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

Did GitHub Copilot increase my productivity?

trace.yshui.dev

Post reply on HN
Print view View original

61–70 of 326 posts

Re: Did GitHub Copilot increase my productivity?

#61

Post by eru »

noduerme wrote: post #25
aimazon wrote: post #16

Earlier quoted context omitted.

I don’t think you’re a dimwit but I read your post[1] with an example and I am curious to whether you feel you’re losing something by telling the LLM it’s wrong and to try again, rather than going through the exploratory/iterative learning process yourself. For example, would you have known to ask about GeoJSON if you had not come across and learned about it pre-LLM? More succinctly: do you feel you’re learning more…

You seem to be writing from a lost world when people enjoyed coding because they enjoyed learning new things with every project. I think that's sadly something only enjoyed at work by a very small minority these days.

I don't think those 'good old days' ever existed. Or rather, the situation hasn't really gotten worse.

Re: Did GitHub Copilot increase my productivity?

#62

Post by yawaramin »

simonw wrote: post #37
aimazon wrote: post #16

Earlier quoted context omitted.

I don’t think you’re a dimwit but I read your post[1] with an example and I am curious to whether you feel you’re losing something by telling the LLM it’s wrong and to try again, rather than going through the exploratory/iterative learning process yourself. For example, would you have known to ask about GeoJSON if you had not come across and learned about it pre-LLM? More succinctly: do you feel you’re learning more…

Absolutely I could have learned more from that particular project if I'd spent more time with it rather than getting the LLM to do the work... but that's why I like it as an example: since it was effectively a distraction (a "side quest") the alternative wasn't learning more, it was not doing it at all (and learning nothing). I'm able to get really great results out of LLMs because I have 20+ years of experience help…

> I have 20+ years of experience helping me know what questions to ask of them.

That seems rather unlikely given that LLMs didn't exist more than seven years ago, much less 20+.

Re: Did GitHub Copilot increase my productivity?

#63

Post by eru »

blibble wrote: post #26
simonw wrote: post #8

Earlier quoted context omitted.

I've been finding this stuff genuinely useful for two years now, across Copilot and ChatGPT and Claude 3 Opus and similar tools. Either I'm a dimwit, easily conned by hype and shiny tools to the point that I can imagine benefits for two years that simply aren't there... or there's something to them.

I wouldn't feel too bad, there's orders more magnitude more money put into this con vs. the last one (NFTs)

Your counting of 'cons' seems pretty idiosyncratic.

You know there have been lots of other technologies people have been working on in the meantime and concurrently? Some of them more questionable, some of them less.

Re: Did GitHub Copilot increase my productivity?

#64

Post by eru »

tylerchilds wrote: post #57
noduerme wrote: post #21

Earlier quoted context omitted.

I think the nomenclature of calling generative AI an "AI" is just hype and that leads to disappointment. After all, it's much more exciting for investors than calling it the world's most expensive textual/audio/visual autocomplete. But that's really what this generatoon currently is. And it can feel like magic, because we're pattern-seeking and pattern-matching creatures, so something that seems to intuit the pattern…

“how much will you pay for a needed autocomplete to do X?” i think this is a super important for anyone playing in the LLM space to calculate. currently “AI” providers are selling electricity at a loss to demonstrate product “value”, so even while asking questions is “free” today, there’s actually a finite resource under the hood that needs to have the bill paid in the end. an estimate, but from historic trends, free…

Even _if_ we don't see our current 'AIs' become more intelligent, I am very certain that we will see the amount of electricity needed to produce 2024-state-of-the-art results drop dramatically over the next few years. People are only just figuring out how any of this works, and they are happy to get any reasonable results at all. They aren't really competing on costs like energy efficiency, yet.

---

However I don't really think that 'if' will come to pass: I expect that we will still see lots of advances in the 'intelligence' of these models, so people will compete on these, and worry about electricity usage afterwards.

Re: Did GitHub Copilot increase my productivity?

#65

Post by eru »

mhh__ wrote: post #11
AnarchismIsCool wrote: post #2

Local maxima go BRRRRRRR AI will get there eventually, but this current paradigm seems increasingly only useful for spam and shitty clip art. Even so, everyone is throwing absurd amounts of investment capital at it in the hopes that something useful will happen. It's a pretty clear depiction of the investor class being so detached from the technical reality of what they're investing in that they just sit around light…

> spam The first 100 lines of any project is spam (usually 0 entropy), so AI is a nice win.

That depends on your language, and project.

Re: Did GitHub Copilot increase my productivity?

