開發(fā)者們說(shuō)AI編碼工具確實(shí)有效，而這恰恰是他們擔(dān)憂的地方

Ars 與幾位軟件開發(fā)人員就人工智能問(wèn)題進(jìn)行了交談，發(fā)現(xiàn)他們熱情中夾雜著不安。

本杰·愛(ài)德華茲

圖片來(lái)源：Aurich Lawson | Getty Images

過(guò)去兩年，軟件開發(fā)人員見(jiàn)證了人工智能編碼工具的演變，它們從高級(jí)自動(dòng)補(bǔ)全功能發(fā)展到如今能夠根據(jù)文本提示構(gòu)建整個(gè)應(yīng)用程序。像 Anthropic 的 Claude Code 和 OpenAI 的 Codex 這樣的工具現(xiàn)在可以連續(xù)工作數(shù)小時(shí)，編寫代碼、運(yùn)行測(cè)試，并在人工監(jiān)督下修復(fù)漏洞。OpenAI表示，它現(xiàn)在使用 Codex 來(lái)構(gòu)建 Codex 本身，并且該公司最近公布了該工具底層工作原理的技術(shù)細(xì)節(jié)。這引發(fā)了許多人的疑問(wèn)：這僅僅是人工智能行業(yè)的又一次炒作，還是這次真的有所不同？

為了探究這個(gè)問(wèn)題，Ars 聯(lián)系了幾位 Bluesky 的專業(yè)開發(fā)者，詢問(wèn)他們對(duì)這些工具的實(shí)際使用感受。調(diào)查結(jié)果顯示，大多數(shù)開發(fā)者都認(rèn)同這項(xiàng)技術(shù)有效，但對(duì)于這是否完全是好事，他們的看法仍存在分歧。雖然樣本量較小，且參與者均為自愿報(bào)名，但他們的觀點(diǎn)對(duì)于該領(lǐng)域的從業(yè)者而言仍然具有參考價(jià)值。

從事銷售點(diǎn)系統(tǒng)開發(fā)的戴維·哈格蒂 (David Hagerty) 一開始就告訴 Ars Technica，他對(duì)人工智能的營(yíng)銷持懷疑態(tài)度。“所有人工智能公司都在大肆宣傳其功能，”他說(shuō)。“別誤會(huì)我的意思——人工智能確實(shí)具有革命性，并將產(chǎn)生巨大的影響，但別指望它們能寫出下一部偉大的美國(guó)小說(shuō)之類的。它們不是這么運(yùn)作的。”

曾為 Linux 內(nèi)核做出過(guò)大量貢獻(xiàn)的軟件工程師 Roland Dreier 告訴 Ars Technica，他承認(rèn)人工智能領(lǐng)域存在炒作，但他一直密切關(guān)注著該領(lǐng)域的發(fā)展。“這聽起來(lái)像是夸大其詞，但目前最先進(jìn)的智能體確實(shí)非常出色，”他說(shuō)。Dreier 描述了過(guò)去六個(gè)月的“飛躍式變化”，尤其是在 Anthropic 發(fā)布Claude Opus 4.5之后。他以前使用人工智能只是為了自動(dòng)補(bǔ)全和偶爾提問(wèn)，而現(xiàn)在他期望告訴智能體“這個(gè)測(cè)試失敗了，幫我調(diào)試并修復(fù)它”，然后它就能正常工作。他估計(jì)，對(duì)于構(gòu)建一個(gè)使用 Terraform 部署配置的 Rust 后端服務(wù)和一個(gè)Svelte前端這樣的復(fù)雜任務(wù)，速度可以提升 10 倍。

目前開發(fā)者們最關(guān)心的一個(gè)問(wèn)題是，所謂的“語(yǔ)法編程”（即手動(dòng)使用既定編程語(yǔ)言的語(yǔ)法編寫代碼，而不是用英語(yǔ)與人工智能代理交流）是否會(huì)在不久的將來(lái)消失，因?yàn)槿斯ぶ悄芫幋a代理會(huì)替他們處理語(yǔ)法。Dreier認(rèn)為，對(duì)于許多任務(wù)而言，語(yǔ)法編程已經(jīng)基本過(guò)時(shí)了。“我仍然需要能夠閱讀和審查代碼，”他說(shuō)，“但我實(shí)際敲擊的Rust或其他我正在使用的語(yǔ)言代碼已經(jīng)很少了。”

