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。

Ask HN: AI Replacing Engineers – Firsthand Stories?

48 点作者 ludovicianul14 天前
I keep reading about many companies moving to a model where they stop recruiting and replace human engineers with AI. Is anyone part of such a company that wants to share some firsthand experience? What changed significantly in your workflow? On top of the "clasic" GitHub Copilot and/or Cursor, any other tools/agents/automated workflows that are used to compensate for additional human effort? Are you 10x more efficient? Is the effort similar as before, but distributed in other areas?

29 条评论

aristofun14 天前
I see much more histeria and false but extremely high hopes, than real deal from where I sit (deceloper at faanglike high tec company).<p>Looks like the higher the management, the farther away from real engineering work — the more excitement there is and the less common sense and real understanding of how developers and llms work.<p>&gt; Are you 10x more efficient?<p>90% of my time is spent thinking and talking about the problem and solutions. 10% is spent coding (sometimes 1% with 9% integrating this into existing infrastructure and processes). Even with ideal AGI coding agent id be only 10% more efficient.<p>Imagine a very bright junior developer. You still are heavily time taxed mentoring him and communicating.<p>Not many non technical people (to my surprise) get it.<p>Based on posts and comments here there are plenty “technical enough” people who don’t understand the essence of engineering work (software engineering in particular).<p>Spitting out barely (yet) working throwaway grade code is an impressive accomplishment for TikTok, but it has very little to do with complex business critical software most real engineers deal with everyday
评论 #43831683 未加载
评论 #43877594 未加载
评论 #43831901 未加载
评论 #43832097 未加载
juancn14 天前
I see the same mistake made everywhere, thinking that in software engineering that the hard part is making new code.<p>A large chunk of the work is dealing with people, understanding what do they really want&#x2F;need and helping them understand it.<p>On the technical side, most of the work is around fixing issues with existing software (protecting an investment).<p>Then, maybe 1 to 10% of the workload is making something new.<p>AI kinda works for the &quot;making something new&quot; part but sucks at the rest. And when it works, it&#x27;s at most &quot;average&quot; (in the sense of how good it&#x27;s training set was, it prefers things it sees more commonly regardless of quality).<p>My gut instinct is that there&#x27;s going to be an AI crash, much like in the late 90s&#x2F;early 2000s. Too much hype, and then, after the crash, maybe we&#x27;ll start to see something a bit more sane and realistic.
byoung214 天前
I am a tech lead and while I don&#x27;t use AI at my company (Disney) for writing code, since I have a team of contractors at my disposal, in my spare time I am working on a side project where ChatGPT is writing all of the code. It is an experiment mainly to see if it can be done. I am getting better at writing prompts to get better results, but I don&#x27;t think we are at the point where a nontechnical project manager could get good results. I feel like a lead or senior can use AI to replace interns and juniors, but more likely it is currently being used to make them more productive rather than replace people. It will be interesting to see the next few years, when it will be possible for one lead or senior to do the work of entire teams.
评论 #43832282 未加载
variadix14 天前
Not directly, but I wouldn’t be surprised if there’s enough of an efficiency improvement to obviate hiring an engineer or two (across 100+ people). In the same way that Google and StackOverflow made people more efficient when compared to having to otherwise search through and read physical documentation (to debug, to understand some API or hardware thing), LLMs have made me more efficient by being able to get tailored answers to my questions without having to do as much searching or reading. They can provide small code examples as clarification too.<p>In many ways LLMs feel like the next iteration of search engines: they’re easier to use, you can ask follow up questions or for examples and get an immediate response tailored to your scenario, you can provide the code and get a response for what the issue is and how to fix it, you can let it read internal documentation and get specialized support that wouldn’t be on the internet, you can let it read whole code bases and get reasonable answers to queries about said code, etc.<p>I don’t really see LLMs automating engineers end-to-end any time soon. They really are incapable of deductive reasoning, the extent to which they are is emergent from inductive phenomena, and breaks down massively when the input is outside the training distribution (see all the examples of LLMs failing basic deductive puzzles that are very similar to a well known one, but slightly tweaked).<p>Reading, understanding, and checking someone else’s code is harder than writing it correctly in the first place, and letting LLMs write entire code bases has produced immense garbage in all the examples I’ve seen. It’s not even junior level output, it’s something like _panicked CS major who started programming a year ago_ level output.<p>Eventually I think AI will automate software engineering, but by the time it’s capable of doing so _all_ intellectual pursuits will be automated because it requires human level cognition and adaptability. Until then it’s a moderate efficiency improvement.
