So let's talk about this whole "vibe coding" thing.
If you haven't heard the term yet, it basically means you just tell the AI what you want and let it write the code for you, like having a junior developer who never sleeps and doesn't complain about your specs. You stay in the driver's seat conceptually, but the machine is doing most of the typing. In theory, you focus on the idea while the AI handles the syntax, the loops, the plumbing.
Now, I'm not gonna sit here and pretend I'm above it. I use AI for code all the time. If I need a quick function to sort an array, import a text file, or hammer out some boilerplate VBA that I've written a hundred times before, yeah, I'll let the AI do it. I already know how it works. I just don't feel like spending 20 minutes typing something I've had memorized since the first Bush administration. At this point it's muscle memory, not intellectual exercise.
That's where vibe coding makes perfect sense to me. As an experienced developer, it's a power tool. It's like using a nail gun instead of a hammer. I still know how framing works. I'm just getting the job done faster, and with fewer smashed thumbs.
Where I start to get nervous is when people try to build entire applications this way without understanding what's under the hood. I see folks spinning up full websites, business apps, database systems, all by prompting an AI and copy-pasting whatever it spits out. And hey, it works... until it doesn't. And when it doesn't, it usually breaks at 4:55 PM on a Friday. Then what? When something breaks, slows down, corrupts data, or starts throwing errors that make no sense, you're stuck. If you don't understand the fundamentals, you don't even know where to start.
It's like calculators in math class. I'm all for calculators. I use one every day. But you still have to learn arithmetic first. You still need to understand why 4 + 4 equals 8 before you trust the machine telling you the answer. Otherwise you're just blindly accepting output. Same thing here. AI is fantastic at producing results. It is not automatically fantastic at teaching you how to think like a programmer. And thinking like a programmer is the real skill. (1)
Back when I was learning, I typed code line by line out of books and magazines. Half the time I didn't fully understand what I was typing, but the act of doing it built the mental wiring. You learned patterns. Logic. Cause and effect. Debugging instincts. That friction mattered. Copy, paste, run skips that entire growth process.
The car analogy fits here too. You don't need to know how an engine works to drive. That's fine. Casual users can absolutely vibe code their way to a small app, a hobby project, something non-critical. No problem there. But if you want to build cars, design engines, or fix one when it explodes, you better know what's under the hood, preferably before the explosion. Same goes for software. If it's business critical, if money, customers, or data integrity are involved, you cannot rely on vibes alone.
And this isn't a new sci-fi warning either. Star Trek covered this decades ago.
In the original series episode "The Apple", Captain Kirk and the crew encounter a society completely dependent on a machine called Vaal. It runs their lives. Keeps the weather stable. Provides food. Maintains order. Sounds great... until it starts malfunctioning. The people have no idea how anything works because they've never had to think for themselves. The Enterprise shows up and has to save the day, which I'm pretty sure violates at least three sections of the Prime Directive. Hey, even in the 23rd century somebody still has to be the sysadmin. OK, Scotty, time to three-finger-salute this entire planet...
Then you've got a similar theme in The Next Generation episode "When The Bough Breaks". Picard and the crew discover a highly advanced but dying civilization that automated everything, including reproduction and societal systems. Over time they lost the knowledge to maintain their own technology. When things started failing, they couldn't fix them. They had the tools but not the understanding.
Sound familiar?
That's the long-term risk of vibe coding if it's taken to the extreme. If we outsource all understanding to machines, we create a generation of builders who can't repair what they build.
Now, to be fair, AI is not the villain here. It's a tool. A powerful one. I actually agree with the more balanced take I've been seeing lately. Use AI as an assistant, not a replacement. Let it help you debug, explain, prototype, and accelerate. But still learn the fundamentals. Still build projects yourself. Still struggle a little. (2)
Because that struggle is where the real learning happens.
I've also been watching discussions about how coding tutorials are supposedly "dead" because of AI. I don't buy that either. The format might be changing. Attention spans might be shrinking thanks to short-form social media dopamine machines that make learning anything longer than 30 seconds feel like homework. The job market might be fluctuating. But the need for structured learning hasn't gone anywhere. If anything, it's more important now. Because when AI gives you an answer, you still need to know if it's a good answer. That's the part people forget.
In my world, especially working with databases like Access and SQL Server, fundamentals matter even more. Data integrity. Relationships. Indexing. Concurrency. Error handling. You vibe code that stuff wrong and you don't just get a bug, you get corrupted data or a business that can't operate Monday morning, which is a great way to meet your clients' lawyers. That's not a weekend hobby app anymore. That's real-world impact.
So yeah, I'm not anti vibe coding. I use it. I enjoy it. It's fun watching AI spit out Python or VBA or PowerShell scripts on demand. It feels like living in the future, minus the flying cars we were promised. But I'm also old school enough to believe you should still learn how to code if you want to call yourself a developer. Otherwise you're just steering the ship without knowing how the engines work. And if the engines fail, you're adrift, staring at error messages like they're written in ancient Klingon.
So that's my take. Use the tool. Embrace the speed. Enjoy the convenience. Just don't skip the understanding part if your livelihood depends on the software you're building.
Curious where you land on this. Are you using AI to accelerate code you already understand, or are you letting it build things you couldn't explain if they broke tomorrow?
LLAP
RR
(1) I remember back in high school and college, a lot of kids were convinced half their classes were pointless. You know the type: "Why do I have to learn this stuff? When am I ever going to use algebra? Or chemistry? Or biology?" On the surface, I understand the frustration. Not everyone grows up to be a scientist or an engineer. But that misses the bigger picture. Those subjects aren't just about the raw information. They're about training your brain. They teach logic, sequencing, cause and effect, and how to follow a structured process. It's algorithmic thinking. You might never use the Pythagorean theorem in your day-to-day life, but learning how to apply a formula, follow a repeatable method, and trust a step-by-step system is the real lesson. That mental wiring carries over directly into programming and problem solving whether people realize it or not. What really gets me is when some of those same folks grow up and start loudly questioning well-established science, like climate change or vaccines, while conveniently forgetting they spent half of science class picking dried glue off their hands or trying to light their book bag on fire with a Bunsen burner. It's hard to respect someone's scientific skepticism when they opted out of the learning part back when they had the chance. "Oh really, Nancy? You've 'done your research?' I remember you skipping science class almost every day to go smoke in the bathroom..."
(2) I don't view AI as some looming villain any more than past technological leaps were villains. Computers didn't "destroy society" because they put typewriter manufacturers out of business. Calculators didn't spell the end of civilization because slide rule and abacus makers had to pivot. Automobiles didn't get framed as evil because they replaced horse-drawn transportation. Every major technology shift displaces certain jobs while creating entirely new industries around it. AI is no different. Yes, it's going to automate some roles. That part is unavoidable. But it's also going to open doors for new specialties, new workflows, and new opportunities that didn't exist before. The real skill, as always, is learning how to adapt, how to use the tool effectively, and how to evolve alongside it, because whether people like it or not, AI isn't going anywhere. Just don't fall into the trap of letting it do your thinking for you, because that's the part of the job that actually matters. Embrace it, learn it, use it but don't outsource your understanding to it. Use it to amplify your skills, not replace the need to have them.
Because the moment the tool becomes the only one who knows how your system works, you've already lost control of it.
Live long and prosper,
RR
No comments:
Post a Comment