當(dāng)被問(wèn)及開發(fā)者是否還會(huì)回歸手動(dòng)編寫語(yǔ)法代碼時(shí)，經(jīng)常在社交媒體上分享人工智能相關(guān)內(nèi)容并構(gòu)建自主代理的開發(fā)者蒂姆·凱洛格直言不諱地表示：“已經(jīng)結(jié)束了。人工智能編碼工具可以輕松處理表面細(xì)節(jié)。” 誠(chéng)然，凱洛格代表了那些已經(jīng)完全接受智能體人工智能的開發(fā)者，他們現(xiàn)在的工作重心是指導(dǎo)人工智能模型，而不是編寫代碼。他說(shuō)，他現(xiàn)在“構(gòu)建三次所需的時(shí)間比手動(dòng)構(gòu)建還要短”，最終得到的架構(gòu)也更加簡(jiǎn)潔。

一位在定價(jià)管理SaaS公司工作的軟件架構(gòu)師（因公司溝通政策要求匿名）告訴Ars，人工智能工具徹底改變了他從事傳統(tǒng)編碼30年后的工作方式。“如果用傳統(tǒng)方法，我可能要花一年時(shí)間才能完成一項(xiàng)工作功能，而現(xiàn)在我只需兩周左右就能交付，”他說(shuō)。對(duì)于業(yè)余項(xiàng)目，他表示現(xiàn)在“只需一個(gè)小時(shí)就能搭建一個(gè)原型，然后判斷它是否值得繼續(xù)開發(fā)或放棄。”

德雷爾表示，降低工作量讓他得以完成一些擱置多年的項(xiàng)目：“‘重寫那個(gè)從相機(jī)SD卡復(fù)制照片的蹩腳shell腳本’這件事，我待辦事項(xiàng)清單上已經(jīng)躺了好幾年了。” 代碼代理最終降低了入門門檻，低到他只需幾個(gè)小時(shí)就能用Rust語(yǔ)言編寫一個(gè)完整的、帶有文本用戶界面和單元測(cè)試的發(fā)布版軟件包。“這沒(méi)什么了不起的，但我絕對(duì)沒(méi)有精力手動(dòng)敲出所有這些代碼，”他告訴Ars Technica。

關(guān)于氛圍編碼和技術(shù)債務(wù)

并非所有人都像德雷爾那樣熱情。人們對(duì)人工智能編碼代理積累技術(shù)債務(wù)的擔(dān)憂，即在開發(fā)初期做出糟糕的設(shè)計(jì)選擇，隨著時(shí)間的推移，這些選擇會(huì)像滾雪球一樣演變成更嚴(yán)重的問(wèn)題，這種擔(dān)憂在2025年初“直覺(jué)式編碼”的首次討論出現(xiàn)后不久便出現(xiàn)了。前OpenAI研究員安德烈·卡帕西創(chuàng)造了這個(gè)術(shù)語(yǔ)，用來(lái)描述在不完全理解最終代碼的情況下，通過(guò)與人工智能對(duì)話進(jìn)行編程的做法。許多人認(rèn)為這是人工智能編碼代理的一個(gè)明顯風(fēng)險(xiǎn)。

微軟高級(jí)軟件開發(fā)工程師 Darren Mart 自 2006 年起就在該公司工作，他向 Ars 表達(dá)了類似的擔(dān)憂。Mart 強(qiáng)調(diào)，他僅以個(gè)人身份而非微軟的名義發(fā)言。他最近在終端中使用 Claude 構(gòu)建了一個(gè)與 Azure Functions 集成的 Next.js 應(yīng)用程序。他說(shuō)，該 AI 模型“根據(jù)我的規(guī)范成功構(gòu)建了大約 95% 的內(nèi)容”。但他仍然保持謹(jǐn)慎。“我只放心地使用它們來(lái)完成我完全理解的任務(wù)，”Mart 說(shuō)，“否則，我無(wú)法知道自己是否會(huì)被引入歧途，并讓自己（和/或我的團(tuán)隊(duì)）背負(fù)沉重的未來(lái)債務(wù)。”

一位從事房地產(chǎn)分析的數(shù)據(jù)科學(xué)家（由于工作性質(zhì)敏感，要求匿名）表示，出于類似的原因，他嚴(yán)格控制著人工智能的使用。他使用 GitHub Copilot 進(jìn)行逐行代碼補(bǔ)全，發(fā)現(xiàn)大約 75% 的情況下都很有用，但他將智能體的功能限制在特定的使用場(chǎng)景：例如，對(duì)遺留代碼進(jìn)行語(yǔ)言轉(zhuǎn)換、使用明確的只讀指令進(jìn)行調(diào)試，以及禁止直接編輯的標(biāo)準(zhǔn)化任務(wù)。“因?yàn)槲覉?jiān)持?jǐn)?shù)據(jù)至上，所以我非常厭惡數(shù)據(jù)被錯(cuò)誤篡改的風(fēng)險(xiǎn)，”他說(shuō)道，“而且，下一行和當(dāng)前行的補(bǔ)全錯(cuò)誤率太高，我無(wú)法放任LLM自由發(fā)揮。”

