WotC Would you buy WotC products produced or enhanced with AI?

[Crimson Longinus]

As someone said early on, these concerns will seem quaint in 5 or 10 years.

Indeed. At that point the last surviving humans probably have more pressing concerns than the copyright.

 

[kevinnc]

[accidental double post]

 

[kevinnc]

Thus far it seems to me that if I had real piracy concerns with AI, I'd also have to have them for google to be ethically consistent. I'm interested if I'm missing anything.

To me, the simplest difference is that there are methods to let Google (Bing, DuckDuckGo, etc...) know that a page should not be crawled or indexed, and search engines (at least all the major ones) will respect that and not do so.

I would imagine there are some more ethical LLMs that follow the same rules, and many that do not. But if someone has a website that is being crawled by search, then they are either choosing to allow it, or are not aware of the method to label it as essentially "do not crawl this".

Ref:

Robots.txt Introduction and Guide | Google Search Central | Documentation | Google for Developers

Robots.txt is used to manage crawler traffic. Explore this robots.txt introduction guide to learn what robot.txt files are and how to use them.

developers.google.com

Block Search Indexing with noindex | Google Search Central | Documentation | Google for Developers

A noindex tag can block Google from indexing a page so that it won't appear in Search results. Learn how to implement noindex tags with this guide.

developers.google.com

 

[EzekielRaiden]

and yet it is much more probable than the ‘truth’ your textual analysis or courts arrive at. So why do you call those truths

Semantics. And, as a result, we're done here. I'm not continuing this conversation with you.

This is the only answer you will receive from me.

 

[The Firebird]

To me, the simplest difference is that there are methods to let Google (Bing, DuckDuckGo, etc...) know that a page should not be crawled or indexed, and search engines (at least all the major ones) will respect that and not do so.

I would imagine there are some more ethical LLMs that follow the same rules, and many that do not. But if someone has a website that is being crawled by search, then they are either choosing to allow it, or are not aware of the method to label it as essentially "do not crawl this".

This is a good point. I think LLMs would benefit from much more transparency and user choice in this regard.

 

[Jfdlsjfd]

This is a good point. I think LLMs would benefit from much more transparency and user choice in this regard.

That's the legal framework of some TDM exception. If you're doing a commercial model, then you should respect opt-out, if the model isn't commercial, then there is no opt-out (since you won't make money on it, there would be no point in sharing revenue anyway).

 

[Faolyn]

It's a new ethical issue. I'm thinking it through.

Thus far it seems to me that if I had real piracy concerns with AI, I'd also have to have them for google to be ethically consistent. I'm interested if I'm missing anything.

Except, as I have repeatedly pointed out, those are two very different things.

OK, let's go with this: you go to your local grocery store. There, you pick up some ammonia-based cleaner and some bleach, because you have some heavy-duty scrubbing to get done.

Whose fault is it if you decide to mix the two together to save time and create deadly chloramine gas?

Hint: it ain't the store's. Or google's.

That's basically what a search engine is: a store, except that (a) this store sells everything, and (b) the customers do the stocking. The managers and employees can their best to keep bad things out of the store, but, well, everything is for sale. So it's a bit hard, because the store is so big it might as well be infinite, at least by human perspective. And there are only a finite number of managers and employees.

Also, there are far more uses for ammonia and bleach than creating deadly gasses. In fact, using them on their own is a good thing, because you can clean and sanitize things very well. So if you remove them from your store, you're also preventing people from using them for their actual purpose.

Kind of how like the majority of wikidot pages do not contain pirated material. Google can't really de-index the 5e wikidot because all the owners would need to do is change the URL. Google would have to de-index the entire website, most of which is completely legitimate.

Also, not only would Google have to do de-index it, but every other search engine as well. I didn't even know Webcrawler still existed! And let's face it, there's more important things for google's employees to do.

Let's exit this analogy, it's not google's fault that a website they don't own has one user using that website to host pirated material. Nor is it google's job to de-index all of wikidot because it contains pirated material.

