← 回總覽

AI 生成代码质量平庸的根源

📅 2026-03-12 16:26 Gergely Orosz 人工智能 3 分鐘 3135 字 評分: 81
AI 编程 代码质量 软件工程 开发者技能 LLM 训练数据
📌 一句话摘要 Gergely Orosz 认为,AI 编程输出之所以平庸,是因为公开代码普遍质量一般,且开发者自身缺乏编写高质量代码的能力,而非工具配置问题。 📝 详细摘要 在这条推文中,Gergely Orosz 挑战了“AI 生成代码差主要是因为工具配置不当”这一普遍观点。他指出,由于 AI 模型是在现有的公开代码上训练的——而这些代码大多质量平庸——因此输出结果自然也反映了这种质量。此外,他强调核心问题在于开发者:甚至在 AI 出现之前,许多人就不具备编写高质量代码的技能。这表明在 AI 辅助开发中,人的因素以及训练数据的“垃圾进,垃圾出”(Garbage In, Garbage
Skip to main content ![Image 1: LogoBestBlogs](https://www.bestblogs.dev/ "BestBlogs.dev")Toggle navigation menu Toggle navigation menuArticlesPodcastsVideosTweetsSourcesNewsletters

⌘K

Change language Switch ThemeSign In

Narrow Mode

The Root Cause of Mediocre AI-Generated Code ============================================

The Root Cause of Mediocre AI-Generated Code ============================================ ![Image 2: Gergely Orosz](https://www.bestblogs.dev/en/tweets?sourceId=SOURCE_6b94cc22) ### Gergely Orosz

@GergelyOrosz

From my inbox

Wrong:

Most AI coding output is not mediocre because the tool was not well co figured:

It’s mediocre because most code out in the open IS mediocre, and (most importantly) most devs don’t know how to write high quality code. And this was true even before AI!!

!Image 3: Tweet image

Mar 12, 2026, 8:26 AM View on X

10 Replies

7 Retweets

76 Likes

3,349 Views ![Image 4: Gergely Orosz](https://www.bestblogs.dev/en/tweets?sourceid=6b94cc22) Gergely Orosz @GergelyOrosz

One Sentence Summary

Gergely Orosz argues that mediocre AI coding output stems from the prevalence of mediocre public code and developers' lack of high-quality coding skills, rather than tool configuration.

Summary

In this tweet, Gergely Orosz challenges the common belief that poor AI-generated code is primarily a result of bad tool configuration. He posits that since AI models are trained on existing public code—which is largely mediocre—the output naturally reflects that quality. Furthermore, he emphasizes that the fundamental issue lies with developers who lacked high-quality coding skills even before the advent of AI, suggesting that the human element and the 'Garbage In, Garbage Out' nature of training data are the primary bottlenecks in AI-assisted development.

AI Score

81

Influence Score 17

Published At Today

Language

English

Tags

AI Coding

Code Quality

Software Engineering

Developer Skills

LLM Training Data HomeArticlesPodcastsVideosTweets

The Root Cause of Mediocre AI-Generated Code | BestBlogs.dev ===============

查看原文 → 發佈: 2026-03-12 16:26:11 收錄: 2026-03-12 18:00:42

🤖 問 AI

針對這篇文章提問,AI 會根據文章內容回答。按 Ctrl+Enter 送出。