說(shuō)到自由發(fā)揮，每天都使用 Cursor 的耐克后端工程師 Brian Westby 告訴 Ars，他認(rèn)為這些工具“好壞參半”。他說(shuō)，它們可以縮短解決明確問(wèn)題的時(shí)間，但“如果我給它太多的自主權(quán)，就會(huì)出現(xiàn)太多意想不到的問(wèn)題”。

傳統(tǒng)代碼生命線與企業(yè)人工智能差距

對(duì)于使用老舊系統(tǒng)的開發(fā)人員來(lái)說(shuō)，人工智能工具就像是翻譯和考古學(xué)家的結(jié)合體。First American Financial 的一名工程師 Nate Hashem 告訴 Ars Technica，他每天都在更新老舊的代碼庫(kù)，因?yàn)椤白畛醯拈_發(fā)人員已經(jīng)不在了，而且文檔往往也不清楚代碼當(dāng)初為什么那樣編寫。” Hashem 說(shuō)，這一點(diǎn)至關(guān)重要，因?yàn)橐郧啊案緵](méi)有時(shí)間去改進(jìn)這些”。“公司不會(huì)給你兩到四周的時(shí)間來(lái)弄清楚所有東西到底是怎么運(yùn)作的。”

在他看來(lái)，在這種高壓、資源相對(duì)匱乏的環(huán)境中，人工智能通過(guò)加快識(shí)別過(guò)時(shí)代碼的刪除位置和方式、診斷錯(cuò)誤以及最終實(shí)現(xiàn)代碼庫(kù)現(xiàn)代化的過(guò)程，使這項(xiàng)工作“變得更加愉快”。

哈希姆還提出了一個(gè)理論，解釋為什么大型企業(yè)內(nèi)部的人工智能應(yīng)用與社交媒體上的應(yīng)用截然不同。他說(shuō)，高管們要求公司“面向人工智能”，但部署包含專有數(shù)據(jù)的人工智能工具可能需要數(shù)月的法律審查。與此同時(shí)，微軟和谷歌集成到Gmail和Excel等產(chǎn)品中的人工智能功能（這些工具實(shí)際上惠及大多數(shù)員工）往往運(yùn)行在功能較為有限的人工智能模型上。哈希姆說(shuō)：“管理層要求普通白領(lǐng)員工使用人工智能，但卻只給他們提供功能糟糕的人工智能工具，因?yàn)檎嬲糜玫墓ぞ咝枰罅康某杀竞头蓞f(xié)議。”

說(shuō)到管理，這些新的人工智能編碼工具對(duì)軟件開發(fā)工作意味著什么，這個(gè)問(wèn)題引發(fā)了各種各樣的反應(yīng)。它會(huì)威脅到任何人的工作嗎？凱洛格熱情擁抱了智能編碼，他直言不諱地表示：“是的，影響巨大。今天是編寫代碼，然后是架構(gòu)，再然后是層層產(chǎn)品管理。那些無(wú)法適應(yīng)更高層次工作的人將失去他們的工作。”

德雷爾雖然對(duì)自己的地位感到安心，但卻擔(dān)憂新人的發(fā)展道路。“教育和培訓(xùn)必須做出改變，才能讓初級(jí)開發(fā)人員獲得所需的經(jīng)驗(yàn)和判斷力，”他說(shuō)，“讓他們像我當(dāng)初那樣實(shí)現(xiàn)系統(tǒng)的一小部分功能，簡(jiǎn)直是浪費(fèi)時(shí)間。”

Hagerty 用經(jīng)濟(jì)術(shù)語(yǔ)來(lái)說(shuō)：“如果我能用像 Sonnet 4.5 這樣的模型以低于最低工資的價(jià)格獲得初級(jí)水平的代碼，那么初級(jí)職位將越來(lái)越難招到人。”

微軟工程師馬特則以更個(gè)人化的方式表達(dá)了這種看法。他說(shuō)，軟件開發(fā)的角色“正在從創(chuàng)造/構(gòu)建突然轉(zhuǎn)向監(jiān)督”，“雖然有些人可能歡迎這種轉(zhuǎn)變，但其他人肯定不歡迎。我堅(jiān)定地屬于后者。”

