Live data from Hacker News

Ask HN: Anyone struggling to get value out of coding LLMs?

news.ycombinator.com

61–70 of 293 posts

Re: Ask HN: Anyone struggling to get value out of coding LLMs?

#61
There's no silver bullet.

It is amazing how in our field we repeatedly forget this simple advice from Fred Brooks.

In my experience, LLMs are way more useful for coding and less problem-prone when you use them without exaggerated expectations and understand that it was trained on buggy code, and that of course it is going to generate buggy code. Because almost all code is buggy.

Don't delegate design for it, use functional decomposition, do your homework and then use LLMs to eliminate toil, to deal with the boring stuff, to guide you on unfamiliar territory. But LLMs don't eliminate the need for you to understand the code that goes with your name. And usually, if you think a piece of LLM generated code is perfect, remember that maybe the defects are there, but you need to improve your own knowledge and skills to find it. Be always suspicious, don't trust it blindly.

Re: Ask HN: Anyone struggling to get value out of coding LLMs?

#62
I am struggling to programmatically get syntactically valid JSON out of LLMs, using both the OpenAI and Vertex apis.

I am using:

  "response_format": { "type": "json_object" }
And with Vertex:

    "generationConfig": {
      "responseMimeType": "application/json"
    }
And even:

  "response_format": {
    "type": "json_schema",
    "json_schema": { ...
And with Vertex:

    "generationConfig": {
      "responseMimeType": "application/json",
      "responseSchema": { ...

Neither of them is reliable.

It always gives me json in the format of a markup document with a single json code block:

    ```json
    {}
    ```
Sure I can strip the code fence, but it's mighty suspicious I asked for json and got markup.

I am getting a huge number of json syntax errors, so it's not even getting to the schemas.

When i did get to the schemas, it was occasionally leaving out fields that I'd declared were required (even if i.e. null or an empty array). So I had to mark them as not required, since the strict schema wasn't guiding it to produce correct output, just catching it when it did.

I admit I'm challenging it by asking it to produce json that contains big strings of markup, which might even contain code blocks with nested json.

If that's a problem, I'll refactor how I send it prompts so it doesn't nest different types.

But that's not easy or efficient, because I need it to return both json and markup in one call, so if I want to use "responseMimeType": "application/json" and "responseSchema", then it can ONLY be json, and the markup NEEDS to be embedded in the json, not the other way around, and there's no way to return both while still getting json and schema validation. I'd hate to have to use tool calls as "out parameters".

But I'm still getting a lot of json parsing problems and schema validation problems that aren't related to nested json formatting.

Are other people regularly seeing markup json code blocks around what's supposed to be pure json, and getting a lot of json parsing and schema validation issues?

Re: Ask HN: Anyone struggling to get value out of coding LLMs?

#63
post #44

Earlier quoted context omitted.

> So I ask ChatGPT and one of the things it said is that historically it meant "measure" This is a great example of why using LLMs to learn is perilous. This is nonsense.

Is it? I didn't bother to check as it wasn't too relevant. I wasn't stuck anymore. Worst-case it's a memnonic that is untrue. Best case: it is. For actual derivations I check the work. This is what it wrote by the way. I posted an image and asked: Why is this m m for angle mangle? ChatGPT: In geometry we often need to distinguish the angle itself (a geometric object) from the numerical size of that angle (a real-numb…

See slope: https://en.wikipedia.org/wiki/Slope

It isn't customarily used for angles (those get Greek letters).

The m stands for mystery.

Edit: ah, but I see that this prefixed m for "measure" is also used sometimes. It appears at https://en.wikipedia.org/wiki/Angle#Combining_angle_pairs though I'm not sure why it's necessary. Maybe because you want to clarify absolute magnitude is meant, to avoid adding negative values.

Re: Ask HN: Anyone struggling to get value out of coding LLMs?

#64
post #56
post #13

There are two kinds of engineers. Those who can’t stop raving about how much of a superpower LLMs are for coding, how it’s made them 100x more productive, and is unlocking things they could’ve never done before. And those who, like you, find it to be an extremely finicky process that requires extreme amount of coddling to get average results at best. The only thing I don’t understand is why people from the former gro…

You’ll usually find three kinds of people talking about LLMs online: 1. The ones writing Medium posts about being 100x more productive. 2. The skeptics, like yourself, who construct made-up examples to prove it’s impossible to get value, framing things in a way where failure is the only plausible outcome. 3. And then there’s the middle group, people quietly getting real, uneven, but worthwhile value. You’d be surpris…

GP didn't "construct made-up examples" anywhere in this thread as of the time of writing this post. Where did you hallucinate that from?

Seriously, why is every comment hyping up AI generated code like this. No concrete evidence, just turtles all the way down. But when actual examples do show up, it's a complete clown show [1][2].

[1]: https://news.ycombinator.com/item?id=44050152

[2]: https://news.ycombinator.com/item?id=43907376

Re: Ask HN: Anyone struggling to get value out of coding LLMs?

#65
I've spent the past week overcoming my fear of Google's Gemini and OpenAI's ChatGPT. Things I've learned:

- Using an AI for strange tasks like using a TTS model to turn snippets of IPA text (for a constructed language) into an audio file (via CLI) - much of the task turned out to be setting up stuff. Gemini was not very good when it came to giving me instructions for doing things in the GCP and Google Workspace browser consoles. ChatGPT was much clearer with instructions for setting up AWS CLI locally and navigating the AWS browser console to create dedicated user for the task etc. The final audio results were mixed, but then that's what you get when trying to beat a commercial TTS AI to doing something it really thinks you're mad to try.