#66

Post by eru »

leptons wrote: post #50
amanzi wrote: post #43

I feel like this bit is important for this discussion: "...I do not use Copilot for my day job. I use it for my own projects only..." My guess is the people that don't use Copilot because they don't trust it, or don't like the code it produces, are probably trying to use it in a professional scenario where they have strict coding standards. For hobbyist use or side-projects, Copilot is an incredible time saver. I use…

I only use Copilot for my day job, because they paid for it. When I'm coding my personal projects on nights and weekends I find myself missing Copilot and I'm contemplating paying for it myself. I don't really trust it for everything, but sometimes it surprises me with how much it gets right. I just can't justify paying $19/month personally for the "Business" plan so they don't use my proprietary code to train their…

> I just can't justify paying $19/month personally [...]

Why not? Is it a significant fraction of your income? I find that the cost of hardware deprecation, electricity and internet access probably amounts to a lot more than that per month for me already. (Not to mention the opportunity cost of my time.)

It's fine to be frugal, but it's also useful to compare your actual costs.

(Though that's not always appreciated. I got into some minor trouble at an employer when I pointed out that the 15 GiB of storage space they 'graciously' gave us should cost them less than the company toilet paper we use every day. And I said 'should' because that's internal storage space, and I used numbers from AWS. But if they are paying significantly more for storage than what AWS charges (with enormous margins for Amazon), we also have a big problem.)

Re: Did GitHub Copilot increase my productivity?

#67

Post by skydhash »

Seb-C wrote: post #45

I still haven't found any interest in this wave of AIs for coding, mostly because writing code hasn't been a bottleneck for me since I learned vim and started using it as my only tool. Knowing what to write and how to organize the code is the really difficult part of our job, and I am unable to do this correctly when reviewing stuff written by others (or by an AI). I have been wondering if people are actually using t…

Same. Most of my time is spent thinking about the domain space, constraints and processing flows. Refreshing my memory about a method, a class or a specific how to is fast based on a combination of Dash (offline docs), PDF Manuals, and opening the libraries pages in advance. And with Vim, I can copy-paste code while having complete control over the code. And thanks to the buffer model, I can have everything I'm thinking about in front of me instead of one single file (VSCode Tab model). Boilerplate has never been a bottleneck.

And I like to have an idea of everything I have to manipulate, variables and property, functions and classes, modules and files. If something has not been written by me, I have some good reasons to trust it, either because I reviewed its source code or the library reputation (I'd probably read the source code anyway because some documentations are not good). IDEs have been able to provide the kind of completion that I like (auto-importation, symbols) and information I want (function signature and class properties).

I'd rather take a moment to read a library's documentation and code examples than see ChatGPT generates something that has a non-zero chance of being an hallucination.

Re: Did GitHub Copilot increase my productivity?

#68

Post by skydhash »

eru wrote: post #60
heyoni wrote: post #52

Earlier quoted context omitted.

Especially when renaming variables that are “immune” to normal refactoring. Copilot handles that pretty well and I don’t have to spend all that focus on such a menial task.

Yes, I often find that copilot is pretty good at picking up on the pattern of refactorings that I am doing. Especially those that are a bit tedious and almost mechanical, but not quite mechanical enough to do with a simple search-and-replace.

Vim and text motions should do the trick.

Re: Did GitHub Copilot increase my productivity?

#69

Post by fzeroracer »

matrix87 wrote: post #23

I don't really get the value proposition, for every second you spend writing something you spend another 10 verifying that it was done properly. Speeding up the process of writing is sort of a waste because of Amdahl's law How does copilot et al. speed up verification and review? For example, a C program is harder to verify than a Go program doing the same thing. Suppose you use it to verify something in an enterpris…

This is where I stand. I think people fall into the trap of believing that it is reliable and useful because it can occasionally spit out boilerplate code that matches what boilerplate code is supposed to look like.

Which is a very dangerous place to be in, because the more you assume that the code generated is reliable the more likely hallucinations are to sneak into your codebase. And now you don't have a good mental model of what your code is doing because you didn't actually write any of it.

I have a feeling that codebases using and relying on copilot are, essentially, pushing disasters later down the line when bugs pop up and no one has the ability to actually debug the system.

Re: Did GitHub Copilot increase my productivity?

#70

Post by Loocid »

yawaramin wrote: post #62
simonw wrote: post #37

Earlier quoted context omitted.

Absolutely I could have learned more from that particular project if I'd spent more time with it rather than getting the LLM to do the work... but that's why I like it as an example: since it was effectively a distraction (a "side quest") the alternative wasn't learning more, it was not doing it at all (and learning nothing). I'm able to get really great results out of LLMs because I have 20+ years of experience help…

> I have 20+ years of experience helping me know what questions to ask of them. That seems rather unlikely given that LLMs didn't exist more than seven years ago, much less 20+.

They mean they have 20 years experience in programming, which helps them ask good questions.
Post reply on HN