protecting sensitive data remain crucial components of software development.
At least for now.
We shouldn’t focus on whether AI will write most of our code. Increasingly, it appears it will happen sooner rather than later.
The real question is whether the industry can keep developing skilled engineers when AI takes over the very work that once taught them how to become skilled engineers.
There’s a possibility that junior developers can be easily replaced. Can they replace the experience they would have gained over the next ten years? This is a much harder problem to automate.