- Working with ChatGPT to interrogate a Javascript library to produce a markdown file summarising the library's functionality and usage, to save me the time of repeating the exercise with LLMs during future sessions. Sadly the exercise didn't help solve the truly useless code LLMs generate when using the library ... but it's a start.

- LLMs are surprisingly good at massaging my ego - once I learned how to first instruct them to take on a given persona before performing a task:

I still fear LLMs, but now I fear them a little less ...

Re: Ask HN: Anyone struggling to get value out of coding LLMs?

#66
post #55

Earlier quoted context omitted.

I find LLMs 100x more productive for greenfield work. If I want to create a React app with X amount of pages, some Redux stores, Auth, etc. then it can smash that out in minutes. I can say "now add X" and it'll do it. Generally with good results. But when it comes to maintaining existing systems, or adding more complicated features, or needing to know business domain details, a LLM is usually not that great for me. T…

> If I want to create a React app with X amount of pages, some Redux stores, Auth, etc. then it can smash that out in minutes. I can say "now add X" and it'll do it. Generally with good results. Not discounting your experience, but a lot of these examples are about frameworks that never had good bootstrapping, such as Rails does/did. LLMs are really good at boilerplate, but maybe this points to these such stacks need…

You can get similar results by downloading a boilerplate template project for that though.

LLM makes that a bit easier and faster to do, but is also a bit more error prone than a template project.

Re: Ask HN: Anyone struggling to get value out of coding LLMs?

#67
post #60
post #33

Earlier quoted context omitted.

Would you be so kind as to link to some of these sophisticated products? Would love to check them out.

Go look at the latest YC batch. 30% of them were built with 90% of the code being written by LLMs.

Where did you see that stat? And why just 30%, shouldn't almost all of them be built by LLM if LLM is that much better?

Edit: This says a quarter, where did you get the 30% figure from?

https://techcrunch.com/2025/03/06/a-quarter-of-startups-in-y...

Re: Ask HN: Anyone struggling to get value out of coding LLMs?

#68
post #66
post #55

Earlier quoted context omitted.

> If I want to create a React app with X amount of pages, some Redux stores, Auth, etc. then it can smash that out in minutes. I can say "now add X" and it'll do it. Generally with good results. Not discounting your experience, but a lot of these examples are about frameworks that never had good bootstrapping, such as Rails does/did. LLMs are really good at boilerplate, but maybe this points to these such stacks need…

You can get similar results by downloading a boilerplate template project for that though. LLM makes that a bit easier and faster to do, but is also a bit more error prone than a template project.

I’m not sure that LLM makes it easier to do. Pain points I’ve seen:

1. You have to remember all the technologies you need included, my company template already has them.

2. LLM doesn’t have a standardized directory structure, so you end up with different projects having different structures and file naming conventions. This makes later refactoring or upgrades across multiple projects less automatable (sometimes this can be solved by having an LLM do those, but they often are unsuccessful in some projects still)

3. LLMs have a knowledge cutoff. If your company has already moved to a version after that knowledge cutoff, you need to upgrade the LLM generated code.

I very much prefer having a company template to asking an LLM to generate the initial project.

Re: Ask HN: Anyone struggling to get value out of coding LLMs?

#69
> I asked friends who are enthusiastic vibe coders and they basically said "your standards are too high".

Sure, vibe coders by definition can't have any standards for the code they're generating because by definition they never look at it.

> Is the model for success here that you just say "I don't care about code quality because I don't have to maintain it because I will use LLMs for that too?"

Vibe coding may work for some purposes, but if it were currently a successful strategy in all cases, or even narrowly for improving AI, Google AI or DeepSeek or somebody would be improving their product far faster than mere humans could, by virtue of having more budget for GPUs and TPUs than you do, and more advanced AI models, too. If and when this happens you should not expect to find out by your job getting easier; rather, you'll be watching the news and extremely unexpected things will be happening. You won't find out that they were caused by AI until later, if ever.

Re: Ask HN: Anyone struggling to get value out of coding LLMs?

#70
post #13

There are two kinds of engineers. Those who can’t stop raving about how much of a superpower LLMs are for coding, how it’s made them 100x more productive, and is unlocking things they could’ve never done before. And those who, like you, find it to be an extremely finicky process that requires extreme amount of coddling to get average results at best. The only thing I don’t understand is why people from the former gro…

I find LLMs 100x more productive for greenfield work. If I want to create a React app with X amount of pages, some Redux stores, Auth, etc. then it can smash that out in minutes. I can say "now add X" and it'll do it. Generally with good results. But when it comes to maintaining existing systems, or adding more complicated features, or needing to know business domain details, a LLM is usually not that great for me. T…

I've recently been able to use LLM on a large-ish internal project to find a bug. The prompt took the form of "here's the symptoms I observe, and some hypothesis, tell me where the code that handles this case is written" (it was a brand new repo that I hadn't looked at before - code written by a different team, that were claiming some weird race condition/ were not really willing to look into the bug). Basically I was asking the LLM to tell me where to look, and it actually found the issue itself.

Not 100x more productive, that's an exaggeration... not even 10x. But it helps. It is an extremely competent rubber duck [1].

[1] https://en.wikipedia.org/wiki/Rubber_duck_debugging

Post reply on HN