AI Didn't Kill Software Development, and Here's Why It Never Will

Goop, Rubber Ducks and Application Enhancements

It’s the first day back after the 2025 holiday season, I’m trying to get my motivation back after nearly two weeks of playing nothing but Rocket League and Arc Raiders and stuffing my face with Swedish Meatballs. My mind, wanting to go back to the comfort of looting goop from the carcasses of my fallen enemies, making it back to Speranza to see the wife and kids once more, feeding my chicken and continuing to add to my rubber duck collection. But knowing that the bills aren’t going to pay themselves and that I haven’t sold enough feet pics to have any impact on my bank account yet, so here I am, trying to figure out what I can be working on to add a little more value to the applications I own and keep my job while I’m at it. And finally, after several hours of what I always refer to as “circling-the-drain” (having no idea where to start, a sort-of programmer’s writers block if you will) I finally found something I felt was pretty useful to work on. Adding more testing and logging info to one of the applications I own. Now, what the hell does any of this have to do with AI? Well, in the last half of the day I found myself using my brain and not AI a lot more than I’ve had to in the last few months. Because what I had chosen to work on isn’t something that I’ve found AI is really any good at yet and may never really be good at. I had to simultaneously solve for a very abstract thought, and at the same time a very in-depth set of technical code and implementations to cover my test cases. This type of situation isn’t merely a “tell AI what to do situation,” it required years of experience and prior knowledge of how and where to start, while learning new concepts along the way.

Sometimes You Gotta Drink the Kool-Aid and Go All-In

I’m going to sidetrack a little by sharing a backstory on how I have had an AI problem in the past, and how I’m finding a healthier balance as of a couple months ago. A few years ago, prior to the massive, big-bang moment for LLMs from OpenAI I had tried out Copilot for the first time. I had heard mixed reviews and wanted to try it for myself. I gave it a couple of days and after some trial and error I determined that it really, really…well it sucked. It was buggy, it got things wrong, it messed up dozens of lines of code. The inline tool (which I still rarely use) was virtually useless. So, I deleted it. Fast forward to the end of 2024 and all through 2025 and it’s night-and-day. In 2024 I got asked to join a team that was going to do some experimenting and report back our findings to the rest of the team. I jumped at the opportunity and installed all the tools again and this time…the results were mind blowing. With the onset of AI agents and the ability to read entire directories of code and projects it was crazy good. So, by about early summer of 2025 I decided to go all in.

I signed up for ChatGPT, Gemini, I started learning to make AI videos, I used AI tools in every Microsoft 365 product they offered it in (and it’s in literally everything) — why? Because tools provide leverage. Leverage provides speed and speed means faster delivery and shorter time to market. It means getting more things done. AI, for me at least, probably increases my productivity by about 60% when it comes to writing code. That may seem absurd but think of how much boilerplate code we write. How much repetition of class setups, mocking unit tests, figuring out and writing unit test scenarios, all that stuff that you have to sit and not only think about, but then pound out letter-by-letter on a keyboard takes so much time. And that stuff to me is boring! I want to make things and provide value; I’ve never been one to get hung up too much on the actual code itself. I want to build things, and I want them to be easy to maintain and deploy. So, that’s to say I don’t favor speed over quality. I favor a balance of speed and quality. Using AI to help me write code gives me all of that ability and more. But I’ll admit at one point in 2025 I went too far, and I am finding my way back to a healthier balance.

A Book!? That Is Some Stone-Age Shiznit

A lot of that was probably a little unnecessary, but I feel like it paints a more holistic view of my own perspective on AI, specifically when used in the world of software. Finally, I’d like to get into the little moment that happened today that will hopefully be the end to my fears I’ve been having of the great AI takeover from calcifying in my mind. I had to do three things today that I haven’t done in months: reference and read from a book specifically related to just my core language (not any frameworks), read several pages of documentation without the help of AI, and I had to dissect, stitch together and determine the proper implementation of the test code I was trying to write. All of this to anyone who’s been doing this for any length of time will not come as any surprise.

But this is my entire point. It was this moment that reminded me that AI isn’t a replacement for human involvement, especially when it comes to writing software. It may replace or reduce the barrier to entry quite a bit for the most rudimentary or entry-level of tasks, like simple websites, basic APIs, or converting from one programming language to another. But when you get an application beyond a local environment, to where it has to scale, be secure, provide logging, be debugged, span multiple discrete systems and all be traced back to a single point of origin…I genuinely bid the folks telling me that AI is going to take my job the best of luck. I’m not unaware of the ridiculousness of saying “today I used my brain” and getting excited about it. But I think for a lot of people in today’s age this honestly is becoming something we relish. The fact that we’re human, the whole point of being here, working in a job we enjoy, doing things we like. It’s easy to get caught up in the hype of AI, and want to be part of that hype for yourself. That’s been me more times than I can count. I want to see what it can do; I want to make stuff faster and use AI to create a conversation around the code I’m writing. I don’t want it to just replace me in my job, but not for the reasons someone might immediately think. Using my brain and solving problems is exactly what drew me to this profession in the first place. And getting that feeling back today for that brief moment, having to read, having to slow down and take time to learn and understand something and not just automate the shit out of it — it reminded me why I care.

The Human Experience

This part might get a little too sentimental for some and if you’re already tired of reading this far, no hard feelings. But what is AI if not literally Artificial Intelligence. If you create a science and technology based literally around how the brain works and how humans think, you cannot tell me it doesn’t then come to a head to such a point where we have to ask ourselves what all this is for, what the point of life is and why the hell we’re all plummeting around a fiery star in the middle of nowhere…you just can’t. So, fellow human, ask yourself this question - what does it mean to you to be human? What is the human experience for you like? What prompted writing this for me today was something that made me feel uniquely human. I had to use my brain and to dig deep to solve a problem. As I see AI transform almost everything around me and have an impact on almost every facet of modern life, I find myself asking this question more and more…what does it mean to be human?

Now, before any of you call the local psych facility or send me links to the anti-unalive hotline, know that I mean this from a caring, exciting and curious perspective. What tickled my brain a little today was that I actually had to use it. I’m not blaming myself or attacking myself over the fact that I went absolutely down the rabbit hole of AI for almost an entire year, no I’m more so thanking myself for doing it. Because in that experience, which is all that life is, I learned about myself. I learned what I really enjoy about being a software developer. And I also learned that sure, AI and robots may be 100 times or even 1000 times faster at things than we could ever be. But that they’ll never replace the feeling you get when you solve a problem on your own, with your own brain power. Most of the time we do things simply because it gives us joy. Even if a robot made Swedish meatballs that taste better than my grandma’s it wouldn’t matter. Because they wouldn’t be my grandma’s Swedish meatballs.

P.S. Thank you if you read this far.

• I really do appreciate anyone who reads my blog. I write it for fun and as a hobby. Becoming a software developer has changed my life in ways I never would have imagined. There are many folks in the world today who have no idea the impact they had on my life by simply making a video or writing a blog post. I want to give back, and I hope I can do the same for someone else.

Thanks for reading, and happy coding,

~ Charles