Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Ask HN: By what percentage has AI changed your output as a software engineer?

news.ycombinator.com

Post reply on HN
Print view View original

81–90 of 96 posts

Re: Ask HN: By what percentage has AI changed your output as a software engineer?

#81

Post by countWSS »

Massive x100 speed-up for writing simple, mind-numbing stuff. Algorithm optimization +20%*(Depends on domain). Obscure APIsx10(deals with APIs much better than docs) Debugging is massively easier(incomparably). I don't know waht HN hate for AI is coming from, but its ideal for prototyping/debugging/refining code. Minor polishing/editing specifics is much easier for me, rather than writting a correct "draft v1.0" from scratch: since i typicall evolve structures from primitive(v0.1) while AI gives me a chance to start from 1.0 and polish it later(optimize/fix/improve/etc). For "from scratch to working prototype" workflow would mean months of coding condense to days(mostly debugging/flaw finding).

Drawbacks:I feel like more code is simplified, no ad-hoc complex structure after AI rewrite. Pure "clean code" with most standard optimizations and control flow. AI can fix "good complex code" into simplified "enterprise ready" control flow charts in code form, it lacks the sense of 'code quality' instead thinking verbose, clean code is the quality marker.

Re: Ask HN: By what percentage has AI changed your output as a software engineer?

#83

Post by bradlys »

Net negative. People keep using AI slop and are driven by shitty metrics from the top to use AI as much as possible. It leads to the entire codebase being garbage. I blame the higher ups. Everyone is just doing their job and has no room to negotiate here. Everything is tracked via metrics and you’re required to do a bunch of stupid AI slop shit otherwise you’re at risk of getting fired.

This is at a faang.

Re: Ask HN: By what percentage has AI changed your output as a software engineer?

#84

Post by omnimus »

jen729w wrote: post #67
omnimus wrote: post #35

Earlier quoted context omitted.

This. LLMs are good at stuff that is very general (is often in the dataset). What i gain most from LLM is when i use it to teach me - like extended documentation. But to make unique solutions you will get pretty random results and worse you are not building understanding and domain knowledge of your program. Claude Code sounds cool until it makes 3 changes at once 2 of which you are unsure if they are required or if…

> What i gain most from LLM is when i use it to teach me - like extended documentation. This, yes. What I do now is use Claude but expressly tell it do not edit my code, just show me, I want to learn . I'm not a very experienced dev so often it'll show me a pattern that I'm unfamiliar with. I'll use that new knowledge, but then go and type out the code myself. This is slower, in the moment. But I am convinced that th…

It is especially helpful when new in some framework where you aim to follow best practices so others can follow and you don't end up reinventing.

Re: Ask HN: By what percentage has AI changed your output as a software engineer?

#85

Post by jaredsohn »

alloyed wrote: post #55

So, in my professional work environment, which is a complex C++ app with long build times and lots of proprietary domain-specific knowledge, LLMs were worse than useless. Not totally surprising, but I originally had hopes for something like Cursor being useful in terms of simplifying the process of large mechanical refactors, which it decidedly wasn't. It could suggest interesting and complex uses of C++ templates, b…

>(my understanding of claude code is that it's a non-interactive agent, which is worse for what i have in mind. iteration and _changing my mind_ are a big part of my process, so even if I let the computer do its own thing for an hour and work on something else, that's less productive than spending even 10 minutes of focused time on the same thing.)

Just use 'plan mode'; it will ask clarifying questions.

Re: Ask HN: By what percentage has AI changed your output as a software engineer?

#86

Post by readthenotes1 »

scioto wrote: post #27

-100%. With this looming Shiny Toy plus other sub-optimizations at my last place of work, I decided to retire. Didn't want to deal with it.

Sounds like it's +100%, assuming you are now doing things that provide beyond food and shelter security...

I would say +110% to remind you of the effort your childhood coaches almost assuredly requested of you, but didn't want to trigger the quibblers who would point out that's impossible (not that there's anything wrong with that)

Re: Ask HN: By what percentage has AI changed your output as a software engineer?

#87

Post by d13z »

From -10x to 5x.

It depends a lot. I work as a web platform engineer so the datasets on some parts of what I do are limited. For a react project, there are millions of projects, code and setups in the training data. To create and manage a monorepo that runs on our specific infrastructure, the story is different. Writing bash scripts to run on GitHub actions: 10x. Ask to setup a monorepo: -3x. Modify a specific business domain: -10x. A couple of weeks ago I was handed a project that was the implementation of a experimentation sdk. The project was made by previous developer with Claude. I couldn't understand a thing. I had to basically start from scratch to understand all the principles of how this experimentation tool worked. This was not fault of the previous Devs, it was a combination of reading code from a third³ person (person through an LLM) plus I'm a bottom up processor.

Re: Ask HN: By what percentage has AI changed your output as a software engineer?

#88

Post by moltar »

d13z wrote: post #87

From -10x to 5x. It depends a lot. I work as a web platform engineer so the datasets on some parts of what I do are limited. For a react project, there are millions of projects, code and setups in the training data. To create and manage a monorepo that runs on our specific infrastructure, the story is different. Writing bash scripts to run on GitHub actions: 10x. Ask to setup a monorepo: -3x. Modify a specific busine…

I manage monorepos using Projen and CC is quite good with that with just a small skill.

Re: Ask HN: By what percentage has AI changed your output as a software engineer?

#90

Post by ikidd »

10X. 100X. I have very, very little time for programming anymore, but I have a lot of ideas for things I never had time to make.

In the last 6 months I've built (with the $20/mo plan from Cursor):

- shared tasks app that tracks fleet maintenance on my farm, with a personal tasks view. We can now search part numbers by equipment or refer back to previous work to look up invoices, as well as actually get jobs lined up for when we have help so they aren't sitting there staring at the walls because nobody has a list of what to do next.

- LoRa telemetry mesh for soil moisture sensors, gate motion sensors, waterbowl temperature and level sensors. This would have taken me months to build and debug, Cursor had it done in minutes, I prototyped and then punched up some boards on kicad and ordered them from JLCPCB. Took a couple days to debug them into a workable state and was in place for winter.

- Fertilizer planning application for figuring out our fertilizer buy while prices are low, based on:

- Soil sampler truck application for Android tablet, that interfaces a sub-3m Garmin GPS connected via bluetooth and tracks all the sample site for the sample truck as samples are punched, so we can go back to the same spot every year and determine soil regeneration results.

- Android application to track and calculate loads for silage feeding of cattle, since as you feed you'll want to rebalance the grain ration as you refill, and you'll often have silage at X %age left on, this app lets you enter the mix you want and calculates how much of each to add in order to rebalance the leftovers on the truck to the new ration. I actually did this one on my laptop while I was feeding cattle over a week.

- An alert app to watch MQTT and set phone alarms off if an alert pops up for telemetry problems or intrusion alerts on above LoRa mesh.

Currently working on a system that lets me run multiple Cursor sessions on a workstation and interact with the chats remotely over my phone so I can work on projects while I'm running equipment or doing chores.

A lot of people hate AI, and for a lot of things I totally get it. Frankly, the ethics of AI and the horrible uses people are finding for it scares the hell out of me. But I would never have found time to do any one of these items before AI. I used to run software projects when I was an IT nerd, so I know how program well enough to be dangerous. I'm enjoying having a team back.

Post reply on HN