Ironically, this fell off the HN front page without enough upvotes...neither hackernews.coffee nor Claude suggested it to me...
Oh, itsdrewmiller pointed us in the right direction: https://news.ycombinator.com/item?id=44551579
AI coding tools make developers slower but they think they're faster study finds
31–40 of 40 posts
Re: AI coding tools make developers slower but they think they're faster study finds
#32My personal experience: I shipped multiple features at work in the past 6 months that I simply wouldn’t have tried shipping otherwise, since my day job is mostly management. AI wrote maybe 80% of the code, I spent a bit of time rewriting some parts. No major bugs so far (ironically, the one big bug the team had to revert was done entirely by me)
I can guarantee I wouldn’t have shipped ANY of it, since it’d require focus blocks I simply don’t have on the job.
I’m also about to ship a Mac app that’s heavily vibecoded. I wouldn’t even try without AI, since I’m not a Swift developer.
Those aren’t “illusions” of performance. I imagine it’s hard to gauge every single scenario, and sensationalist takes like this research elicit an emotional response on the anti-AI crowd, but denying the impact is simply ignorance at this point…
Re: AI coding tools make developers slower but they think they're faster study finds
#33Earlier quoted context omitted.
This is literally what the study says: the developers said they felt they were faster after having completed the task, when in fact they were slower. Your "lived experience" just follows the same bias.
Ok let me rephrase. Actual tangible real results say different. I know the anti AI folks would like to believe AI coding results are a mass hallucination but the completed software that I build in record time with deep functionality says otherwise.
How did you measure your before and after?
Re: AI coding tools make developers slower but they think they're faster study finds
#34My personal experience: I shipped multiple features at work in the past 6 months that I simply wouldn’t have tried shipping otherwise, since my day job is mostly management. AI wrote maybe 80% of the code, I spent a bit of time rewriting some parts. No major bugs so far (ironically, the one big bug the team had to revert was done entirely by me) I can guarantee I wouldn’t have shipped ANY of it, since it’d require fo…
> Those aren’t “illusions” of performance.
Given that your job is not software development but management, you spending time delivering features is effectively removing time from doing your job.
If you had spent managing the same time you spent vibe-coding, maybe it would have been a force-multiplier for your reportees and your team might have been more productive as a whole than the added productivity of your vibe-coding.
This is absolutely an illusion of performance.
Re: AI coding tools make developers slower but they think they're faster study finds
#35My personal experience is that in well and quickly autocompleted languages I know well (python) then I am roughly the same speed, I'd say...it's just a bit less annoying having AI do simple boilerplate for me, but sometimes annoying whenever I try to use it for larger refactoring where it gets style and structure incorrect. In languages and libraries I know less well - vuejs+myriad of (especially) js libraries, I wou…
> do simple boilerplate
I just don't buy this.
Everytime you say AI is useless because it can't solve complex problems, people will bring up "oh but it writes boilerplate code for me".
How often do you exactly write boilerplate code??? Do you know what else writes boilerplate code for you? Libraries and framework.
Boilerplate code is a solved problem since way before GenAI was in the public eye.
Re: AI coding tools make developers slower but they think they're faster study finds
#36My personal experience: I shipped multiple features at work in the past 6 months that I simply wouldn’t have tried shipping otherwise, since my day job is mostly management. AI wrote maybe 80% of the code, I spent a bit of time rewriting some parts. No major bugs so far (ironically, the one big bug the team had to revert was done entirely by me) I can guarantee I wouldn’t have shipped ANY of it, since it’d require fo…
> Those aren’t “illusions” of performance. Given that your job is not software development but management, you spending time delivering features is effectively removing time from doing your job. If you had spent managing the same time you spent vibe-coding, maybe it would have been a force-multiplier for your reportees and your team might have been more productive as a whole than the added productivity of your vibe-c…
I tend to agree here. I haven't been allowed to code for a few years now but I spend a lot of my time talking through code with developers. I find many of the people on my teams lack a perspective I can provide to frame a problem or evaluate an approach.
I also help them get to the heart of problems quickly simply because I'm not stuck in the code all day. For example, if I see a developer taking too long to identify the source of a bug, I'll get on a call and get them to take me through that code and get them prove any assumption ("ok, show me the code that checks that value is greater than zero").
