I am not a lawyer, but my understanding is the truth of legal systems is that they are a gray area of interpretation and precedent. In a strict definition of copyright, copying a file from a server into the cache in your ram is a copy that theoretically could be ruled an unauthorized copy. This being obviously a ridiculous idea, but I believe if you streamed content from a website this is the only way they could fine you for copyright violation. Generally with the networks you mentioned, the distribution is prohibited, and most being peer to peer systems they cite you for that.

The more charitable take on why they are not the same is; anything on the open web is assumed to give the right to copy to a closed ‘local’ system and use for your own system, as that is fundamental to web browsing. Further in the same way you can make a local copy of a movie to cut it up and use in a review, and that is fair use as a transformative work, you can make copy of the open web content and make a transformative work for it.

At least that’s how I would argue it, if I was paid to be OpenAI’s legal puppet. All of that is to tiptoe around the reality that the legal system is a tool of the rich. So the law becomes firm for the purpose of piracy of a movie, but long and difficult for the purpose of a large companies profits.

I think the core of the fair use argument is that the AI models that are being trained are transformative products of the original works.

Might be a hot take here but I basically agree. I still believe it was theft and that the realities of the legal framework we had don’t really stand up to the evolving problems, but under the current laws there is really no justification for saying that, taking the input of a bunch of images and giving the output of a set of statistical correlations of pixels based on descriptions, isn’t transformation.

But only resist if it’s safe to do so.

It turns out the requirements for using your second amendment to protect your home from an ICE raid also very quickly enables suicide.

You can still push for folks to learn to use a gun and train with others to feel comfortable with it, while still not believe that people with suicidal tendencies should have an answer to the question of how they would kill themselves 1 minute away.

When they stop supporting fascism.

Where else would it come from? 🤷‍♂️

In either case, it’s such a great bit.

I am aware of private trackers. I guess to me, even in the space of “it’s neat to think about” something like Loops and built on activity pub makes more sense.

Now if we imagined folks offering home hardware or ‘seed boxes’ to help an instance out, that I could see being useful tech, but constantly uploading from your phone seems a bit silly.

Ah okay, that was my understanding as well. I was curious if anyone had tried making an ad.

When you say there was a push, do you mean there was a time when they did have ads, or do you mean the community push with the twitter troubles?

I am aware of the problems and have had similar thoughts about how best to deal with the taxing reality of streamed video, but I think the reality of it is, while already fighting the network effect and ad budgets, someone that downloads an app and see it saps half their battery for that day because they liked 5 videos and left, they are going to uninstall it.

I think instanced makes the most sense, and even that would be a hard ask if popularity every spikes.

I think the fact that Peertube which I think of as more a PC interface, where bandwidth and power consumption are less an issue, but still chooses to limit the peer connections to active watching speaks to how discordant the idea is with what people expect from streaming media.

I would happily use a desktop app as you described, so if it ever exists, let me know 😁

Is there any discovery or list of other instances?

I think people would be pretty quickly upset by the batter and storage drain from this. I have to imagine if picking an instance is a barrier to mastodon, most folks are not interested enough to learn the mechanics of why likes suddenly use system resources, and see it as a failing of the app.

Not really. There is a right wing podcast with a rotating host of fascists and Newsom as the only guest where they just dog walk him.

What gives you faith in this system?

If you have faith in these systems, I can see how one might see this as progress. But it’s like saying “the wheels are spinning, we must be moving” while stuck in mud. I consider it progress with they do something. When the people that are inflicting violence on random members of the community are physically stopped. When police stop treating unmasked men with guns violating the constitution as allies. When they can’t shuffle away murders to relocate them another state, and are instead stopped at the airport. Maybe then I could muster some faith for the judicial system, but as it sits, I don’t care if their on the ‘double plus super notice squared of the 5th degree warning’ they are going to stop nazis in the street for real this time.

then pump up the price or flood every prompt with ads.

“Sure I found that document you needed, and with it, I also found this great new game I know you’ll love. Raid: Shadow Legends, It’s a free to pla…”

I cannot wait for companies spending 300 dollars per user per month for this convenience.

Well like much analysis of economic impact, that’s hard to quantify, so we can just assume it’s trivial.

why would they have used it in the first place?

Because 30 seconds with an image generator looks nicer than 30 seconds in MS paint, the deeper point being the deciding factor is that it took 30 seconds of time.

It shaved off work hours.

I think this only makes sense in some abstract of a net aggregate of artistic labor hours. The reality though is that this work was never done by the artists, never given significant time allocation, and would never lead to hiring more artists.

It was placeholder art. They didn’t reduce the artist hire because they weren’t going to have the artist make orange boxes and MSpaint character icons.

The reductio ad absurdum is equally silly the other way. “Does the seeded algorithmic generation of a cloud texture disqualify anything that uses it as AI???” This is a debate stage level talking point, and is unconvincing in reality.

Super cool that we build a system that is only held accountable months later through courts that were taken over.