Framing it as property is the wrong angle though. It's an ecosystem, not a cache of good writings that was stolen. That ecosystem was hurt severely and its recovery is uncertain.

I think we'll not have people writing good content for a long time (there's no reason or incentive to), and the effects of this will splash back heavily on AI companies themselves.

You can see AI as a battery for intelligence that took a long time to charge and it's being used right now. For years, it was charged with all sorts of novel content that went undiscovered and AI is making available. That charge is the production of novel content, new insights, cross-pollination between areas, slowly driven by humans.

My view also draws a conclusion about recursive self-improvement: it is impossible for a battery to re-charge itself. I don't particularly think it can be done with this technology (LLMs).

I could be wrong though, but I don't think I am, and we'll know within our lifetimes. If things stall, it's likely because it has ran out of seeds/charge/substrate and not a technical limitation. It is in the long-term interest of AI companies to make incentives for people to generate novel public insights, they just don't know that yet.

Microsoft is one to talk.... Remember when MS trained copilot on all your github code?

I'm not sure how to be upset over this. For decades copyright has been extended and extended. Meanwhile, the ease and speed of spreading published works across the globe have increased massively. Don't forget that when copyright was first created, it could take multiple years for first editions to make it across the globe. In that environment, multiple decades of copyright makes sense. Whereas today that would actually stifle innovation and creativity rather than incentivizing it. I mean, because of the length of copyright we've gotten all of these live action remakes of Disney films or superhero movies rehashing the same stories.

I find it a little hard to be upset about AI. Supposedly stealing copyrighted works when the vast majority of those works. Probably should have been in the public domain to begin with. I have a faint hope that this scuffle between the AI companies and the publishing industry will result in more reasonable copyright laws, but I think it's more likely that exceptions will be made and AI will be treated as a special case.

As if the transatlantic slave trade wasn't a thing. What a perfect illustration of our industry's self-absorption and self-regard.

Has anyone found a meaningful discussion of how scraping is theft?

Obviously, reproducing works in whole is infringement. That's not what AI is doing, so the question becomes: how is scraping different from ordinary reading? Is it just that site owners want to play back history and retroactively create high-cost licenses for scraping?

Copyright infringement, if this even were that, is not theft. Chattel slavery is the largest theft of labor in human history.

This is a great basis for a dividend from AI revenue to be paid back to society in a more inclusive form than stock. As more money flows into AI companies, data centers, and other related infrastructure, an amount should be extracted and redistributed in the name of balancing this equation.

I think it's okay to advance humanity, but they can GTFO when they then try to ban distilling and open models.

Transatlantic slave trade calling - we’ll hold. I know there’s a memory shortage but history books are cheap.

Funny. The end of copyright and patents is by far the best thing about AI to me. All of it is nonsense. Great that you drew a picture of a mouse once, I really fail to see why I couldn’t draw it and sell it either. It was moronic from the get go.