After two years of vibecoding, I'm back to writing by hand (atmoio.substack.com)

Article summary

The article's author claims to have spent two years 'vibecoding', but the meaning of this term is disputed in the comments. Some understand 'vibecoding' to mean writing code without looking at it, relying solely on AI assistance. Others use the term more loosely to describe using AI to write code, with or without reviewing the output. The author's experience and reasons for stopping 'vibecoding' are not detailed in the available comments.

Main themes

• Vibecoding definition
• AI-assisted coding
• Code quality and maintainability
• LLM limitations
• Programming skills and responsibility

What commenters say

• The term 'vibecoding' originally meant writing code without looking at it, but its meaning has been watered down over time to include any use of AI assistance.
• Some argue that true 'vibecoding' requires ignoring the code and only considering the end product, while others see it as a spectrum of AI involvement.
• Using AI to write code can lead to poor code quality and maintainability if not properly reviewed and managed.
• The effectiveness of AI-assisted coding tools like GitHub Copilot is disputed, with some finding them useful and others considering them underwhelming or obstructive.
• There is a need for a clearer distinction between 'vibecoding' and other forms of AI-assisted programming, such as 'agent-assisted coding'.
• The use of AI in coding raises questions about the role of human programmers and the importance of understanding the code they are working with.
• Some commentators believe that 'vibecoding' is not a viable or responsible way to write code, especially for serious projects or in professional settings.
• Others see AI-assisted coding as a valuable tool that can improve productivity and facilitate learning, but requires careful management and review to ensure quality.