By doing this I'm using my coding experience directly without actually coding. I'd consider coding a huge waste of time for me, but spending 30 minutes to unstick a developer when I am sure they should have found the problem by now seems like a really good use of my time.
It also lets people know they can't just spend three days on something that should take a couple hours without someone checking in, which I don't live having to do but it's a reality for some teams I work with.
Re: AI coding tools make developers slower but they think they're faster study finds
#37My personal experience: I shipped multiple features at work in the past 6 months that I simply wouldn’t have tried shipping otherwise, since my day job is mostly management. AI wrote maybe 80% of the code, I spent a bit of time rewriting some parts. No major bugs so far (ironically, the one big bug the team had to revert was done entirely by me) I can guarantee I wouldn’t have shipped ANY of it, since it’d require fo…
> Those aren’t “illusions” of performance. Given that your job is not software development but management, you spending time delivering features is effectively removing time from doing your job. If you had spent managing the same time you spent vibe-coding, maybe it would have been a force-multiplier for your reportees and your team might have been more productive as a whole than the added productivity of your vibe-c…
> Given that your job is not software development but management, you spending time delivering features is effectively removing time from doing your job.
It most definitely isn’t. With all due respect, I know my job and my schedule more than you and your baseless assumptions.
Re: AI coding tools make developers slower but they think they're faster study finds
#38My personal experience is that in well and quickly autocompleted languages I know well (python) then I am roughly the same speed, I'd say...it's just a bit less annoying having AI do simple boilerplate for me, but sometimes annoying whenever I try to use it for larger refactoring where it gets style and structure incorrect. In languages and libraries I know less well - vuejs+myriad of (especially) js libraries, I wou…
> do simple boilerplate I just don't buy this. Everytime you say AI is useless because it can't solve complex problems, people will bring up "oh but it writes boilerplate code for me". How often do you exactly write boilerplate code??? Do you know what else writes boilerplate code for you? Libraries and framework. Boilerplate code is a solved problem since way before GenAI was in the public eye.
Absolutely every layer of abstraction reduces boilerplate, but it's not as if it's entirely gone. I should say, I use boilerplate as a euphemism for 'trivial code'. For example recreating backed request models in the frontend, a simple "create db record" service function, etc.
I mostly use vuejs at the front and fastapi for the backed, and unless I'm missing something big, there is still plenty of 'trivial code' that I find is done a bit faster.
bug creation rates of AI trival code vs my own trivial code has not been evaluated
Re: AI coding tools make developers slower but they think they're faster study finds
#39With all due respect to the people who have studied this, I do not care. I'm relatively young, and yet the pain of typing has started to become unmistakeable and chipped away at my productivity, motivation, and sense of invincibility. You could tell me AI coding makes me 50% slower. I'm taking it. I refuse to grind my wrists to dust.
> and yet the pain of typing has started to become unmistakeable
As in _physical pain_? If you haven't already, go to a doctor; that's not normal (it's somewhat common, but it is far from inevitable and there are things you can do about it).
Re: AI coding tools make developers slower but they think they're faster study finds
#40Earlier quoted context omitted.
> do simple boilerplate I just don't buy this. Everytime you say AI is useless because it can't solve complex problems, people will bring up "oh but it writes boilerplate code for me". How often do you exactly write boilerplate code??? Do you know what else writes boilerplate code for you? Libraries and framework. Boilerplate code is a solved problem since way before GenAI was in the public eye.
Absolutely every layer of abstraction reduces boilerplate, but it's not as if it's entirely gone. I should say, I use boilerplate as a euphemism for 'trivial code'. For example recreating backed request models in the frontend, a simple "create db record" service function, etc. I mostly use vuejs at the front and fastapi for the backed, and unless I'm missing something big, there is still plenty of 'trivial code' that…
> For example recreating backed request models in the frontend
https://pypi.org/project/pydantic-to-typescript/
> a simple "create db record" service function
https://en.wikipedia.org/wiki/Object%E2%80%93relational_mapp...
> unless I'm missing something big
You are missing a lot of things, yes.