即使宏觀層面仍存在諸多不確定性，有些人出于個(gè)人原因，仍然非常享受使用這些工具，而忽略了其更深遠(yuǎn)的影響。“我非常喜歡使用人工智能編碼工具，”一位在定價(jià)管理SaaS公司工作的匿名軟件架構(gòu)師告訴Ars Technica。“我成年后從事傳統(tǒng)編碼工作（大約30年），現(xiàn)在我比以前做傳統(tǒng)編碼時(shí)更有樂(lè)趣。”

本杰·愛(ài)德華茲

資深人工智能記者

本杰·愛(ài)德華茲是 Ars Technica 的高級(jí)人工智能記者，也是該網(wǎng)站人工智能專區(qū)（成立于 2022 年）的創(chuàng)始人。他同時(shí)也是一位擁有近二十年經(jīng)驗(yàn)的科技史學(xué)家。閑暇時(shí)，他喜歡創(chuàng)作和錄制音樂(lè)、收藏老式電腦以及親近自然。他現(xiàn)居北卡羅來(lái)納州羅利市。

Developers say AI coding tools work—and that’s precisely what worries them

Ars spoke to several software devs about AI and found enthusiasm tempered by unease.

BENJ EDWARDS – JAN 30, 2026 2:04 PM

Software developers have spent the past two years watching AI coding tools evolve from advanced autocomplete into something that can, in some cases, build entire applications from a text prompt. Tools like Anthropic’s Claude Code and OpenAI’s Codex can now work on software projects for hours at a time, writing code, running tests, and, with human supervision, fixing bugs. OpenAI says it now uses Codex to build Codex itself, and the company recently published technical details about how the tool works under the hood. It has caused many to wonder: Is this just more AI industry hype, or are things actually different this time?

To find out, Ars reached out to several professional developers on Bluesky to ask how they feel about these tools in practice, and the responses revealed a workforce that largely agrees the technology works, but remains divided on whether that’s entirely good news. It’s a small sample size that was self-selected by those who wanted to participate, but their views are still instructive as working professionals in the space.

ADVERTISING

David Hagerty, a developer who works on point-of-sale systems, told Ars Technica up front that he is skeptical of the marketing. “All of the AI companies are hyping up the capabilities so much,” he said. “Don’t get me wrong—LLMs are revolutionary and will have an immense impact, but don’t expect them to ever write the next great American novel or anything. It’s not how they work.”

Roland Dreier, a software engineer who has contributed extensively to the Linux kernel in the past, told Ars Technica that he acknowledges the presence of hype but has watched the progression of the AI space closely. “It sounds like implausible hype, but state-of-the-art agents are just staggeringly good right now,” he said. Dreier described a “step-change” in the past six months, particularly after Anthropic released Claude Opus 4.5. Where he once used AI for autocomplete and asking the occasional question, he now expects to tell an agent “this test is failing, debug it and fix it for me” and have it work. He estimated a 10x speed improvement for complex tasks like building a Rust backend service with Terraform deployment configuration and a Sveltefrontend.

A huge question on developers’ minds right now is whether what you might call “syntax programming,” that is, the act of manually writing code in the syntax of an established programming language (as opposed to conversing with an AI agent in English), will become extinct in the near future due to AI coding agents handling the syntax for them. Dreier believes syntax programming is largely finished for many tasks. “I still need to be able to read and review code,” he said, “but very little of my typing is actual Rust or whatever language I’m working in.”

When asked if developers will ever return to manual syntax coding, Tim Kellogg, a developer who actively posts about AI on social media and builds autonomous agents, was blunt: “It’s over. AI coding tools easily take care of the surface level of detail.” Admittedly, Kellogg represents developers who have fully embraced agentic AI and now spend their days directing AI models rather than typing code. He said he can now “build, then rebuild 3 times in less time than it would have taken to build manually,” and ends up with cleaner architecture as a result.

One software architect at a pricing management SaaS company, who asked to remain anonymous due to company communications policies, told Ars that AI tools have transformed his work after 30 years of traditional coding. “I was able to deliver a feature at work in about 2 weeks that probably would have taken us a year if we did it the traditional way,” he said. And for side projects, he said he can now “spin up a prototype in like an hour and figure out if it’s worth taking further or abandoning.”

Dreier said the lowered effort has unlocked projects he’d put off for years: “I’ve had ‘rewrite that janky shell script for copying photos off a camera SD card’ on my to-do list for literal years.” Coding agents finally lowered the barrier to entry, so to speak, low enough that he spent a few hours building a full released package with a text UI, written in Rust with unit tests. “Nothing profound there, but I never would have had the energy to type all that code out by hand,” he told Ars.

