
Artificial intelligence (AI) can write your app in a weekend – just don’t expect it to work come Monday.
-
AI can generate code quickly, but it lacks architectural vision and long-term maintainability.
-
“Vibe coding” creates security vulnerabilities and technical debt that non-programmers can't fix.
-
Experienced developers find AI useful as a tool, but warn against blind copy-pasting.
-
AI-generated code often breaks in production, requiring skilled developers to clean up.
“I hate fixing AI-written code. It solves the task, sure. But it has no vision,” said a backend developer at the office. I looked up. That line stuck with me.
He wasn’t talking about broken logic or bad formatting. He meant that something deeper was missing, any sense of architecture, intention, or care. It worked, technically and for now, but felt soulless.AI coding tools are fueling a new trend in the tech scene: vibe-coding. The term was allegedly coined by Andrej Karpathy, co-founder of OpenAI, posting about “giving in to the vibes and forgetting that the code even exists.” The vibe part got stuck. Vibe coding is essentially prompt-based programming. You describe what you want to AI tools like Cursor, Copilot, or GPT-4, and the model spits out the code. No need to understand syntax, architecture, or how a for-loop works. It is coding without actually coding.
The trick is going viral. Anysphere, the AI startup behind Cursor, just bagged a $900 million funding round at a $9.9 billion valuation. Founded by ex-OpenAI and Tesla engineers, the company claims to be generating a billion lines of code a day.
“The drunk uncle walks by after the wreck and gives you a roll of duct tape before asking to borrow some money to go to Vegas,”
one Redditor described vibe coding.
Some see this as a revolution, one that makes building software faster, more accessible, even fun. The startups and businesses seem to be all in. AI might sound like faster prototypes, fewer hires, and reduced development costs. Why build a whole engineering team when one "AI whisperer" can ship the project in a weekend?
Others, especially the folks who have to maintain, secure, and scale these Frankensteinian projects, are sounding the alarm. And probably quietly having heart attacks.
In the trenches on Reddit, veteran developers, DevOps engineers, and cybersecurity pros are tearing into vibe coding with the kind of raw, unfiltered honesty.
AI does not know how to code
“What’s up with the vibe coding?” asked one Redditor in a thread turned into a heated discussion. “Nothing’s up,” came the reply, “it’s all downhill lol.”
That pretty much sums up the backlash. While AI-generated code is being hailed as the future by startups, developers on the ground are watching “cheap business people” and “people who don’t understand coding, data structures, or systems design” betting it all on AI to ship entire apps. “Vibe coding is a dumb name for letting AI build everything for you,” another Redditor ranted.
Another Redditor compares vibe coding to a drunk uncle who “knows a thing or two about racing” and helps you build a race car. When things inevitably go wrong – there’s no oil in the car, the wheels aren’t tightened, and you crash it immediately – you’re left alone to fix a wreck with no tools, no manual, and no real understanding of how it came together.
“The drunk uncle walks by after the wreck and gives you a roll of duct tape before asking to borrow some money to go to Vegas,” compares Redditor, being applauded by others in a digital agora for providing the best summary of vibe coding.
At the heart of the problem with vibe coding is the fact that AI doesn’t know what it’s doing. And if a coder doesn’t either, there is double trouble. One developer put it bluntly: “The amount of bloat is crazy… it makes mistakes that you have to correct, but if you don’t know how to code, how are how is someone vibing that?”
Another problem is that AI follows its own logic while coding, which might be tricky to understand, troubleshoot, or build upon. One Redditor described a classmate’s AI-generated code as “disgusting.” The only reason it didn’t get flagged was because the professor didn’t look at it.
Even seasoned pros using AI assistants aren’t immune. A game modder said, “For every one thing that goes right, it gives me ten failures.” A data analyst admitted it “overcomplicates formulas” and sometimes gives “straight up wrong” answers.
“The more experience you have as a developer, the more AI can help. Better code, better documentation, and complete unit test coverage including property tests,”
an experienced developer said.
Another developer, after building 80% of a product with Cursor, GPT-4, and sheer vibes, watched the whole thing unravel in production: “Dropdowns didn’t dropdown. The Save button erased data. Ghost CSS from hell.” The developer shared that he hired a React developer to clean up the mess.
Behind all the jokes about Mercury retrograde crashing production and code with “bad aura,” there’s a deeper anxiety. “AI makes shit up,” one user said flatly. “You really don’t want to use AI code in prod.”
While a security professional chimed in: “I have so much anxiety around this. It’s insane what we’re willing to give up for convenience.”
Security nightmare, hackers' heaven
For security teams, the vibe isn’t chill. Redditor nailed the mood in a single sentence: “God it's gonna be a good time to be a hacker in a year or two.” Another corrected them: “It already is.”
The growing adoption of AI-generated code could be seen as a live opportunity for exploitation. “Time to poison the AI models and inject nefarious code,” one person suggested, half-joking but entirely plausible. “It would be a fascinating graduate study experiment.”
Security pros joked that it is a “good time to get into cybersecurity, because all these vibe-coded apps are going to make it to production soon enough.” “Sounds like job security to me,” a commenter shrugged. “Years of tech debt created in a few months when they realize these people have no idea what they're doing and need good devs to come in and fix their mess.”

Security teams and SREs, those tasked with keeping the backend stable and the data secure, are eyeing the vibe coding hype with growing concern. “Our Principal SRE Engineer retired and took up goose farming 23 minutes after I showed him this post,” one user deadpanned.
Others aren’t laughing: “Good luck putting this horse back in the barn. They're not going to vibe pen test their app because they don't know what they don't know.”
The core issue is that if vibe coders don’t understand the code AI wrote or the logic behind it, they can not fix bugs. “What do you think is going to happen when you vibe fix the issues? You're going to vibe create more issues,” one commenter warned.
In other words, using AI to patch up AI-generated code eventually leads to more breakage.
Some experienced coders love AI
Not everyone’s out here screaming into the void about vibe coding. For a growing number of seasoned developers, AI isn't the end of coding, but is more of their cheat code. If you know what you’re doing, AI doesn’t replace your brain; it supercharges it.
“I have 30+ years of coding experience and a MS in CS,” one veteran chimed in, “and the comments here are hysterical.” For him, tools like Cursor are game changers: “The more experience you have as a developer, the more AI can help. Better code, better documentation, and complete unit test coverage including property tests.”
“AI coding is a tremendous learning tool if you already know how to program. The issue begins where it suggests things and it just gets copy/pasted without understanding what it does,”
Reddit user said.
“I started using ChatGPT back in January 2022, and since then, my coding efficiency has increased by about five times,” another user claimed.With AI models like DeepSeek, Grok, Qwen, and ChatGPT dropping, the velocity of AI-assisted workflows is only picking up. “AI coding is a tremendous learning tool if you already know how to program,” one user noted. Still, another warned: “The issue begins where it suggests things and it just gets copy/pasted without understanding what it does.”
And while it’s cute to dunk on the vibe coders, some veterans are betting that this is just the warm-up act.
“I mean, this is the worst the technology will ever be,” one commenter warned. “I have zero doubts that AI will be able to program the entirety of some pretty big apps without much finagling or manual code fixes within the next decade.”
It’s not a question of if AI changes coding forever, it’s when. But for now, as one Redditor put it: just because you know how to swing a hammer doesn’t mean you know how to build a boat.
Your email address will not be published. Required fields are markedmarked