ChatGPT/AI slop: not here, please!

just fyi there’s another tool that does pretty much the same but for way more instruments, and regarding that specific post, wouldn’t feel too bad about it, no “good willing soul” would share a tinyurl link to track them clicks…

and btw, if you already a OAI user, isn’t it as simple as uploading a pdf and asking questions? and you probably should do it in private because the manuals are copyrighted.

This manual is copyright © 2023 Elektron Music Machines MAV AB. All reproduction, digital or printed, without written authorization is strictly prohibited.

2 Likes

I’ve never used that tool or anything else that does similar things. What i don’t understand is why people willingly train a commercial tool that will make their lives more expensive and difficult (believe it or not). Like the story of people training tools that will replace them in that company.

People then go and complain about billionaires, accumulated wealth, fake news, copyright infringement etc.

my guess cause it’s free and because no many care enough to think why is it free and who really benefits from this…

2 Likes

I don’t think it matters. The manual is online. Pretty sure AI can access it.

AI will never swing a golf club for me, take a walk in the woods for me, or hug my girl for me….the important stuff.

1 Like

I’ve been called a tool many, many times, but hardly ever called helpful. So, thank you.

7 Likes

welp, I thought same thing about previous os and overbridge versions but elektron had other thoughts…

1 Like

? You mean the copyright stamp? The manual can be gotten from some music store sites. That isn’t infringement. And I don’t think AI gives a shit about that little symbol.

The manual is also available in more places than Elektron site. There’s some bullshit sites that offer manuals for just about anything and everything. One page at a time. I fucking hate those sites, but I was able to get an obscure manual off one of those junk sites. I would think AI is referencing pretty much all of everything.

not being said argumentatively. Just with distaste for AI in 73.56% of respects

I know it’s available from other sites but fact it that it is copyrighted and:

All reproduction, digital or printed, without written authorization is strictly prohibited.

so while someone can go and put that pdf into llm instance it’s not permitted by elektron, and publishing such link on the forum is not allowed.
he could’ve asked for permission to do that, maybe they would allow it, but otherwise it’s not allowed.

I don’t know about anywhere else but in the US there are laws surrounding what is called “fair use” and something like manuals online are essentially an educational tool so it’s right on the border of what is considered fair use.

I think it’s sort of a gray area depending on if anyone is making money off it, the context of the use and if anyone files a DMCA takedown or any other type of cease and desist.

This is what google said:

Fair use is a legal doctrine that allows the use of copyrighted material without permission in certain circumstances. It’s an exception to copyright law that balances the rights of the copyright holder with the rights of the public.

Fair use can be applied to a variety of uses, including:

criticism, commentary, news reporting, teaching, scholarship, and research.

To determine if a use is fair, U.S. judges consider four factors:

  • Purpose and character of the use

  • Whether the use is transformative

  • The nature of the copyrighted work

  • The effect of the use upon the potential market for, or value of, the copyrighted work

I know that the language on anything which is registered as intellectual or written property is intentionally strong to protect holders of copyright against unfair use, but the “fair use” laws also apply to a variety of situations like a copyright music which might apply to a song in the background of a video but which is not the focus of a video (or even reaction videos I guess, otherwise they could not legally make those and there’s a lot of them). I’m sure that there’s still accreditation required but my point is that I’m sure AI companies and those manual sites are exploiting a loophole.

I’m not judging one way or the other because I’ve bought a printed manual for something off of ebay and I assume that if ebay or the copyright holder were heavily biased towards this specific use, then at least the major marketplaces would have this on their forbidden list.

At the same time, some businesses are extremely sensitive to it, like some copy/print shops won’t reproduce an artwork if it has a signature other than your own, or any portion of a book which is copyright and I think the reason being that they can’t guarantee what you intend to do with the material, however the digital domain contains a huge gray area for all sorts of things.

I just think the chatgpt style AI, despite where it can be helpful for some people, is at it’s core a device which exploits loopholes.