Wikidot is, as far as I know, a privately-owned website, meaning they can host whatever they want to host. If WotC sent them a C&D, then they'd have to listen (or deal with whatever consequences there are for not following it).

Since it's one of the first things that comes up when you search for D&D material, it would be shocking if WotC wasn't aware of it. And wikidot isn't some sooper-sekrit pirate site. It's right there, out in the open. It claims to be the third largest wiki farm out there.

Also, it's a Polish site. I don't know anything about how American copyright laws are treated over in Poland, although apparently Poland has "among the most modern copyright laws in the world," whatever that means, and that they were a signatory of the Berne Convention ("The Berne Convention, adopted in 1886, deals with the protection of works and the rights of their authors.")

So two possibilities:

(1) WotC can't legally do anything about a Polish website using their material, because this site isn't breaking Polish laws, in which case, why would google de-index a site that's not breaking its country's laws and that WotC isn't going after legally?

(I have no idea if the 5e wikidot breaks any Polish copyright laws)

(2) WotC is fine with the 5e wikidot, meaning that they have given their consent.

And just to reiterate: writers and artists did not consent to having AI scrape their work.

I think the discussion has a broader scope than that. We can think it is unethical to use AI for creative material in commerical products while thinking it has loads of very useful applications elsewhere.

And we're not talking about those other useful applications. We're talking about gaming books "produced or enhanced" with AI instead of paying actual writers and artists.

I'm sorry Faolyn but this statement seems to be outright false. The largest source of (weighted) training data for chatGPT-3, for example, was Common Crawl, which is not going to differ substantially from google. Maybe google is primarily using pirated stuff. But you rejected that, and in that case LLMs are not built almost entirely on pirated material.

From Wikipedia:

The Common Crawl dataset includes copyrighted work and is distributed from the US under fair use claims. Researchers in other countries have made use of techniques such as shuffling sentences or referencing the Common Crawl dataset to work around copyright law in other legal jurisdictions.

Hmmm.

Did you get confused by their terminology and think that Common Crawl used only public domain material?

I guess this gets back to why I care. I see some of the statements in this thread. And critics of AI just get stuff wrong all the time in their rush to attack it. The last few posts talk about burning books and say "AI has brought nothing of interest or of value".

Because it doesn't. AI takes away from actual human work and fails to pay the humans whose work was scraped, and the people who are using AI are not hiring actual humans to do the work.

That is why AI has brought nothing of interest or value, because anything it can do, a human can do better.

 

[The Firebird]

Except, as I have repeatedly pointed out, those are two very different things.

I don't think this discussion is progressing. I don't think your example is a good one.

The managers and employees can their best to keep bad things out of the store, but, well, everything is for sale. So it's a bit hard, because the store is so big it might as well be infinite, at least by human perspective. And there are only a finite number of managers and employees.

I've discussed ways to implement content moderation before.

Kind of how like the majority of wikidot pages do not contain pirated material. Google can't really de-index the 5e wikidot because all the owners would need to do is change the URL.

A forum poster would just make a new account...

Google would have to de-index the entire website, most of which is completely legitimate.

That seems a reasonable trade off.

Also, not only would Google have to do de-index it, but every other search engine as well. I didn't even know Webcrawler still existed!

There are a lot of discussion boards. They all need moderators?

Did you get confused by their terminology and think that Common Crawl used only public domain material?

No. My point was that google and common crawl look at similar data.

That is why AI has brought nothing of interest or value, because anything it can do, a human can do better.

Even if that's true that doesn't mean it provides nothing of value. Speed and efficiency matter. And I know it provides things of value because it provides value to me.

 

[Bedrockgames]

I think a big company like WOTC has the resources where AI makes no sense (at least for things like books). I would have more understanding if it were an indie company trying save money. But WOTC can afford art, writing, design, etc. And they charge enough that my expectation is I am getting something made by a human

 

[Faolyn]

OK, @The Firebird, we're done.