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

GitHub Copilot

copilot.github.com

Post reply on HN
Print view View original

961–970 of 1001 posts

Re: GitHub Copilot

#961

Post by rantwasp »

machiaweliczny wrote: post #917
rantwasp wrote: post #841

i’m sort of a Luddite so take this with a grain of salt. I don’t see this going anywhere. If it’s really good and we can actually have 1 developer instead of 2 now why would any developer want to do this? this would basically be a piece of automation that diminishes the value we are creating. If it’s crap it’s going to create mountains of verbosity and code written to pump up the LOC numbers. It’s terrible. After tha…

As long as supply > demand then 1 programmer able to do job for 2 equals higher salary. Obviously supply shortage will end though, sooner or later.

I don’t necessarily agree with you that if supply > demand and 1 programmer does the job of 2 this will lead to a higher salary.

You can see this today in most places where you can walk on water, but you still need to be in your pay band.

Re: GitHub Copilot

#962

Post by ganzuul »

rictic wrote: post #779
billti wrote: post #688

Earlier quoted context omitted.

The landing page for it states the below, so hopefully not too much of an issue (though I guess some folks may find a 0.1% risk high). > GitHub Copilot is a code synthesizer, not a search engine: the vast majority of the code that it suggests is uniquely generated and has never been seen before. We found that about 0.1% of the time, the suggestion may contain some snippets that are verbatim from the training set.

If you pulled a marble out of a bag ten times a day, with a 0.1% chance each time that it was red: after a day you'd have a 1% chance of seeing a red marble, the first week you'd have a 6.7% chance, the first month you'd have a 26% chance, and the first working year you'd have a 92.6% chance of having seen at least one red marble. Probabilities are fun!

Well within the margin of fair use.

Re: GitHub Copilot

#963

Post by rvz »

rvz wrote: post #903
reilly3000 wrote: post #229

Earlier quoted context omitted.

Right. If you’re comfortable giving access to your source files to GitHub+OpenAI, then go for it. I’m not sure how this would apply to secret keys or flat files with customer data/PII, but in any case that makes it a non-starter for me. Their “Please do not share this URL publicly.” Banner at the top of the page which disclosed this info makes my skin crawl a bit… If I were only working on public projects I would be…

> Am I being to paranoid here? No. They already admitted that they send telemetry of the code you give it and its training set already has personal information in it anyway, despite what is being hyped up here by the fanatics, even when someone said that 'Copilot guesses the exact code I want to write about one in ten times' [0] No thanks and certainly no deal. [0] https://news.ycombinator.com/item?id=27676845

So am I being too paranoid here to say that a bot (or something) somewhere on HN is instantly downvoting my very good questions and substantiated claims?

I always ask whoever disagrees to have the courage to sit down and discuss, but they always run away and never explain themselves.

Look really suspicious of either bot behaviour or just some angry hater don't you think?

Re: GitHub Copilot

#965

Post by adn »

visarga wrote: post #631
6gvONxR4sf7o wrote: post #235

Earlier quoted context omitted.

> Yet, isn’t that effectively what human brains do too? If I want to watch a bunch of movies, I have to pay the theater for each movie, or pay netflix, or whatever. The screenplay I write afterwards belongs to me, but the learning process involved me paying for access to others' work. That's what's often missing here. But at the same time, if you train on legally public data, there's no 'theater' to be paid. (Often,…

> Maybe we'll start seeing licenses with a section saying "not for use as training data for commercial models." Considering that the impact of a single example is extremely small in training a model, and that it is trained on an ungodly amount of examples, then I wonder if the effort of forbidding its use has any real benefits.

Yes, of course it does, because if every user opted out then the model would not work as well as it does, and github would not be able to profit off the work of others to the degree they are (or will be). Just because they are taking code on a massive scale does not mean the outcome is inevitable: don't get it twisted, copilot only works because of the code human beings have written.

Re: GitHub Copilot

#966

Post by ganzuul »

chimtim wrote: post #738
fzaninotto wrote: post #40

I've been using the alpha for the past 2 weeks, and I'm blown away. Copilot guesses the exact code I want to write about one in ten times, and the rest of the time it suggests something rather good, or completely off. But when it guesses right, it feels like it's reading my mind. It's really like pair programming, even though I'm coding alone. I have a better understanding of my own code, and I tend to give better na…

Unfortunately, one in 10 times is far from good enough (and this is with good prompt engineering which after using large language models for a while, one starts to do). I feel like the current generation of AI is bringing us close enough to something that works once in a while but requires constant human expertise ~50% of the time. The self-driving industry is in a similar situation of despair where millions have bee…

I think 1/10 is incredible. If it holds up it means they may have found the right prior for a path of development that can actually lead to artificial general intelligence. With exponential improvement; humans learning to hack the AI and the AI learning better suggestions, this may in theory happen very quickly.

We live in a very small corner of the space of possible universes, which is why finding a prior in program space within it is a big deal.

Re: GitHub Copilot

#967

Post by adn »

bootlooped wrote: post #663
mihaifm wrote: post #637

I can see in their FAQ that there are plans to monetize this. So they're building an ML model that feeds on the work done by millions of people and then selling it? How is this even ethical? Not to mention we'd be feeding the model while using it. Guess this is another instance where we are becoming the product.

If I spent a lot of time reading open source repos on GitHub to teach myself to code, and then went out and got a high-paying job based on that knowledge, is that ethical? This seems roughly analogous to what the machine is doing.

Regardless of the legality, one of these situations is clearly ethical compared to the other. In the case were you get a job based on your knowledge of GPL software, you still must respect the license if you use that code commercially (i.e. at your new job). And yes, if you reproduce GPL code you "learned" from, you are violating the license.

A company ingesting an entire GPL codebase without warning or any way to opt-out in order to create a closed-source feature that they and only they will profit off of is clearly not the same as an individual reading the code and getting a job based on those ideas.

Re: GitHub Copilot

#970

Post by toastal »

hnarn wrote: post #423

I'm sorry if I'm derailing the discussion here but "copilot" really is a much better phrase to use for assisting software than "autopilot". If more companies would choose phrases like these that accurately emphasize that the human is not being replaced but assisted (no names mentioned) I think it would benefit everyone in terms of clarity. Sure, you might say "it's all marketing and if AP exists nobody would buy CP",…

Copilot is to make you feel like you're the one in control, when in reality you, the user, are training it for free.
Post reply on HN