评论 #43851446 未加载
jmisavage14 天前
We’re in the early stages of this transition. There’s no formal hiring freeze, but leadership has made it clear we should exhaust AI options before considering new hires. At the same time, raises and promotions are frozen this year, which has definitely caused a lot of frustration internally.<p>As part of this AI-first shift, all engineers now have access to Cursor, and we’re still figuring out how to integrate it. We just started defining .cursorrules files for projects.<p>What’s been most noticeable is how quickly some people rely too much on AI outputs, especially the first pass. I’ve seen PRs where it’s obvious that the generated code wasn’t even run or reviewed. I know this is part of the messy adjustment period, but right now, it feels like I’m spending more time reviewing and cleaning up code than I did before.
markus_zhang14 天前
For now it&#x27;s more like AI boosting productivity so company doesn&#x27;t have to hire more.<p>We are a team of 5 down from 8 a few months ago, and we are working on more stuffs. I would not be able to survive without AI writing some queries and scripts for me. It really saves a tons of time.
baq14 天前
Regardless of its realized effectiveness improvements it froze the intern&#x2F;junior hiring pipelines.
评论 #43831763 未加载
评论 #43831682 未加载
评论 #43831760 未加载
gitfan8614 天前
Most but not all companies are bottlenecked by organizatial issues not speed of completing a jira ticket. A lot of those companies have moats or sales issues that prevent competitors from easily taking market share.<p>So instead of seeing mass drop in job openings you will see companies that are not bottlenecked by org issues start to move very fast. In general that will create new markets and have a positive effect on kobs
评论 #43845989 未加载
chrisgd14 天前
Duolingo made this announcement today about replacing contractors with AI<p><a href="https:&#x2F;&#x2F;www.theverge.com&#x2F;news&#x2F;657594&#x2F;duolingo-ai-first-replace-contract-workers" rel="nofollow">https:&#x2F;&#x2F;www.theverge.com&#x2F;news&#x2F;657594&#x2F;duolingo-ai-first-repla...</a>
评论 #43832109 未加载
评论 #43831645 未加载
ilaksh14 天前
I use AI as much as possible for programming, but the specific wording &quot;replacing&quot; is not quite there yet. The confusing thing for people is it probably will be at full replacement level within a couple of years.<p>The leading edge models surpass humans in some ways, but still make weird oversights routinely. I think the models will continue to get bigger and have more comprehensive world models and the remaining brittleness will go away over the next few years.<p>We are early on in a process that will go from only a few jobs to almost all (existing) jobs very quickly as the models and tools continue to rapidly improve.
评论 #43832093 未加载
评论 #43831984 未加载
joshuanapoli14 天前
In software development, we usually don’t really have a firm scope that gets completed in a clean way. So when developers get more efficient (from high level languages, OOP, Agile, Internet, AI, etc.) I think that we normally slide into a bigger scope, rather than finishing sooner or reducing the team size. Everyone usually gets the same boost from a coding productivity innovation at about the same time. So team size for products in competitive markets isn’t affected much by productivity. Improvements received by the customers accelerate, rather than developer job cuts.
评论 #43833184 未加载
评论 #43832073 未加载
评论 #43851476 未加载
zooom13 天前
My impression is that LLMs, at least currently, are just like any other modest increase in the power of our tooling. CRUD apps get even easier. Simple UIs get even easier. And on the business side, no-code&#x2F;low-code (no coder, no human) solutions are hyped to the moon.<p>But the end result will be, once (if) the economy becomes healthy again, businesses will just become more ambitious and software get more hardware intensive and slower. Same ole same ole.
jonplackett14 天前
Is &#x27;replacing&#x27; the right way to think of it though?<p>I don&#x27;t see any AI yet anywhere near good enough to literally do a person&#x27;s job.<p>But I can easily see it making someone, say 20%-50% more effective, based on my own experience using it for coding, data processing, lots of other things.<p>So now you need 8 people instead of 10 people to do a job.<p>That&#x27;s still 2 people who won&#x27;t be employed, but they haven&#x27;t been &#x27;replaced&#x27; by AI in the way people seem to think they will be.
评论 #43831812 未加载
评论 #43831778 未加载
评论 #43831749 未加载
scarface_7414 天前
My anecdote is that I am in cloud consulting specializing in app dev. 90%+ of my projects are greenfield development.<p>Before LLMs got good enough, there were projects I would scope with the expectation of having one junior consultant do the coding grunt work - simple Lambdas, Python utility scripts, bash scripts, infrastructure as code, translating some preexisting code to the target language of the customer.<p>This is the perfect use case for ChatGPT. It’s simple well contained work that can fit in its context window, the AWS SDKs in various languages are well documented, there is plenty of sample code, and it’s easy enough to test.<p>I can tell it to “verify all AWS SDK functions on the web” or give it the links to newer SDK functionality.<p>I don’t really ever need a junior developer for anything. If I have to be explicit about the requirements anyway, I can use an LLM.<p>And before the the gate keeping starts, I’ve been coding as a hobby in 1986 and started coding in assembly language then and have been coding professionally since 1996.
