Vibe Coding의 열풍, 도대체 왜?
What's the hype with Vibe Coding?
핵심 요약
AI 코딩 도구의 실효성에 의문을 제기하며, 반복적인 수정 작업으로 인한 비효율성을 지적하는 글.
- AI 코딩 도구 — 실무 적용 시 잦은 오류와 수정 작업으로 인한 비효율성 호소
- 작업 방식 논쟁 — 단순히 코드를 짜는 것이 아니라 오류를 빠르게 수정하는 속도의 문제
- 사용자 숙련도 — AI를 제대로 활용하지 못하는 것은 결국 사용자의 역량 문제라는 의견
- 자동화의 비유 — 비행기 자동 조종 장치처럼 완벽하지 않아도 생산성을 높이는 도구로 인식
Perhaps I have not learned how to use AI effectively for coding, but my experience has been mixed. I have used AI tools since they became widely available, and my most productive period was during the early reasoning-model era, around o3-mini-high. At the time, my tasks may have been more narrowly scoped.
More recently, however, even when using advanced tools like Codex and high-capability models with detailed instructions and documentation, I still encounter obvious mistakes. Even when I ask these tools to create simple functions, the output can contain errors, which makes the process feel inefficient. By the time I correct the result through repeated prompting, I often could have completed the work myself.
AI can be useful for simple prototypes, basic web pages, and polished writing, but I do not yet see the same value in coding work that many others describe. Perhaps my prompts are not explicit enough, but from my perspective, AI coding tools still require too much supervision to be consistently productive.