4 Likes

I doubt that an llm consuming the content of a PDF can be regarded as ‘reproduction’. (But I’m not a lawyer).

Not to say there isn’t a problem here of course.

EDIT: Oops, @shigginpit and I posted at the same time which makes this slightly redundant.

People are shockingly lazy at times, and ChatGPT proselytizers will use almost any excuse to make it “useful.”

2 Likes

@bibenu

I completely get what you’re saying but I wouldn’t consider uploading it to 3rd party platform that you have no idea who’s behind and how they use it to be fair use, it’s not like I took the manual and made a cheatsheet - which I’d consider a fair use, fact is you have no idea how this uploaded document can be used by the platform you’ve uploaded it to.

“I made a ChatGPT instance” - do you host it? do you own it? nope, hence, not exactly fair use.

2 Likes

I think it’s less about ethics and more about actionability. These companies for the most part are rolling deep with legal support because the people responsible know that what they’re doing has the potential for gross misuse by the end user (even if AI itself doesn’t eventually become the enemy of humanity).

That’s why despite the attempts at good PR for their technology, I don’t really think these are good companies. It’s an interesting technology but it’s being wielded like an open flame in a dry field of grass. Very hard to control what is burnt and what is not.

I think it’s about ethics too, consider something like someone downloaded a track from your soundcloud, which is copyrighted, and uploaded it to OAI or any other LLM for whatever reason.

someone submitted your music against your permission that later on can be used by the platform for commercial purposes, would you consider that ethical?

is that fair use?

me too, and the fact that they allow submissions of copyrighted material is bonkers.

4 Likes

Our own ethics are only in control of what we choose to do in this situation. The companies can operate because there is apparently enough leeway legally to continue without fear of unavoidable prosecution. Unless and until someone is able to change the legal precedents which govern that system, ethics is only a matter of opinion.

I can say with confidence I have 100% never used chatgpt. But I have looked up manuals online and for my own education, I don’t feel ethically challenged for the use. Whether I’m contributing to the problem by using a site with an obscure manual that even the manufacturer doesn’t provide (or manufacturer no longer exists) it’s up to whoever is in a position to judge me for it.

Whether or not they are using it under fair use or whatever legal clause they can do it, the internet makes it really difficult for any authority to distribute justice aside from takedown notices which probably only really impact stuff like google and youtube. In some instances the sites which are not AI but host such documents may themselves be of vague ownership.

The internet is out of control, the ethical thing to do would be to go back in time and stop it before it became chat gpt. I’m bout to go work on that right now!

Uploading manuals into the cloud (or however chatrtfm works), is probably unarguably a bad idea though, just in my opinion.

2 Likes

The loophole Gen ML (it isn’t AI…which is just a “groovebox” term for what is basically statistics or metaheuristics on steroids) developers aim to exploit is definitely fair use. Their argument tends to be that it isn’t copying content, it’s learning content, and then synthesizing it. Which, no. It’s more like it’s averaging content. It relies on variety for its “creativity”.

We do have a couple of reasons to be hopeful. The longer LLMs and diffusion models are around, and the more dross they pump into the data sphere, the more they become trained on their own output. On its current course, the gen ML centipede just eats its own garbage, and collapses into its own averaging and bias.

Additionally, every LLM query uses an order of magnitude more energy than a basic search. Right now, all these companies getting huge investments to develop these tools are doing the “we run at a loss now to grow, but will be profitable soon I promise” techbro economic model, and running at huge losses to grow the market. So enshittification hasn’t interrupted the honeymoon yet.

5 Likes

Agreed, but none of those sites are showing off about it on Elektron’s forum.

You wouldn’t use an unlicensed Stevie Wonder sample in a track, then go play it outside his house all day, would you…

How do you enshittify something that is already dog shit?

5 Likes

It’s like a Russian Doll of shit. The shit product gets wrapped in layers of increasingly economically exploitative shit.

10 Likes

I DEFINITELY would!

I don’t know where he lives

1 Like