Review the code yourself, tell it what to fix, and it will fix it. For me it often takes like 5 rounds of this before the code is fully polished to the point I’m proud of it. And you know what? It’s still MULTIPLE times faster than typing code by hand. And at least for me, the quality is higher, because I have 12 different agents that review the work too and they catch additional issues that even I missed.
If you or others ship shitty code don’t blame the LLM, the issue is entirely the engineer using it wrong
I told my boss I wouldn’t use AI to help write code and sign off on the commits with my name. He told me to use it to write the documentation… it was bad. Essential concepts were not mentioned and obvious shit was explained five times in slightly different phrasing. I am now writing our documentation as an obsidian vault myself again.
I only have it compile a change log from the commit messages of commits and merges on the main and development branch. I know our commits are well written (because I established the standards for them in our repo myself) and that’s concrete and rigid enough that it can’t fuck it up enough to matter.
But honestly there are build tools that could do that for me, I don’t need to buy tokens for that.
Not just beginners unfortunately. One of the things I have to almost always teach to new colleagues, even experienced ones, is to put at least some minimal effort into making the code itself readable instead of relying on comments as crutches. Just basic things like picking proper names for variables and functions.
You can add instructions to not comment, you can also have it explain what it does at every step, not everyone just doesn’t care about learning. It can be a very effective teaching tool if you use it that way. 🤷
I’m definitely still a noob but I’ve done hundreds of hours of debugging on code in the past few months, and my job for the past 8 years is basically to troubleshoot issues, though the past year I got to start doing devops/code work on the side.
Then you should know that code not working is the absolute easiest fuckup to catch. It’s literally not one to be concerned about.
One in a million chance of an edge case that doesn’t throw an error at all, but does something unexpected? Good luck if you don’t know how the system works.
Agreed it’s why if I’m asking for llm assistance I will generally start with a design for a component and ask it to follow that and comment accordingly usually leads to much better results than blanket asking it to do something for you
I’ve seen the comments AI adds and yeah… No… It’s often pointing out the obvious or even in some cases just misleading.
Out of the box yes this is true, but:
If you or others ship shitty code don’t blame the LLM, the issue is entirely the engineer using it wrong
Yea the comments are awful. More useless than dog shit.
Skill issue tbh
I told my boss I wouldn’t use AI to help write code and sign off on the commits with my name. He told me to use it to write the documentation… it was bad. Essential concepts were not mentioned and obvious shit was explained five times in slightly different phrasing. I am now writing our documentation as an obsidian vault myself again.
I only have it compile a change log from the commit messages of commits and merges on the main and development branch. I know our commits are well written (because I established the standards for them in our repo myself) and that’s concrete and rigid enough that it can’t fuck it up enough to matter.
But honestly there are build tools that could do that for me, I don’t need to buy tokens for that.
Yes. This is where every “have you tried Agentic AI?” conservation lands, for me.
They tell me I could pay daily, for worse results, to give up the bash script that has worked perfectly for five years. Oh, gee. Tell me more! Haha.
Sounds like what a lot of beginner programmers do as well.
Not just beginners unfortunately. One of the things I have to almost always teach to new colleagues, even experienced ones, is to put at least some minimal effort into making the code itself readable instead of relying on comments as crutches. Just basic things like picking proper names for variables and functions.
a = aa + aaa .* a3(A(aa).a2);
//assign a as a function of aa,aaa,A,a3, and a2
Nailed it
What the fuck are doing here agent? Go tenet those glass towers already.
I had to tell the ai to reduce the comments it made. Since it was bunch of not important information.
You can add instructions to not comment, you can also have it explain what it does at every step, not everyone just doesn’t care about learning. It can be a very effective teaching tool if you use it that way. 🤷
AI is a terrible way to learn something. It will do something wrong, explain it incorrectly, and you will have no idea.
AI is only useful if you are able to spot and correct the mistakes it makes. Because it will make mistakes.
Very effective teaching tools already exist if you want to learn.
“You’ll have no idea” until it doesn’t work lmfao
Someone hasn’t done enough debugging in their life. I wish the lesson be as painless as possible
I’m definitely still a noob but I’ve done hundreds of hours of debugging on code in the past few months, and my job for the past 8 years is basically to troubleshoot issues, though the past year I got to start doing devops/code work on the side.
Its fine though, I get why you guys are scared.
Then you should know that code not working is the absolute easiest fuckup to catch. It’s literally not one to be concerned about.
One in a million chance of an edge case that doesn’t throw an error at all, but does something unexpected? Good luck if you don’t know how the system works.
All that and you still think that “doesn’t work” is the only problem that can happen?
When a product is so absolutely terrible that uncommented code is the best it can do…
Y’all are delusional
Agreed it’s why if I’m asking for llm assistance I will generally start with a design for a component and ask it to follow that and comment accordingly usually leads to much better results than blanket asking it to do something for you
So… indistinguishable from the statistical mean of human generated comments.
There is one exceedingly important difference: who gets paid.