Staring at my computer’s display, I’m watching a Python program run exactly to the specifications I wrote before I started coding. No debugging. No mysterious crashes. No three-hour hunts for a missing semicolon buried in line 1400. It just worked.

And instead of feeling triumphant, I’m wondering if I just witnessed the beginning of the end of my desire to learn Python. That realization landed with all the emotional warmth of a tax audit.

Sorry, let me back up a little.

Call it a disability if you want, but I learn best by doing. Give me a project, let me make mistakes, let me break things badly enough that I have to understand why they broke. Those lessons stick. I can sit through classes, watch endless tutorial videos and read books thick enough to stop small-caliber ammunition, but until I actually need the skill for something real, it lives in short-term memory right next to forgotten passwords and grocery lists.

Early in my career, I wrote a lot of software for my designs. Back then, understanding the tradeoffs between hardware and software wasn’t optional if you wanted robust products. The microprocessors I used didn’t have luxurious high-level language support. If you wanted code, you learned assembly language. Period. The processor didn’t care about your feelings. Fortunately, I enjoyed it.

I was already writing software for my home computer, so writing it professionally felt less like work and more like getting paid for a hobby. For the geeks in the audience, the first program I ever sold was a 6800 StarTrek game modified for ADM3 cursor controls complete with custom sound effects.

Primitive by today’s standards, of course, but at the time it felt like I had personally kicked open the doors to the future.

A few years later, high-level languages like C, Ada and eventually Python started appearing. Unfortunately, by then I had drifted out of design engineering and into management. Once you become a manager, your programming language changes from assembly code to budget meetings and performance reviews. Somehow that felt like a downgrade.

Fast forward twenty years.

Now my projects are dictated mostly by curiosity, hobby and occasionally my wife. I’ve written several times about my Arduino projects. I can still fumble my way through the Arduino flavor of C provided things don’t become too exotic. But Python and I never really bonded. Our relationship resembles two coworkers trapped in an elevator politely pretending to enjoy each other’s company.

Over the years, I’ve taken multiple Python courses trying to fix that problem. I did the exercises. Passed the tests. Felt reasonably competent.
Then a year later I’d have to dig through reference books just to print a message to the screen. Clearly I needed a real project. Something with enough complexity, frustration and caffeine consumption to hammer the knowledge into long-term memory.

So I decided to build a remote monitoring system for my well house.

With two well pumps and a thousand-gallon storage tank, there are plenty of opportunities for disaster. In fact, “opportunities” may be underselling the situation. Mechanical systems on rural property eventually fail with the reliability of sunrise. The idea was simple enough: monitor the pumps, trigger alarms if they run too long and remotely transmit status back to my computer.

Naturally, while designing it I realized there were several other places a remote monitor would be useful. Like confirming I actually locked the chicken coop before some local predator held an all-you-can-eat buffet.

At that point the project began expanding the way all engineering projects do when nobody in management is around to stop you.

This time, instead of immediately diving into code like a raccoon attacking a trash can, I decided to document the system properly before writing a single line. A genuine systems engineering approach. Proof that those years as a systems engineer were not entirely wasted. Yes Andy, I was listening.

Two weeks later I had a fairly complete specification.

And then I made the decision that changed everything.

Why not ask ChatGPT for a starting framework?

At first the discussion was surprisingly productive. ChatGPT pointed out a few holes in my requirements, suggested several reasonable approaches and generally behaved like an enthusiastic junior engineer who had consumed dangerous amounts of caffeine.

Ten minutes later it handed me a functioning program that would have taken me a couple of months to write on my own.

That should have been my first warning sign.

Getting it running required a little fumbling because Linux pathing occasionally feels like a practical joke designed by bitter monks living in caves. But once I fixed the paths, the program ran beautifully.

Not “pretty good.”

Not “close enough for a first draft.”

Beautifully.

 

Sure, there were a few cosmetic issues caused by my own vague specifications, but there was no real debugging. No suffering. No late-night battles with mysterious crashes. No existential despair while staring at log files at 2 AM wondering whether the computer or I had become mentally unstable.
And honestly? That bothered me.

Because somewhere in the middle of all this, I realized I no longer had a compelling reason to learn Python deeply.

I already know enough to tweak and repair code. Anything more than that and ChatGPT can generate it faster than I can type it. The entire learning curve suddenly feels optional. That’s a deeply unsettling thought for someone who spent decades believing technical skill had intrinsic value.

Five years from now, programming may become a niche skill like using a slide rule, balancing a carburetor or remembering ten-digit phone numbers without consulting a glowing rectangle. Sure, a few specialists will still exist, but most people won’t need the skill badly enough to learn it.

And once human programmers have either retired, switched careers or allowed their abilities to atrophy, how long do you think the data centers will wait before monetizing that dependency? History suggests: not long.

This won’t stop with programming. Writing. Artwork. Music. Illustration. Advertising. Editing. The easier AI makes a skill, the less incentive beginners have to struggle through the ugly early stages of learning it. That’s the real danger here.

I still believe AI will never surpass the very best professionals. The truly gifted people will survive because creativity, judgment and originality still matter. But beginners? Beginners are in trouble.

The people currently being displaced by AI aren’t masters of their craft. They’re the people who were supposed to become masters someday. Junior programmers. Copy editors. News writers. Graphic artists. The people who traditionally learned by doing mediocre work until experience slowly transformed them into experts.

Now companies are increasingly saying, “Why hire a beginner when AI is good enough?”

We’re outsourcing our training grounds to machines.

That’s not a small cultural shift. That’s a long-term infrastructure problem disguised as efficiency. Sure, companies will save money in the short term. Unless, of course, the AI accidentally deletes all their files or invents legal citations.

But eventually the talent pool shrinks. Expertise becomes rare. Rare becomes expensive. Very expensive!

And those gigantic data centers powering AI? They aren’t charity projects. They consume staggering amounts of electricity, hardware and money. At some point the bill comes due.

I use AI constantly myself. It edits nearly everything I write. It helps create illustrations, narration and music for my projects. And now it handles most of my heavy lifting in Python too. So naturally I find myself wondering: what skills should I still bother honing? Writing? Critical thinking? The ability to swear creatively at malfunctioning hardware?

Because unlike my old calculator, AI isn’t going to remain a cheap convenience forever. When I switched from a slide rule to a calculator, the calculator didn’t bill me per digit. AI absolutely will.

Industry isn’t investing all that money in data centers to mine for bitcoin, someday we’ll all be staring at monthly subscription tiers for “Premium Thought Processing” while nostalgically remembering when humans still practiced skills simply because they could.

---

And of course, today’s song from Songer… Why Train When AI Reigns

© 2026, Byron Seastrunk. All rights reserved.