651–660 of 751 posts
Sorry to ask a shallow question. His "photo" is so interesting. It feels exactly like old school Wall Street Journal "photos" from 1990s. Is there a plug-in or service to create this type of image from a photograph?
1 - https://www.wsj.com/articles/whats-in-a-hedcut-depends-how-i...
Oh god please no. GitHub Copilot is a wonderful technology. I am not taking anything away from you if Copilot suggests code that is similar or identical to your copyrighted code. You were not going to sell it to me anyway. The following is supposed to be OK: somebody reads your GPLed code, learns abstract concepts from it, teaches it to me, I write code that uses the same algorithm. But it's not OK to abbreviate the…
The trouble is that this apparently is not what Copilot is always doing. If it had only "learned abstract concepts" from GPL'd (or any other form of copyright) code, then that would not be a problem, and of course that is kind-of what Copilot purports to be doing, supposedly learning the association between concepts described in comments and corresponding forms of implementation.
However, apparently Copilot is sometimes NOT generating it's own code based on the concepts it has learned, but is instead just regurgitating chunks of potentially copyright-protected code verbatim. It'd be interesting to know if it is doing this deliberately (to maintain the coherence of what it is generating) or not - I guess the more of something it has already copied exactly the more it is likely to continue copying since that is the best "predict next word" continuation. Of course while it would be interesting to learn more about the mechanics of Copilot, that doesn't change the legality, or not, of what it is doing, another aspect of which (although IANAL) is how much of the original work is being copied.
At the end of the day it shouldn't matter whether it's you or Copilot either learning from or copying someone else's code - exact same copyright protections apply.
Earlier quoted context omitted.
I see this basic logic in almost AI ethics threads, and it starts with a big assumption: "humans learn from copyrighted source material without copyright violation". This then gets tenuously extended to "ai also learns, so it must not be in violation of copyright law". The first assumption is highly flawed though. Humans routinely do violate copyright law. Plagiarism is a huge problem in many sectors; un-cited direct…
+1. And let's not forget too that "AI", that is, ML models, are not "autonomous" in the way that humans are autonomous. Sure, we use the word "learn" to describe what they do, which is one word that we also use to describe what people do. But ML models are always wielded by people or corporations for particular purposes. If a corporation was to directly publish some copy that appears plagiarized, we'd call that plagi…
That said I think the results are often useful and sometimes fascinating. We should not fool ourselves about the learning that these large neural nets do, though.
I don't like that opensource code is being used in a commercial product. I feel concerned about NNs learning about stuff they aren't really "supposed to" learn, because somebody published something by mistake a long time ago. But this general argument about reproducing copyrighted code is stupid, and actively trying to shut Copilot down because of that is why lawyers are cancer. Basically, what Copilot (or anything l…
The vast majority of open source code would be almost entirely worthless (or more likely, would straight up not exist) if it couldn't be used in commercial products.
Open source software licenses were a mistake.
Agree about the rest.
I think 'training an AI' is actually a distinctly new use of IP, and should probably be considered under a specific kind of 'AI-use' license. Open Source licenses should be updated to indicate whether they allow or do not allow AIs to be trained on covered work as well as the other rights they allow.
Earlier quoted context omitted.
>If human beings had much more accurate memories, copyright would be quite a bit more intrusive (and/or quite a bit less effective) because, following any exposure to some kinds of works, we could use our own memories to reproduce those entire works from scratch for our own use or pleasure without obtaining authorized copies from elsewhere. I don't know the name, but I remember some sci-fi story about some academy wh…
There is a short story by Orson Scott Card which has that theme called 'Unaccompanied Sonata'. You can read it here: https://b-ok.cc/book/4395497/b2fb2e
If there were no copyright problems then why didn't Microsoft train Copilot on its own source code like windows, visual studio, sql server, etc?
Why would anyone want to stop Copilot is beyond me. Reinventing the wheel, millions of time a day, is an atrocity. Millions of (wo)man hours, wasted, every single day, on writing solutions to problems that have already been solved. There is a partial solution to this, and it's making people angry, it's crazy. If you put your code publicly on the internet, you should expect that people will reuse your code at some poi…
You want to use my code, without ever knowing I wrote it? You want to use my hard work, regurgitated anonymously, stripped of all credit, stripped of all attribution, stripped of all identity and ancestry and citation? FUCK YOU! Training must be opt in, not opt out. Every artist, every creative individual, must EXPLICITLY OPT IN to having their hard work regurgitated anonymously by Copilot or Dall-E or whatever. If y…
I'm in favor of this. You can't ingest code that says "you cannot use this without attribution", put it through a bunch of if statements that strip the license, and then say it's "AI-generated". I don't care about most of our generic CRUD apps or the 15th rewrite of a sorting algorithm, but I do care about those smart enough to advance the field and come up with novel solutions. If we take away the incentive for attr…
I'm probably not the first person to say it through this whole debacle, but that might be me: https://news.ycombinator.com/item?id=33242619
>> "Copilot was Microsoft's first test of people's trust after the GitHub acquisition. It's going very, very, very poorly. There were ways to do this with consent and collaboration with the people and projects it takes code from, but they're acting like classic Microsoft here."