I was lurking around dev.to because I was bored and I saw a post made by @sylwia-lask on the topic:
What stood out to me was the "Vibe Coding Trap" in particular. We all have Vibe Code something at least once and to be honest, it is quite useful when it comes to prototyping, building template projects, etc. However, it can be damaging if we are not careful on how much we rely on AI.
and spoiler alert (from the title), a lot can go wrong. Let me explain.
You use AI excessively? You should Caallllmmm dooowwwnnnn.
AI is everywhere. There is no doubt about it and we all use it in our workflows. The problem is because it is hard to ignore, we are prone to use it...maybe too much.
As a person searching for work, I am realizing that I need to tone down the amount of the use of AI when it comes to contributing to Open Source and my personal projects. Here is the main reason.
Can you explain what you built in an Interview?
You can build a very polished project using AI. It may sound impressive on paper and sure, you can use AI to provide you a summary for others to understand.
But here is the thing.
Once you start getting into an interview and try to explain what you built in detail, your ankles will be broken to the point where Bumgumi's half-baked domain expansion is more impressive than your explanation to the project that you vibe coded.
As my friend, John, states: "It's like saying you did a back flip in Chinese, but you don't know Chinese".
It is important that you own your code by not relying on AI too much to the point where you have no idea on what you built. You can for sure use AI when you are creating a simple template, but it is not prefer to use AI just to "one-shot" a project and claiming that you built all that.
There are exceptions. For example, if you are building a project using AI and showcasing it for fun, than it is fine. However, if you are building a Vibe Coded Project and listing it on your resume, good luck explaining in detail about how and why you built it that way.
Overall, use AI as a tool and ensuring that you are still learning the same way as if you are using StackOverFlow back in the days.
Speaking of, it reminds me of a post by @greggyb:
He mentions about the workflow it used to be for developers where:
For most of my career, the standard loop was: write some code → hit a wall → Google it → find a StackOverflow answer → copy, paste, tweak. That ran from the late 90s to roughly 2022. Then everything changed.
What stood out to me was "What I do know is this: there has never been a better time to invite someone new into programming. The barrier that built up over decades is coming down fast.".
Even though it is great to try something new, we have to acknowledge if it will affect us in a good or bad way. For this case, I believe that AI is a great addition to have. However, it is always good to balance it out.
I always have this statement in mind when doing anything "Anything too much or too little is not good for you". If you use too much AI, you are not learning...you are relying. However, using AI too little depends on situations such as if you decide to write manual test data instead of asking AI to create mock data for you. In other words, being on that side of the coin can be time consuming for you if you don't use AI to speed up the process.
Again, use AI efficiently and ensure you are still learning and growing.
What do you guys think about this? Let me know your thoughts!
Top comments (2)
The real danger of vibe coding isn't buggy code.
It's when the interviewer asks, "Why did you do it this way?"
And your only honest answer is:
"Because ChatGPT said 'Excellent choice!'" 😂
I love the interview example! Great reminder for begginers and job seekers.