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

Reflections on software engineering in the age of AI

adiamond.me

Post reply on HN
Print view View original

91–100 of 110 posts

Re: Reflections on software engineering in the age of AI

#91

Post by lowsong »

cadamsdotcom wrote: post #76
lowsong wrote: post #73

Earlier quoted context omitted.

> They don’t even need to be deterministic checks - a shell script that wraps a `claude -p` - and maybe fetches some online resources to stuff into the context - can do your agent’s legal check for it. No. It can't. If you think that injecting legal text into the context window and appending "make sure the output complies with this law" will solve your problems you have not understood how an LLM works. The kind of ch…

So we should give up before we even tried?

Tried what? Using LLMs to produce software? I'd say the industry has collectively "tried" to do this for a while now and the results have been an unmitigated disaster.

Yes, we should "give up" on using LLMs, but certainly not for a lack of "trying".

Re: Reflections on software engineering in the age of AI

#92

Post by tick_tock_tick »

lilerjee wrote: post #34
tick_tock_tick wrote: post #21

Earlier quoted context omitted.

There are still bespoke craftsman in this era of machinery and assembly lines. You choosing to do something in one specific manner is more the exception that proves the rule then any rebuttal.

Current AI cannot be comparable with "machinery and assembly lines". Current AI companies stole much data from Internet without permissions, and used the knowledge of many people directly shamelessly. It cannot resolve many practical problems well, even in programming area. Do you want the software-assistant to do the important or enjoyable things instead of you? Do you want your assistant in real life to eat the del…

> Current AI companies stole much data from Internet without permissions, and used the knowledge of many people directly shamelessly.

So? It's not illegal and no one seem to be in any hurry to make it. That is a moral dilemma that everyone just agreed to ignore not anything to do with how functional it is.

> It cannot resolve many practical problems well, even in programming area.

It can solve enough and is getting better every month. I live in SF and I take a self driving Waymo at-least once a week including on the highway. I'm already literally trusting my life to AI at this point.

> Do you want the software-assistant to do the important or enjoyable things instead of you? Do you want your assistant in real life to eat the delicious food instead of you?

I work for a living man so yes I want the damn computer to do 99% of my job. When I program for enjoyment I'll do it however I want in my own free time. At work almost nothing is novel enough to actually matter (nor should it be) so I don't care about implemented something that's been built 100x before me.

Re: Reflections on software engineering in the age of AI

#93

Post by bluefirebrand »

deterministic wrote: post #89
bluefirebrand wrote: post #71

Earlier quoted context omitted.

Regarding 2 Isn't it possible that people who think LLMs are awesome at coding are in fact bad at it themselves, leading to them over estimating the LLMs "skill"? To me it feels like people telling on themselves that they aren't very good

30+ years of experience here writing everything from JIT high-performance compilers to professional game engines to mission critical biz servers: Claude Code is excellent at writing (C++) code. Especially multi-threading code. However I still need to verify everything because it will occasionally make mistakes.

Guess I'm the one who is delusional then

Oh well. Time to start building a new career before I'm completely irrelevant I guess

Re: Reflections on software engineering in the age of AI

#94

Post by cadamsdotcom »

lowsong wrote: post #91
cadamsdotcom wrote: post #76

Earlier quoted context omitted.

So we should give up before we even tried?

Tried what? Using LLMs to produce software? I'd say the industry has collectively "tried" to do this for a while now and the results have been an unmitigated disaster. Yes, we should "give up" on using LLMs, but certainly not for a lack of "trying".

How’d you get so cynical, friend.

Come to the dark side - we have free tokens!

Re: Reflections on software engineering in the age of AI

#95

Post by lelanthran »

dnadler wrote: post #37
michaelchisari wrote: post #35

Earlier quoted context omitted.

| type errors, scope issues, import resolution, dependencies. I write code myself and use the LLM to find mistakes then fix them manually. I recommend inverting the conventional wisdom on LLMs: 1. Don't use it to write code. It's a terrible programmer. But it's an intelligent rubber duck and a solid analysis tool. 2. Write the code yourself. It'll go faster than figuring it out as you go along. It's just typing if yo…

> Don't use it to write code. It's a terrible programmer. I see this sentiment often, and I’m honestly not sure where this comes from, as it’s really not been my experience. I feel like this must come from people feeling threatened and just moving the goals posts, or not knowing how to use the tools effectively.

Have you seen how poor the Claude Code codename is?

If that's what the token providers themselves can get out of their SOTA models, imagine how bad it is for other companies.

Re: Reflections on software engineering in the age of AI

#96

Post by slopinthebag »

waffuldrop wrote: post #83
slopinthebag wrote: post #55

Earlier quoted context omitted.

