TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Why Vibe Coding Makes Me Want to Throw Up

7 点作者 jerhewet大约 2 个月前

3 条评论

WuxiFingerHold大约 2 个月前
Yes, vibe coding is part of the great LLM marketing campaign. I&#x27;m old enough to be skeptical of things that are overhyped.<p>That doesn&#x27;t mean it&#x27;s a fad or generally bad. It&#x27;s a great way to implement ideas and then validate them. A dream for startups. It doesn&#x27;t matter if founders can&#x27;t even write their &quot;hello world&quot; when validating an idea.<p>The downside is as with all AI generated code comes during the lifecycle of the code: Who&#x27;s responsible? Who knows what the code does? Who&#x27;s going to maintain it beyond AI? Is it secure? Reliable? I don&#x27;t care about the next Tic Toc clone being written by AI, but how about software in the nuclear plants that is powering all the AI? Or in transportation, medical devices?
techpineapple大约 2 个月前
I certainly don&#x27;t hate vibe coding as much as this person, but having tried it a few times, the thing that really worries me is &quot;the slippery slope&quot; he describes at the end and that&#x27;s true with AI assistance in general. I think learning new programming languages probably expands your mind in ways that is somewhat immeasurable, and using AI to work with unknown languages takes away from that, and in general, I wonder what happens if you go from troubleshooting problems like 5 times a week, to once every 3 weeks because most problems fix themselves. And I do wonder if that will have a broader impact on the quality of your thinking as an engineer.
sherdil2022大约 2 个月前
I used ChatGPT to help me write some complex functionality in Rust - which I am relatively new to - and the initial efficiency was amazing. I learnt a lot and I also ran into some cases where ChatGPT wasn’t much help.<p>For me, it is a better search engine - but we still have to do hard work and due diligence - and debugging.<p>Overall, we have to embrace LLMs - and whatever tech comes next - and ensure it is used for betterment of humanity.<p>Tech has also always been used to subjugate others, create deep fakes, wage wars, kill, surveil, discriminate and worse - and AI tech is going to be no different. We have to ensure that guardrails are put and the tech is not abused.