2. Copy&Pasting Code by manual search exists.
3. This is just a smart tool so you don't have to figure out yourself what to copy&paste (in the best case) and save a lot of time.
Sometimes I truly wonder how people can genuinely be upset about things like this. What is broken are copyright and patent laws in the 21st century.
I guess Copilot could address this by checking the licenses of the projects it uses. Even when combining code, it could pull in the required attribution or avoid GPL licensed code (unless enabled) for example.
Tell me you regularly plagiarize without telling me you regularly plagiarize.
Of course it is. Plagiarism is “the practice of taking someone else's work or ideas and passing them off as one's own.”
It’s unethical and it will get you fired at any reputable company.
I’m deeply disturbed that you think this form of plagiarism is universal — I can assure you that is not the case.
I work at a FANG currently, and plagiarism is absolutely not tolerated.
In fact, plagiarism has been considered a fireable offense at every other company I’ve worked at over my 25 year long career, and prior to that, considered a serious form of academic misconduct in school.
It’s clearly unethical and I’ve never plagiarized in my life.
I’ve only run into one instance of someone else plagiarizing code in my career, and that individual was fired.
I feel like you're arguing in bad faith. So, whatever.
SO’s license requires attribution.
If you don’t want to be a plagiarist, you either need to include attribution, or you need to rewrite the solution in entirely your own words.
> You should write code sometimes to solve a complex problem
There’s a very high chance you posted your comment from a device using code I wrote.
Glueing together plagiarized code copied from SO, or stolen from OSS projects on GitHub, is not software engineering.
This thread is an eye opener for me too. Do engineers not get trained on their legal obligations? My company is old and not a tradition tech company but we have been running workshops on the issue for years. Even if they don't, what about their legal teams? Or CI tools to scan for licence violations? Some of the responses here are so naive it's crazy. I hope no one is identifying the companies they work for.
But that's something entirely different from small code snippets, changed and adapted to solve the same problem a thousand other people already had. Nothing else are developers doing going on GitHub, StackOverflow or any other website to find answers to their questions. That's not naivety, that's how coding works (partially). If you would have to re-invent the wheel everytime you build something new, good luck.
Snippets aren't exactly defined but I see them as more than just a single line like "here's how to flatten a list in Python", it's some functionality - e.g. an algorithm implementation or some task.
It's universal in any company that doesn't take measures against it. So basically startups, small, medium, and even some large companies.
Code Plagiarism is taken very seriously by every company I have worked with. Multiple companies have been sued for violating the GPL. The SFC is currently fighting Vizio in court for example. While not commonplace, to say it's "almost impossible" is a stretch. Every large company complies with code copyright obligations for a reason. My company publishes changes to GCC and a dozen other GPL projects. Entire products like Protocode and BlackDuck exist to ensure code compliance. Even small code snippets are flagged.
Over the past few years the source code for Windows, SQL server, Bing and Cortana have all been leaked. If someone built a product using that code, how long do you think it would take Microsoft to sue? CoPilot is one rule for mega-corps and another for everyone else.
Now you might say, what about GPL code you don't own. You are allowed to redistribute it (upload to github). But because you are not the owner you can't license it to Github under new terms (that allow them to use it for ML training). But the question still is, is there anything in the GPL that forbids it's code being used for ML training? Even if the generated model is proprietary, has no attributions, etc?
If not, then it's kinda stupid to argue the point about the lack of knowledge, since lack or not lack of knowledge clearly doesn't matter. Github will happily continue using confidential code even from trigger happy companies like Qualcomm for copilot.
It's kind of like some employee that worked at Qualcomm and has seen the code. Do you retrain him (aka hit his head until he forgets) after leaving the company?
The comparison might seem silly but as AI advances I expect more and more arguments (especially in court) to come from analogies of humans and AIs.
Creating an output filter based on millions lines of DMCAd code that would not cripple the copilot output completely at the same time, sounds like one of those hard problems. Especially if there's no agreed upon definition of copyright "violation" here.
Are you saying that I would need all the original authors consent to upload a repo to github even if I include all the original attribution and licenses? Because what you are implying is that when uploading I'm granting github a license far outside the bounds of the license included, which only all the contributors can do. For example, would the linux project need to contact each and every contributor ever to upload a mirror to github, since their contributions were under GPL but you are implying that the license given to github is much, much broader?
This would make any project not originally started on github and with a few contributors basically impossible to host there.
> 2. Copy&Pasting Code by manual search exists.
The question is who is doing the infringement here. Github copilot is obfuscating the copying and telling it's users that the code is theirs to use, own, etc. as they please but is also taking large chunks of code it does not have the right to redistribute, even less grant licenses to.
90% of Twitter is just inventing new ways to whine about things