大型语言模型最大化论者的不安全布道
The insecure evangelism of LLM maximalists

原始链接: https://lewiscampbell.tech/blog/260114.html

## LLM 生产力:一个怀疑论者的观点 这位作者是一位资深开发者,他对当前使用 LLM 的“代理编码”状态表示失望。虽然发现 LLM 在研究和简单代码片段等任务中很有用,但他们发现“提示驱动开发”速度慢、容易出错,并且需要过多的监督——最终感觉效率反而降低。 核心沮丧源于围绕 LLM 驱动开发的积极宣传。作者认为将怀疑论等同于害怕过时是一种令人费解的暗示,因为他们*希望*未来能够专注于设计和规范。 他们开始怀疑这种热情的推广并非真正的生产力提升,而是那些感到自己编码技能价值降低的人的一种不安全感的投射。作者仍然愿意改变想法,承认自己存在技能差距,但挑战 LLM 倡导者考虑这种技术可能并非普遍的生产力助推器,并且可能不适合所有开发者或所有任务。

## 黑客新闻讨论:LLM 爱好者的“不安全宣传” 最近黑客新闻上出现了一场讨论,中心是技术社区内对大型语言模型(LLM)的热烈推广,以及支持者们表现出的一种防御姿态。最初的帖子质疑,为什么一些LLM倡导者会在他人没有分享他们积极体验时表现出敌意。 许多评论者同意,LLM在不熟悉的语言或领域中执行任务表现出色,并且可以显著提高某些人的生产力。然而,经验丰富的程序员经常发现LLM生成的代码需要大量修改,并且可能引入细微的错误或低效之处。人们对依赖具有潜在长期责任的“廉价”代码表示担忧。 一个关键点是,LLM是强大的工具,但并非普遍优越。它们的价值取决于任务和用户的技能水平。有些人认为它们对样板代码或探索很有帮助,而另一些人则担心它们会降低代码的整体质量并削弱基本的编程技能。 讨论还涉及到了互联网上普遍存在的强烈、分裂的观点倾向,以及不安全感可能驱动热情倡导 *和* 怀疑论。几位评论员注意到,这与过去的科技辩论(Python vs. C等)相似,并告诫人们不要轻易否定不同的观点。
相关文章

原文
1/14/2026

I am an LLM productivity skeptic.

I find LLMs useful as a sort of digital clerk - searching the web for me, finding documentation, looking up algorithms. I even find them useful in a limited coding capacity; with a small context and clear guidelines.

But doing "prompt-driven development" or "vibe coding" with an Agentic LLM was an incredibly disapointing experience for me. It required an immense amount of baby sitting, for small code changes, made slowly, which were often wrong. All the while I sat there feeling dumber and dumber, as my tokens drained away.

Of course that was my experience, and my preference. I genuinely don't mind if other people vibe code. Go for it! I do not deny this kind of coding is enabling a lot of people - who aren't experienced devs - to create things they would never otherwise be able to create. (Also, sometimes they pay me to clean them up afterwards, which is nice.)

But that is not enough for the vocal proponents. It's the future! You'll be left behind! Software has changed forever! And then, inevitably, comes the character evaluation, which goes something like this:

You - a senior dev - are resisting this change due to deeply held psychological fears of being made irrelevant and/or having to learn new things. You are stuck in your ways and unwilling to change them because you are afraid.

This has always baffled me, because quite frankly I like the idea of agentic coding. I often feel the actual implementation is a bottleneck to the things I want to create. I would love it if I could just sit around making specs (yes I am a programmer who enjoys this) and have little machines implement it for me perfectly. It's a wonderful fantasy world, and I wish I could inhabit it. That's why I was so disappointed.

And it made me think - why are these people so insistent, and hostile? Why can't they live and let live? Why do they need to convince the rest of us? And to be honest, I am developing my own character evaluation. It's not very charitable, but it is making a lot of sense to me:

You tried agentic coding. You realised it was better at programming than you are. You see a lot of accomplished, prominent developers claiming they are more productive without it. Could they just be that much better at programming than I am? No! They are just threatened. They are the ones who are insecure! I'm a great developer!

It's projection. Their evangelism is born of insecurity.

I am still willing to admit I am wrong. That I'm not holding the agents properly. That doing this is it's own skill and I have not spent enough time with it. I have changed my mind on tech before, and I'm sure I will do so again.

LLM evangelists - are you willing to admit that you just might not be that good at programming computers? Maybe you once were. Maybe you never were.



I'm available for hire.
联系我们 contact @ memedata.com