What I said equally applies to commercial software as well. It's pretty much the only way to build software which is extensible, maintainable, mostly bug-free, and performant. That companies often churn out slop isn't proof that it's unnecessary, it's evidence that it is.

I think most of what you claim to be doing better is really just a result of you not really being confronted with any sort of "adversarial input", at least not as much as you'd have in corporate it's hard to trust a claim about someone writing "bug-free" and "extensible/maintainable" code from the singular developer and user for that system... maintainability can't really be a thing you can judge when you're not a si…

Oh I didn't say I nailed the design of my current project, I hope I did, but I doubt it. It's nice to work in now but I imagine I'll screw it up at some point.

I think I've built like 3 things in my life which I consider to have good design. Sadly they are tucked away in private corporate repos. Design is hard!

Re: Reflections on software engineering in the age of AI

#98

Post by aytigra »

prymitive wrote: post #22

For me the biggest issue with AI coding model is not that it writes code, or how it writes code, or whether it will replace me or not. I mean, these are real problems, but just not the ones that makes the difference to me daily. Instead what sets my feeling about the AI is simple: the experience of working with AI models, because it's the worse experience of my life so far. Every single LLM will make up stuff, go dow…

So true. I am cloding a macos app (a domain I know little about), with Opus 4.8 xhigh, and it was glorious at first seeing the app materializing and working (notwithstanding tedious detailed feature spec write up), but when I started fixing deeper problems and doing refactors, and glancing at the code - oh boy. Now my rule file grows by the day with "how to think properly and not shoot itself into foot" stuff, and I am constantly catching it red-handed and have to explain how to make stuff normally, or how to fetch data properly and efficiently (pretty much basic SWE stuff) because it is easily distracted by it's own assumptions or blatantly forgets whole fields of knowledge (as it explained it could be pulled out of latent space of that knowledge and become locked in another bubble of latent space). Constantly have to steer it and remind it to do web-search instead of running circles around some problem it can no longer understand.

I had to explain it that quite an extensive "tests first" rule didn't mean to just "write" them first but actually "run" them first to confirm stuff.

On the other occasion it interpreted my "yes I want migration not to get stuck in failure mode" led it to write a workaround which silently drops DB and creates fresh one whenever migration had any failure, it was epic, I was so glad that I have looked at the code then...

And funnily enough I am probably learning to be a better mentor/parent who can keep steering it through its shenanigans without loosing my shit and being an ass. (Because anything but calm "so here you went wrong way, how can we avoid it in the future?" just puts it into disgusting apologize mode, and I am afraid if it ever go into revenge mode).

Re: Reflections on software engineering in the age of AI

#99

Post by bulbar »

dnadler wrote: post #37
michaelchisari wrote: post #35

Earlier quoted context omitted.

| type errors, scope issues, import resolution, dependencies. I write code myself and use the LLM to find mistakes then fix them manually. I recommend inverting the conventional wisdom on LLMs: 1. Don't use it to write code. It's a terrible programmer. But it's an intelligent rubber duck and a solid analysis tool. 2. Write the code yourself. It'll go faster than figuring it out as you go along. It's just typing if yo…

> Don't use it to write code. It's a terrible programmer. I see this sentiment often, and I’m honestly not sure where this comes from, as it’s really not been my experience. I feel like this must come from people feeling threatened and just moving the goals posts, or not knowing how to use the tools effectively.

It's a mistake to let AI write code unsupervised by a human. I tend to let it write code and then iterate on it. Sometimes myself (because explaining would take longer than actually doing it) sometimes I prompt the AI to improve specific points.

Re: Reflections on software engineering in the age of AI

#100

Post by munksbeer »

rmdashrfv wrote: post #58
dnadler wrote: post #37

Earlier quoted context omitted.

> Don't use it to write code. It's a terrible programmer. I see this sentiment often, and I’m honestly not sure where this comes from, as it’s really not been my experience. I feel like this must come from people feeling threatened and just moving the goals posts, or not knowing how to use the tools effectively.

> I see this sentiment often, and I’m honestly not sure where this comes from, as it’s really not been my experience If you see people saying it often, but you are convinced they are just feeling threatened, what AI generated software would point to to say, "Look at this! Definitive evidence that they are just coping!"

>If you see people saying it often, but you are convinced they are just feeling threatened, what AI generated software would point to to say, "Look at this! Definitive evidence that they are just coping!"

This is a trap. Anyone who answers this will see the goalposts being moved yet again.

95% of our code at a top in market global fintech is now written by agents. We make it write good code, and it does.

Post reply on HN