throwaw1214 天前
&gt; where they stop recruiting and replace human engineers with AI<p>I don&#x27;t think it is possible NOW.<p>But for specific areas, productivity gain you get from a single developer with LLM is much higher than before. Some areas I see it is shining:<p><pre><code> * building independent React&#x2F;UI components * boilerplate code * reusing already solved solutions (e.g. try algorithm X,Y,Z. plot the chart in 2D&#x2F;3D,...) </code></pre> &gt; What changed significantly in your workflow? Hiring freeze, because leaders are not sure yet about the gains from AI, what if we hire bunch of people and can&#x27;t come up with projects for them (not because we are out of ideas, because getting investment is hard if you are not AI company), while LLM is generating so much code.<p>&gt; Are you 10x more efficient? Not always, but I am filtering out things faster giving me opportunity to get into the code concepts sooner (because AI is summarizing it for me before I read 10 page blogpost)
Balgair14 天前
My company isn&#x27;t backfilling positions anymore and we had a ~10% company wide layoff about 2 weeks ago. My team was told to use AI to fill in the roles that were lost on the team [0].<p>What&#x27;s changed in the workflow is a lot really. We do a lot of documentation, so most of that boilerplate is not done via AI based workflows. In the past, that would have been one of us copy-pasting from older documents for about a month. Now it takes seconds. Most of the content is still us and the other stakeholders. But the editing passes are mostly AI too now. Still, we very much need humans in the loop.<p>We don&#x27;t use copilot as we&#x27;re doing documentation, not code. We mostly use internal AIs that the company is building and then a vendor that supports workflow-style AI. So, like, iterative passes under the token limits for writing. These workflows do get pretty long, like 100+ steps, just to get to boilerplate.<p>We&#x27;re easily 100x more efficient. Four of us can get a document done in a week that took the whole team years to do before.<p>The effort is more concentrated now. I can shepherd a document to near final review with a meeting or two from the specialist engineers, that used to take many meetings with much of both teams. We were actually able to keep up and not fall behind for about 3 months. But, management see us as a big pointless cost center of silly legal compliance, so we&#x27;re permanently doomed to never get to caught up. Whatever, still have a job for now.<p>I guess my questions back are:<p>- How do you think AI is going to change the other parts of your company than coding&#x2F;engineering?<p>- Have you seen other non engineering roles be changed due to AI?<p>- What do your SOs&#x2F;family think of AI in their lives and work?<p>- How fast do you think we&#x27;re getting to the &#x27;scary&#x27; phase of AI? 2 years? 20 years? 200 years?<p>[0] I try to keep this account anonymous as possible, so no, I&#x27;m not sharing the company.
nusl14 天前
Salesforce is one company I&#x27;m aware of that announced this sort of thing.<p>I&#x27;m personally only more productive with the help of AI if one the following conditions are met;<p>1. It&#x27;s something I was going to type anyway, but I can just press Tab and&#x2F;or make a minor edit<p>2. The code produced doesn&#x27;t require many changes or time in understanding, as the times where it has required many changes or deeper understanding probably would have been faster to just code myself<p>Where it has been helpful, though, is debugging errors or replacing search engines for helping out with docs or syntax. But, sometimes it produces bullsh*t that doesn&#x27;t exist and this can lead you down a rabbithole to nowhere.<p>More than once it&#x27;s suggested something to me that solved all of the things I needed, only to realise none of it existed.
biggestdoofus14 天前
I think it just depends on the complexity of what you are working with. For the simpler stuff it seems to work very well. However it becomes a gigantic time sink if you try to use it for more complex tasks, you just go in circles while it has no real idea of what to do. It&#x27;s not just more complex code it struggles with, it could be simple code in complex systems as well, where a foundational understanding of the different parts is essential.<p>The people writing boring crud apps should be scared (but I think it&#x27;s a failure in our industry that this is still a thing).<p>The technical debt that will be amassed by AI coding is worrying however. Coworkers here routinely try to merge inn stuff that is just absolute slop, and now I even have to argue with them on the basis that they think it&#x27;s right because the AI wrote it...
anonzzzies14 天前
For frontend, we don&#x27;t need people anymore: backend, especially complex stuff, LLMs really waste our time by not getting it&#x2F;going in circles so we just don&#x27;t really do that anymore as plowing through 100s of lines of code or prs that are wrong, not conform to standards, has libs included we don&#x27;t need etc. It is pretty useless as agents will just go in circles until it&#x27;s so messy that it would&#x27;ve been many times easier just writing it with code. And often it&#x27;s hard to get out besides just rollback.
