I donβt disagree that the line is blurry, I just donβt think itβs nonexistent
05.06.2025 20:31 β π 0 π 0 π¬ 0 π 0@s3rios.bsky.social
Free-thinking determinist
I donβt disagree that the line is blurry, I just donβt think itβs nonexistent
05.06.2025 20:31 β π 0 π 0 π¬ 0 π 0I think it absolutely is something artists care about, especially when that exposure is going to someone else. Someone else putting their name on something I made feels bad
05.06.2025 20:30 β π 0 π 0 π¬ 1 π 0It's saying to them that you don't think their labor was worth giving recognition to. People generally like it when what they do is recognized in some way, and I don't think that's just a product of the property relation.
02.06.2025 21:11 β π 0 π 0 π¬ 1 π 0I think plagiarism is bad for more than just its harm to the people being plagiarized as well. I think it inhibits informational flow to plagiarize, I think itβs dishonest, I think it leads to sloppy work, etc.
28.05.2025 02:07 β π 0 π 0 π¬ 1 π 0I mean I think all else being equal itβs bad to make people feel bad. Itβs not ALWAYS bad but I see no compelling reason to think that there is some good served by plagiarism that outweighs the disrespect inherent in it, not to mention that there are lots of other ways in which itβs bad.
28.05.2025 02:05 β π 0 π 0 π¬ 1 π 0What is βthis situationβ?
27.05.2025 21:26 β π 0 π 0 π¬ 1 π 0β¦trying to reconstruct their belief system so that you can understand WHY they believe that. βThis person is dumb in this screenshotβ is worthless if you canβt explain the conditions that led to that statement being made, which requires more specific apprehension of their overall output.
07.05.2025 18:00 β π 2 π 0 π¬ 1 π 0How do you become alerted to new things they say if youβre not following them? Seems more efficient to just follow them instead of constantly checking their page.
Also, studying a bad person isnβt as simple as just finding dumb things they say and calling it a day. Whatβs far more interesting isβ¦
OpenAI accused deepseek of using ChatGPT in its training somehow, and while I donβt know if thatβs true, it shows that these companies are in fact worried about IP. They will circumvent it when it benefits them, but they still rely on it.
07.05.2025 05:36 β π 0 π 0 π¬ 1 π 0But it is literally still a different image. Like if you had used that image instead of the protected one in the dataset, it would be the exact same. So in other words it isnβt strictly determined by the data like you said it was.
07.05.2025 05:32 β π 0 π 0 π¬ 0 π 0How do you think these companies would survive without IP? Do you think Adobe, Disney, or most tech startups can do what they do if they donβt have patent protections or own any copyright?
07.05.2025 05:31 β π 0 π 0 π¬ 2 π 0Many in the petit bourgeois are just trying to survive. Mom and pop shops run on very slim margins. It doesnβt change their class position; in this case, the expansion of private property rights.
07.05.2025 05:29 β π 0 π 0 π¬ 0 π 0Leftists have opposed copyright for a long time.
07.05.2025 03:03 β π 2 π 0 π¬ 1 π 0My overall point is made in the video: practically, the *expansion* of copyright in this case would not fix the problem of displacement and consolidate power in the hands of bad actors. I believe training should be considered fair use (with that caveat that outputs CAN be infringement in specific
07.05.2025 00:53 β π 64 π 7 π¬ 15 π 8My reading of what you said is that the infringing images can be replaced with images that encode the same structure and you will get the same model out the other end. Is that a misinterpretation?
07.05.2025 03:01 β π 0 π 0 π¬ 1 π 0The crypto scene was never populated by artists. People who sell NFTs as an investment and AI bros who hype the tech arenβt artists, though they are the bourgeoisie. Iβm talking about the PETIT bourgeois artists.
07.05.2025 03:00 β π 1 π 0 π¬ 1 π 0So in other words, my friend was right, itβs not about the images themselves but the structures.
07.05.2025 02:17 β π 0 π 0 π¬ 1 π 0Copyright is inherently in favor of large companies, the same way that private property is inherently in favor of the bourgeois. Sometimes private property will protect you (say your house is burgled) but the net effect is negative.
07.05.2025 02:07 β π 0 π 0 π¬ 0 π 0Even when βevenly enforcedβ copyright is bad. Itβs bad for artists especially, who lose control of their work. The people who benefit most from IP are large corporations and petit bourgeois artists who are in a position to negotiation for their rights.
07.05.2025 02:06 β π 0 π 0 π¬ 2 π 0If it produces the same result without any of the original training data does that not lend support to the claim made by my friend that the specific training data isnβt that important for the weights?
07.05.2025 02:03 β π 0 π 0 π¬ 1 π 0β¦scope of copyright will actually do anything at all to solve the problem. Indeed, it could easily make it worse, as Alex explained.
06.05.2025 23:08 β π 1 π 0 π¬ 0 π 0β¦spread around. It will likely only get worse too, which I find extremely depressing.
The difference is that I can recognize when the cure is worse than the disease. Copyright and IP more generally have been a blight on society far longer than genAI and I am wholly unconvinced that expanding theβ¦
I actually hate chatbots btw, I think the write poorly, I hate that people use them instead of doing real research, I hate that every student just uses it instead of learning. I also think that image generators have a net negative effect on the world because of all the bullshit misinfo that getsβ¦
06.05.2025 23:05 β π 1 π 0 π¬ 1 π 0To be clear, I was referring to the data laundering thing.
06.05.2025 22:59 β π 0 π 0 π¬ 1 π 0The fact that you think something like this would be written by a chatbot shows how little you actually understand them. Chatbots donβt make obvious spelling errors. Iβm not going to talk to you if youβre just going to assume Iβm a liar.
06.05.2025 22:55 β π 0 π 0 π¬ 2 π 0Iβm unclear on what you mean by this. Can you link an example of this?
06.05.2025 22:11 β π 0 π 0 π¬ 1 π 0Why is βfriendβ in scare quotes? Do you think Iβm lying?
06.05.2025 22:03 β π 0 π 0 π¬ 1 π 0I decided to ask a friend of mine who knows way more about what heβs talking about than I do. This is what he said.
06.05.2025 21:32 β π 0 π 0 π¬ 1 π 0Itβs not a βmixtureβ of the training data. Thatβs just wrong. It isnβt combining things.
06.05.2025 20:18 β π 0 π 0 π¬ 1 π 0The reason this one in particular showed up is because it was in a lot of news articles.
06.05.2025 20:02 β π 0 π 0 π¬ 0 π 0