Will AI Replace Programmers? A Veteran Engineer on the Future of Software Jobs

Meet Dave: Four Decades in Software (and Still Coding for Fun)

Dave reflects on his enduring passion for coding: “Would I do it all again? Of course! I've been incredibly fortunate—and I'd trade it all… for just a little bit more. I just love to code!” For Dave (and many like him), the thrill of building things hasn’t dulled, and being paid for it is just a bonus.

But as a father whose son has just started a software engineering degree, Dave’s perspective is not just professional—it’s personal. “This advice is as much for my own family as it is for you,” he shares. And if you enjoy Dave’s style? He reminds you (with characteristic humor): “No Patreon here! I'm just in it for the subs and likes, so don’t forget to hit both and turn on notifications for future episodes.”

Which Programming Jobs Are at Risk—and Which Will Survive?

“I'm less worried about any specific language or trendy area,” Dave begins. Instead, he’s focused on one thing: overall competence. He paints a picture—imagine the skills of all programmers graphed in a bell curve. Those at the far right? The most competent, the rare debug wizards? “They’re doing very well today and will be for a long, long time.”

Here's the reality, Dave explains: Even if AI code generators like ChatGPT triple in their ability to churn out neat Python snippets, that’s not what the top engineers are doing all day. “If you’re the kind of person who can debug the call stack for an SRAM cache miss in an interrupt handler of a microcontroller—because of a reference counting bug in a shared pointer off in a library—you’re going to be okay. There just aren’t many humans who are excellent at debugging those kinds of complex synchronization issues. And it’ll be a long time before AI can handle it.”

But what if your job mostly involves small, repetitive tweaks? “If you're the person assigned to change the year in the mail-merge header every January... well, maybe it's time to think about truck driving school." And with a chuckle, Dave acknowledges that even trucks might drive themselves soon. “Actually, I think they'll run in convoys, with a pro driver in the front like a train engineer. Fewer truckers, but with higher-end skills. And programming will be similar.”

The key point: The difference between programming and trucking is that AI-driven programming is about to explode. Even if half of today's programming jobs vanish, just as many—or more—may be created to harness the power of AI. The field is on the cusp of transformation, not extinction.

Programming Isn’t Just About Writing Code

Here’s a nuance that’s lost in most discussions: Being a valuable software engineer is barely about with banging out line after line of code. Drawing on his own tenure, Dave estimates: “I spent, at best, 20% of my time coding. The other 80%? Debugging and validating that code.” That’s not because of poor code quality—it’s because “with rare exceptions, like my personal projects Task Manager and Zip folders, I wasn’t writing brand-new code. Good-money gigs for cranking out fresh main.cpp files are surprisingly few.”

Instead, Dave describes real engineering work—a year spent porting the Windows 95 shell over to Windows NT. That job was all about creative problem-solving for backward and forward compatibility:

• Unicode compatibility: Making sure file shortcuts created on Windows 95 (using ASCII) would still work on NT (using Unicode), and vice versa—even after renaming shortcuts to complex kanji strings.
• Zero room for errors: If shortcuts taken from a floppy disk didn’t work or, worse, crashed Windows 95, they couldn’t fix it with an update—the older OS was "carved in stone."

“Every aspect of the interaction had to be handled on the NT side, in a way that was backward compatible but forward-looking. It wasn’t mainly about writing code—it was figuring out what code even had to be written.”

Dave predicts: In a few years, AI will handle the grunt work—the easy 90% of porting. But the tough parts, like subtle compatibility and making nuanced trade-offs, will be a “very human problem.”

The Evolving Role: From Coding to AI Collaboration

Let’s imagine AI continues its current trajectory for another decade. What then? Dave believes it’s “probably a given” that AI will move from writing subroutines to full turnkey solutions and applications. At that point, the question becomes:

1. Will software engineers treat AI-created solutions as “black boxes”?
2. Or will they jump in, debug the underlying code, or revise the AI’s specification and output until it’s just right?

Either way, the future still involves an engineer collaborating with AI—tuning requirements, testing results, and ensuring the solution matches the need. For the next few years, Dave foresees vast source code trees with lots of AI-generated and AI-assisted code long before the era of fully-formed, test-passing, customer-ready apps. “Some of the more mundane programming jobs that don’t require a lot of skill will go away. But new, technical jobs will appear—like being the liaison between the customer and the AI, writing prompts, and tuning the software output.”

Bottom line: Many entry-level programming jobs will be replaced by AI, but the future for software engineers is still incredibly bright.

Where Human Programmers Still Shine: Your Competitive Edge

So, why will skilled engineers remain essential?

• Complex problem-solving: AI still cannot independently tackle unstructured, deeply creative problems that demand innovation.
• Human insight and intuition: A big part of engineering is understanding human needs, behaviors, and preferences—something AI, lacking intuition, struggles to design around.
• Ethics and nuanced decision-making: Many decisions require judgment and sensitivity to social norms—territory where human values trump algorithms.
• Customization and personalization: Building software tailored to a client’s unique context demands deep, sometimes tacit knowledge that AI can’t yet fully parse.
• AI needs you: Ironically, the evolution of AI means that talented software engineers will be in even greater demand to design, build, and fine-tune AI itself.

In a nutshell: AI can automate and assist, but it won’t replace programmers entirely. Instead, it will push human engineers to focus on the most creative, complex, and interpersonal aspects of software—where our brains still have the upper hand.

How to Secure Your Software Engineering Future (and Maximize Your Value)

Dave’s advice? “Be as far to the right on the competency curve as you can.” The jobs most likely to be displaced are the ones held by those on the lower end of the skill spectrum; the safest bet is to become indispensable.

To illustrate, he shares a hypothetical: “Suppose you have a team of 10 engineers porting old COBOL business code to a modern language. In two years, it might be an AI... and just three engineers to run it. My take? Be in the top third, not the bottom.”

There's still “plenty of gold in these hills,” Dave quips. But the easy days of snagging nuggets right from the stream are over. For the persistent and innovative, though, “it’s the beginning of a whole new gold rush.”

Join the Conversation (and an Extra Resource on Autism & Engineering)

Dave closes by inviting your perspective: “I’d love to hear your thoughts. I do read all the comments.” In a personal aside, he points out his book on Asperger’s and ASD—aimed at anyone working with, living with, or managing someone on the spectrum, informed by his experience and lessons he wishes he'd learned sooner.

And, if you’re wondering: Yes, Dave is just recovering from the flu—so if he “sounded or looked a little funny,” now you know why! 🙂