LÖVE

ivan wrote: ↑Thu Jul 06, 2023 5:58 pm ...

Generally speaking, you cannot just copy/paste code that you found on the internet unless it includes an appropriate license.

Yes, but, it's not always case. Nor anyone really asks about source of data that code comes from

far as I know you cannot sue an AI.

Just you wait before 2077 year!

All I am saying is that we should be especially critical towards its developers.

Yes, but don't forget about regular, *human* users, who might do same thing, but since this users is not considered as AI we pay less attention to what they post and from where. It's not like ChatGPT devs invented copy-and-pasting code from someone without consensus, lol.

It depends on where users A,B and C are located - each country has its own copyright laws.

So, there is no universal, "cross-countries" way to determine that?
Huh, interesting, probably.

Nobody will take legal action if you copy/paste a few snippets of code.
On the other hand, ChatGPT is a bot and can be trained by scraping terabytes of data from the internet (including these forums).

So, there is no universal, "cross-countries" way to determine that?

You can obtain a copyright/trademark across multiple countries, but for example North Korea is not obligated to recognize Coca Cola's claims over their brand.

PS. A minor correction: North Korea was recently included in WIPO according to Wikipedia but you get the idea.

Even the creators of ChatGPT are concerned about the legal aspects of their creation, including copyright aspects. I would be concerned about copyright issues if I used code «generated» by it. To me, when you look into it objectively, ChatGPT is a glorified mix-and-match tool with a superior language interface. There's no real creativity, the code is not its own, it's just good at piecing together parts of the programs in its database (because to me, a trained neural network IS a database, since it has the data and is able to retrieve it, just like someone who learns a list of numbers is effectively a database of these numbers) in a way that may make sense, or not.

That it's not a data_base, it's association_base, the way to describe content on the intuitive level, not just key and value definition.

pgimeno wrote: ↑Fri Jul 07, 2023 6:23 am Even the creators of ChatGPT are concerned about the legal aspects of their creation, including copyright aspects

In my opinion, they do not seem very concerned.
ChatCPT's replies are generated based on information collected from third parties, but it does not cite its sources or provide attribution.

ivan wrote: ↑Fri Jul 07, 2023 4:59 pm In my opinion, they do not seem very concerned.
ChatCPT's replies are generated based on information collected from third parties, but it does not cite its sources or provide attribution.

Does asking something like "hey, could you give me source of your reply?" give something useful?

GVovkiv wrote: ↑Fri Jul 07, 2023 6:32 pm

ivan wrote: ↑Fri Jul 07, 2023 4:59 pm In my opinion, they do not seem very concerned.
ChatCPT's replies are generated based on information collected from third parties, but it does not cite its sources or provide attribution.

Does asking something like "hey, could you give me source of your reply?" or something like that.

Yes, completely made up as usual.

ChatGPT uses libs for Löve that had never existed. For example it uses vector math as it was already overloaded.

dusoft wrote: ↑Sat Jul 08, 2023 10:35 am Yes, completely made up as usual.

Kind of expected, but, still.
I'm kind of sure, that it can provide source for something, that had "source: https://example.com" in text. Does anyone can prove that?

it doesn't matter what any of us think about what is legal or how these systems work. no one with real power making these decisions is going to care. they will write the laws as they see fit.

what you should be concerned with is how to get your hands on these tools and make the most of them as soon as you can. LocalGPT is interesting too, for making queries from your own data sources.