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

Ask HN: What hacks/tips do you use to make AI work better for you?

news.ycombinator.com

Post reply on HN
Print view View original

81–90 of 108 posts

Re: Ask HN: What hacks/tips do you use to make AI work better for you?

#81

Post by handzhiev »

wruza wrote: post #74
fragmede wrote: post #44

Earlier quoted context omitted.

I’m exhausted by these types of posts. They never include concrete details on what they're trying to do. what languages they're using, what frameworks, which LLM. They occasional state which tool, but then don't go into detail how they're using it. There's never any links to chats/sessions showing the prompts they're giving it and the answers they're finding so unacceptable. Imagine if you got bug reports from custom…

No one includes complete detail at saying it’s useful and life-changing too, so that’s fair. It might turn out that what works for those for whom it works is trivial “code” not worth the ssd blocks it occupies. This is actually my current theory, cause LLMs (all of them, yes we tried all of them) are capable of what I tend to not think about as programming but as industry nonsense which should have been automated/abs…

"No one includes complete detail at saying it’s useful and life-changing too"

There are at least 3 posts in this very discussion sharing details and githup repos with code written mostly by LLM.

Re: Ask HN: What hacks/tips do you use to make AI work better for you?

#82

Post by emgeee »

yen223 wrote: post #2

I like using LLMs for building non-critical tools that make me more productive. Things like shell scripts, Github actions, or one-off tools for visualising some problem space. The kind of thing where code quality doesn't really matter, but which will save you a ton of time in the long run

I've found them incredibly useful for writing Dockerfiles or other bits of infra config like K8s yaml

Re: Ask HN: What hacks/tips do you use to make AI work better for you?

#83

Post by emgeee »

Commenting code and generating documentation.

I like to copy entire python modules into the context window and say something like "add docstrings to all methods, classes, and functions".

You can then feed the code into something like sphinx or pdoc to get a nice webpage.

Re: Ask HN: What hacks/tips do you use to make AI work better for you?

#84

Post by flykespice »

andrewinardeer wrote: post #75
Sateeshm wrote: post #66

Earlier quoted context omitted.

Are you confident that it's getting them right everytime?

No. I'm not confident they are getting them right everytime. Human is in the loop.

What is the rate of miss?

Re: Ask HN: What hacks/tips do you use to make AI work better for you?

#86

Post by paulddraper »

joblemblem wrote: post #11

I’m exhausted by these types of posts. I am a developer with >25 of professional experience. I am unable to get these things to do anything useful. I’ve tried: different models, limiting my scope, breaking it down to small tasks, prompt “engineering”; and am still getting less than useless results. I say less than useless, because I will then additionally waste time debugging or slamming my head against the wall the…

Sorry.

Hope the other posts here help you.

Re: Ask HN: What hacks/tips do you use to make AI work better for you?

#87

Post by miningape »

eternityforest wrote: post #36

If AI doesn't understand something I assume it's too clever, even if I do get the AI to figure it out I won't understand it in a month and other devs won't understand it at all unless I somehow convince them to read the code(Like as if that's going to happen!). I think it's definitely improved my own code, because it's like always working with a pair programmer, who represents a very average developer with insane typ…

> unless I somehow convince them to read the code(Like as if that's going to happen!).

code is read more often than its written. I spend maybe 60-70% of my time reading code when I'm "writing code". I don't think it will take much convincing if you're working on a project with others.

Re: Ask HN: What hacks/tips do you use to make AI work better for you?

#89

Post by screye »

Not a hack as much as a PSA :

There is no shame in dumping your whole codebase into claude. Use any tool (aidigest, Simon's tools). It just works. Especially useful for complex code bases.

Same applies for RAG. Unless you're damn sure what you're looking for, don't load that 1 chunk in. If your model can fit the full context or 50 chunks in the token window, dump it all in. Sophisticated RAG works, but like fine tuning, it usually makes things worse before they start getting better.

Speaking of finetuning. Same PSA. Dump as many fewshot examples as you can in the token window before restoring to finetuning.

Re: Ask HN: What hacks/tips do you use to make AI work better for you?

#90

Post by OnionBlender »

I've been struggling to make use of these tools for C++ projects. My boss keeps asking me to try using these tools to generate documentation or unit tests, but the results are pretty worthless. At best, the comments it generates are the low value kind that should be obvious from the function or parameter names. The kind of redudent comments that people would write if they are required to write a comment. At worst, it straight up lies about what the function does.

The code it writes reminds me of the kind of code my former Java co-workers would write when our company switched to C++.

I find these tools okay for creating simple python scripts or showing me how to do something in Powershell or bash.

Post reply on HN