Would you buy WotC products produced or enhanced with AI?

[Art Waring]

A class action suit is probably the right means in this case.

My point is that there are already numerous lawsuits in motion ATM.

 

[SlyFlourish]

I feel like it would be hard to argue that it wasn't at least a copyright violation, since almost definitionally these models are creating derivative works. The companies, of course, argue that they are sufficiently transformed.

In the end I think that @SlyFlourish has the right of this: if material was used to train the model, it must have value, and therefore the copyright holder of that material is entitled to some sort of compensation. What would that look like? Probably an offer of a pittance one time fee.

Asking permission also would have been nice.

 

[Art Waring]

These forums are sadly not often the place for retrospection among people. Just be happy “no” is a lot more common than “yes”.

Oh I am aware, thanks.

I'm not trying to change minds, but I also don't like the idea of sitting by while folks are doing the "this is fine" act, while at the same time members of the forums are being subjected to harm.

As it is, publishing becomes more of a grim prospect by the day. From tariffs to gen-ai, it ain't easy being a publisher right now.

Also, I found it pretty disturbing that among those voting "yes," only 4% voted for "only using ethical data."

 

[Reynard]

Asking permission also would have been nice.

Sure, but that ship has sailed.

As near as I can tell (I moderately follow the industry, but not obsessively) AI firms are way over valued and are nowhere near profitability. I wonder how or if that will factor in to whatever solutions we (collectively) come up with for training them.

 

[The Firebird]

piracy then? People copying music also is not stealing in the traditional sense

Yeah, indisputable that piracy has been used in many cases and by many different companies.

If the pirated datasets like LibGen were removed and the AI was only trained on things like Common Crawl, would we find their actions ethical?

 

[SlyFlourish]

Sure, but that ship has sailed.

As near as I can tell (I moderately follow the industry, but not obsessively) AI firms are way over valued and are nowhere near profitability. I wonder how or if that will factor in to whatever solutions we (collectively) come up with for training them.

who knows. I wouldn't be surprised if it all collapses but we have a bunch of open LLMs floating around we can use for little things like formatting tables and writing mundane code we don't feel like writing ourselves.

 

[SlyFlourish]

Yeah, indisputable that piracy has been used in many cases and by many different companies.

If the pirated datasets like LibGen were removed and the AI was only trained on things like Common Crawl, would we find their actions ethical?

I would.

• They further exasperate income inequality
• They trick bosses into believing they can fire people
• They use tons of power and water
• They continue to evoke a false sense of accuracy
• They displace human work with cheap trash
• They artificially inflate stock valuations for publicly traded companies

 

[Reynard]

who knows. I wouldn't be surprised if it all collapses but we have a bunch of open LLMs floating around we can use for little things like formatting tables and writing mundane code we don't feel like writing ourselves.

Aren't those trained on the same material.

 

[Faolyn]

The Atlantic had a lookup. My books were there.

Search LibGen, the Pirated-Books Database That Meta Used to Train AI

Millions of books and scientific papers are captured in the collection’s current iteration.

www.theatlantic.com

Thanks for the link. It looks like a large number of my dad's books were used as well.

 

[Faolyn]

I don't think the work was stolen.

They used it without permission or compensation, and without providing credit.

What would you call that?