Vibecoding rattles programming world

Computer Science instructor Ivann Grande’s daughter’s birthday was a couple days away.
He’d planned to work on his present during spring break, but Grande had been busy writing third-quarter comments and grading eighth grade coding projects. And his daughter’s special day had come quicker than he thought.
Grande had a gift in mind: a ‘Harry Potter’-themed escape room, built entirely with code. He’s built countless websites throughout his career, and he wanted to use talents and experience to create something memorable for his daughter.
But if he’s learned one thing about coding, it’s that programming requires long nights and dedication. He needed time. And he was running out of it.
So he pumped preliminary questions into ChatGPT to give him the base structure of his virtual treasure hunt. It was quick and easy. Copy and paste. Without giving it a second thought.
If the code had an error, no problem. Grande would just reword his question and try it again. Of course, Grande could do this all on his own, and he still coded pieces of his gift throughout the process. But what usually took half of a week became just a day with the help of artifical intelligence.
And Grande’s daughter loved her birthday present.
Popularized in February 2025, ‘vibecoding’ is a form of AI use: pumping and dumping entrees from AI to assist with hobbies or goals. Vibecoding is a term that has exponentially grown in use, surprisingly by some who are outside of the coding community — they can program without actually knowing how to program.
“It’s a means of generating code for the sake of generating it,” Grande said. “Even programmers will vibecode. They know how to do it, but they don’t want to spend the time, they just need a quick answer.”
In Grande’s dilemma with his birthday gift, he utilized a form of vibecoding. He was looking for a quick and streamlined solution and ChatGPT responded instantly to his demands. In this case, it was harmless. But there’s a fine line between coding for personal reasons or taking credit publically for artificially-done work.
“When people saw my gift, they asked me, ‘Did you program all of this?’” Grande said. “Of course, I had to say I didn’t program it all. You can’t take credit for something you didn’t do from scratch.”
In general, Grande believes vibecoding is a creative, casual way to use AI. However, when using artificial intelligence to code starts leaking into one’s profession, then it’s unacceptable.
“For those that just plug and play, especially if it’s just a one-time thing, it’s not a big deal,” Grande said. “It’s when it starts becoming a part of your profession. When you’re getting paid to do work. That’s a problem.”
Junior Eric Zhang has been competitively coding for several years, and since the introduction of AI two years ago, his productivity has skyrocketed. While he already knows the fundamentals of several different coding languages, he believes that AI is helpful in all situations and has become very mainstream.
“There’s a very popular tool called GitHub Copilot and it’s based off of the GPT technologies that we’ve seen come out,” Zhang said. “Ever since I’ve started using that tool, I’ve been able to triple or quadruple the efficiency of which I can write code, because there’s a lot of simple things that will automatically fill it out for you.”
In addition to assisting him in filling out basic code or searching for keywords, Zhang believes that AIs can be helpful in acting as an assistant that one can communicate with while coding, with the AI able to fix items that the coder leaves comments about.
While AI has made it easier to code, Zhang believes that it does not allow true beginner coders to start coding as its suggestions may be far too complicated for the beginners to understand.
“It’s almost like learning how to Google and knowing which keywords will give the correct output,” Zhang said. “So, inherently, I feel like that is sort of a skill that people have developed. I think for a beginner, it wouldn’t really help because they would use code that they don’t know how it works.”
This newest wave of developing AIs have produced a lot of fear among employees about their job security. However, Zhang believes that this point is far from occurring and AI has a long way to develop before reaching this point.
“(AIs) just look at all of the data that’s available and try to predict what happens next,” Zhang said. “However, for these models to improve, they need more data, so the curve that they improve upon is logarithmic. You need exponentially more data in order to improve. If we really wanted to improve by one or two percent, we would need tens of times more data than humanity has ever generated.”