评论 #43832112 未加载
评论 #43832072 未加载
评论 #43832057 未加载
ojr14 天前
My friend wants me to build his app for free because he heard the cost of software labor has gone to zero. He gave me ChatGPT responses, thinking that they could help me add features after I told him the complexity was too high.<p>In a sprint planning scenario, I think tasks that were 1,2,3,5,8,13, etc., get put down a notch, nothing more, with the invention of AI. AIs have not made an 8-point task into a 3-point one at all. There is a 50&#x2F;50 chance that an old 8-point task before AI remains 8 points, with it sometimes dropping to 5.
anant9014 天前
The AI-powered &quot;replacement of engineers&quot; that everyone keeps talking about will look less like existing engineers being laid off, and more like reduced hiring of recent engineering graduates. And, as with any large-scale technology trend, it will take a while before we can say we&#x27;ve come out of the innovator&#x2F;early adopter phase, which we are clearly still in. In my opinion, it&#x27;s always easier to invent new technology than to get people to change the ways they currently do things.
评论 #43832130 未加载
vvojd12 天前
There are two main places where AI still differs from humans. Long enough thought chains and proper memory&#x2F;learning.<p>On the first, you really have to consider a number of options when refactoring or adding to a codebase. On the latter, you may be able to get away with having an extremely detailed manual but ultimately a lot of day to day things aren’t suitable to a RAG.<p>So no, no one’s getting fired anytime soon.
fhd214 天前
Can&#x27;t directly answer your question, since I&#x27;m not working at a company that makes any claims about hiring less human engineers because of automation.<p>But I think the central question is not how much of software development can be automated. It&#x27;s rather how many engineers companies _believe_ they need.<p>Having spent some time in mid sized companies adjacent to large companies, the sheer size of teams working on relatively simple stuff can be stunning. I think companies with a lot of money have overstaffed on engineers for at least a decade now. And the thing is: It kinda works. An individual or a small team can only go so far, a good team can only grow at a certain rate. If you throw hundreds of engineers at something, they _will_ figure it out, even if you could theoretically do it with far less, by optimising for quality hires and effective ways of working. That&#x27;s difficult and takes time, so if you have the money for it, you can throw more bodies at it instead. You won&#x27;t get it done cheaper, probably also not better, but most likely faster.<p>The mere _idea_ that LLMs can replace human engineers kinda resets this. The base expectation is now that you can do stuff with a fraction of the work force. And the thing is: You can, you always could, before LLMs. I&#x27;ve been preaching this for probably 20 years now. It&#x27;s just that few companies dared to attempt it, investors would scoff at it, think you&#x27;re being too timid. Now they celebrate it.<p>So like many, I think any claims of replacing developers with AI are likely cost savings in disguise, presented in a way the stock market might accept more than &quot;it&#x27;s not going so well, we&#x27;re reducing investments&quot;.<p>All that aside, I also find it difficult as a layperson to separate the advent of coding LLMs from other, probably more consequential effects, like economic uncertainty. When the economy is stable, companies invest. When it&#x27;s unstable, they wait.
rcarmo13 天前
I want to read about AI replacing C-levels, because most of the non-coding output I see has the same verbosity and general vagueness that comes with too much abstraction.
orwin14 天前
Let&#x27;s say that as long my as the complexity is low, it is really worth using AI, and you can be twice as effective, maybe more, because it can take care of most of the coding&#x2F;testing&#x2F;integration which are 80% of project, and can help you with the architecture part as long as it is easy&#x2F;standard.<p>As the complexity grow, the usefulness of AI agents decrease: a lot, and quite fast.<p>In particular, integration of microservices are a really hard case to crack for any AI agent as it often mix training data with context data.<p>It is more useful in centralised apps, and especially for front dev, as long as you don&#x27;t use finite state machines. I don&#x27;t understand why, even Claude&#x2F;Cursor trip on otherwise really easy code (and btw if you don&#x27;t use state machines for your complex front end code, you&#x27;re doing it wrong).<p>As long as you know what your agent is shitty at however, using AI is a net benefit as you don&#x27;t loose time trying to communicate your needs and just do it, so it is only gains and no loses.
babyent14 天前
I use AI&#x2F;LLM to run thought experiments randomly or learn about topics. Sometimes I use it to help me with code (not to write code).
jakeoverflow12 天前
AI is a great 20% productivity booster - but 20% and 20x are orders of magnitude apart.
teeray14 天前
It will be interesting if and when these companies reach a “find out” stage with AI where their entire codebase is incomprehensible slop that not even the AIs can help them with.
kypro14 天前
Depending on the project Devin.ai should be able to replace ~50% of development. No one should be hiring junior devs in 2025 imo.
评论 #43831806 未加载
评论 #43831770 未加载