Of vibe coding and technical debt

Not everyone shares the same enthusiasm as Dreier. Concerns about AI coding agents building up technical debt, that is, making poor design choices early in a development process that snowball into worse problems over time, originated soon after the first debates around “vibe coding” emerged in early 2025. Former OpenAI researcher Andrej Karpathy coined the term to describe programming by conversing with AI without fully understanding the resulting code, which many see as a clear hazard of AI coding agents.

Darren Mart, a senior software development engineer at Microsoft who has worked there since 2006, shared similar concerns with Ars. Mart, who emphasizes he is speaking in a personal capacity and not on behalf of Microsoft, recently used Claude in a terminal to build a Next.js application integrating with Azure Functions. The AI model “successfully built roughly 95% of it according to my spec,” he said. Yet he remains cautious. “I’m only comfortable using them for completing tasks that I already fully understand,” Mart said, “otherwise there’s no way to know if I’m being led down a perilous path and setting myself (and/or my team) up for a mountain of future debt.”

A data scientist working in real estate analytics, who asked to remain anonymous due to the sensitive nature of his work, described keeping AI on a very short leash for similar reasons. He uses GitHub Copilot for line-by-line completions, which he finds useful about 75 percent of the time, but restricts agentic features to narrow use cases: language conversion for legacy code, debugging with explicit read-only instructions, and standardization tasks where he forbids direct edits. “Since I am data-first, I’m extremely risk averse to bad manipulation of the data,” he said, “and the next and current line completions are way too often too wrong for me to let the LLMs have freer rein.”

Speaking of free rein, Nike backend engineer Brian Westby, who uses Cursor daily, told Ars that he sees the tools as “50/50 good/bad.” They cut down time on well-defined problems, he said, but “hallucinations are still too prevalent if I give it too much room to work.”

The legacy code lifeline and the enterprise AI gap

For developers working with older systems, AI tools have become something like a translator and an archaeologist rolled into one. Nate Hashem, a staff engineer at First American Financial, told Ars Technica that he spends his days updating older codebases where “the original developers are gone and documentation is often unclear on why the code was written the way it was.” That’s important because previously “there used to be no bandwidth to improve any of this,” Hashem said. “The business was not going to give you 2-4 weeks to figure out how everything actually works.”

In that high-pressure, relatively low-resource environment, AI has made the job “a lot more pleasant,” in his words, by speeding up the process of identifying where and how obsolete code can be deleted, diagnosing errors, and ultimately modernizing the codebase.

Hashem also offered a theory about why AI adoption looks so different inside large corporations than it does on social media. Executives demand their companies become “AI oriented,” he said, but the logistics of deploying AI tools with proprietary data can take months of legal review. Meanwhile, the AI features that Microsoft and Google bolt onto products like Gmail and Excel, the tools that actually reach most workers, tend to run on more limited AI models. “That modal white-collar employee is being told by management to use AI,” Hashem said, “but is given crappy AI tools because the good tools require a lot of overhead in cost and legal agreements.”

Speaking of management, the question of what these new AI coding tools mean for software development jobs drew a range of responses. Does it threaten anyone’s job? Kellogg, who has embraced agentic coding enthusiastically, was blunt: “Yes, massively so. Today it’s the act of writing code, then it’ll be architecture, then it’ll be tiers of product management. Those who can’t adapt to operate at a higher level won’t keep their jobs.”

Dreier, while feeling secure in his own position, worried about the path for newcomers. “There are going to have to be changes to education and training to get junior developers the experience and judgment they need,” he said, “when it’s just a waste to make them implement small pieces of a system like I came up doing.”

Hagerty put it in economic terms: “It’s going to get harder for junior-level positions to get filled when I can get junior-quality code for less than minimum wage using a model like Sonnet 4.5.”

Mart, the Microsoft engineer, put it more personally. The software development role is “abruptly pivoting from creation/construction to supervision,” he said, “and while some may welcome that pivot, others certainly do not. I’m firmly in the latter category.”

Even with this ongoing uncertainty on a macro level, some people are really enjoying the tools for personal reasons, regardless of larger implications. “I absolutely love using AI coding tools,” the anonymous software architect at a pricing management SaaS company told Ars. “I did traditional coding for my entire adult life (about 30 years) and I have way more fun now than I ever did doing traditional coding.”

BENJ EDWARDS

SENIOR AI REPORTER

Benj Edwards is Ars Technica's Senior AI Reporter and founder of the site's dedicated AI beat in 2022. He's also a tech historian with almost two decades of experience. In his free time, he writes and records music, collects vintage computers, and enjoys nature. He lives in Raleigh, NC.