Post B54Ho6VDZyxvu8M8ga by pojntfx@mastodon.social
 (DIR) More posts by pojntfx@mastodon.social
 (DIR) Post #B54ETQtyEyuXEBNyEq by wezm@mastodon.decentralised.social
       0 likes, 0 repeats
       
       Huh Limine has moved back to GitHub from Codeberg https://codeberg.org/Limine/Limine
       
 (DIR) Post #B54ETR7nPZyBv3r0mu by wezm@mastodon.decentralised.social
       0 likes, 0 repeats
       
       Wonder if it has anything to do with this https://codeberg.org/small-hack/open-slopware/pulls/263
       
 (DIR) Post #B54ETRKYe8B6YdpCgC by pojntfx@mastodon.social
       0 likes, 0 repeats
       
       @wezm I'm ngl Codeberg doing weird things in regards to some kind of anti-LLM ToS is like my No. 1 worry with the site lol
       
 (DIR) Post #B54ETRZRkm5VIon5t2 by lumi@snug.moe
       0 likes, 0 repeats
       
       @pojntfx @wezm they are doing that? sources?
       
 (DIR) Post #B54F20ZY7mVFhiJw4O by pojntfx@mastodon.social
       0 likes, 0 repeats
       
       @lumi @wezm There is no official indication of it yet from the eV, but check out the posts of Forgejo's developers and Codeberg's maintainres here on fedi, it's pretty clear what their stance on the issue is
       
 (DIR) Post #B54F20kBUF0gEhIQe8 by lumi@snug.moe
       0 likes, 0 repeats
       
       @pojntfx @wezm i hope they do eventually get to banning llms (and genai more broadly)but if they do, i hope it's done in a sane way. innocent until proven guilty. i would rather have a genai project slip through the cracks than have an innocent one get punishedbut if there is an obvious llm booster thing with an AGENTS.md that clearly promotes and accepts llm contributions (or contributions using other genai models, like genai assets) then that should imo be against the rules
       
 (DIR) Post #B54Fj0gTaK8OVE4704 by pojntfx@mastodon.social
       0 likes, 0 repeats
       
       @lumi @wezm Again, I strongly disagree with this. Every relevant OSS project uses these clearly useful tools now, including Linux and systemd. I know they were useless just a few months ago. Things have changed.
       
 (DIR) Post #B54Fj0yCWQJROCMGcy by lumi@snug.moe
       0 likes, 0 repeats
       
       @pojntfx @wezm i tend to object on moral grounds rather than efficacy groundsthis doesn't mean i think they work well, or don't work well. don't really have strong opinions therein some ways proprietary software can be effective, as well. that still doesn't mean i accept proprietary software
       
 (DIR) Post #B54GDUzHTyBJArOoi0 by pojntfx@mastodon.social
       0 likes, 0 repeats
       
       @lumi @wezm Fair, I'd even push back on the moral grounds these days. Abolition of copyright etc. have long been goals of a lot of people who work in OSS. Energy issues are probably better solved by making it possible to build out renewables/to get abundance like China is doing instead of trying to regulate consumption. Really, really good open-weightmodels that are better than (GLM 5.1) or competitive (Gemma 4) with the proprietary ones exist. But whatever. I'm not an eV member
       
 (DIR) Post #B54GDVLc8w2uI7qeWG by lumi@snug.moe
       0 likes, 0 repeats
       
       @pojntfx @wezm i'm also completely for the abolition of copyright and i only use copyright as a way to go against genai. i do not agree with copyrightthe issue, to me, is more fundamental, i think it is not okay to take the creative output of humanity and then regurgitate it. even if/when this regurgitation is effective. trying to replicate creativity is offensive to mei want a more human world, and this goes completely against it
       
 (DIR) Post #B54GWZ5WSbUvOVPUXY by KaitlynEthylia@void.lgbt
       1 likes, 0 repeats
       
       @lumi @pojntfx @wezm I agree with this but also this goes for other big data technologies that existed before the LLM boom, algorithms that attempt to profile and quantify people are equally gross.
       
 (DIR) Post #B54Gbv7RycJBAcridc by lumi@snug.moe
       0 likes, 0 repeats
       
       @KaitlynEthylia @pojntfx @wezm definitely true. genai is only the most recent thing in a big chain going back until the tech industry startedand if you go even further, you eventually will get to capitalism
       
 (DIR) Post #B54Ho5VBIdLGnjKazQ by pojntfx@mastodon.social
       0 likes, 0 repeats
       
       @lumi @wezm > i think it is not okay to take the creative output of humanity and then regurgitate itIdk. I'd argue that code is creative, and that we've been regurgitating it by learning what others have written and then writing our own things on top of what we've learned for decades. That's as fundamental to humans as it is to LLMs. Outlawing learning doesn't sound like it will lead to anything good here.
       
 (DIR) Post #B54Ho5oK9Sedl6HspM by lumi@snug.moe
       0 likes, 0 repeats
       
       @pojntfx @wezm when humans learn to code, we do so by trying things out, feeling our way around and seeing what works. we do so by building mental models and refining them. this whole journey is something very personalthis then leads to every human having their own tastes; their own way they think about things and like to approach problemsllms have none of that, it's just backpropagation on a multilayer perceptron network (am oversimplifying a bit, but it's nothing like how humans learn)
       
 (DIR) Post #B54Ho6VDZyxvu8M8ga by pojntfx@mastodon.social
       0 likes, 0 repeats
       
       @lumi @wezm LLMs just show us how ridiculous the whole concept of copyright was by building a technology that is only possible to exist on top of pre-existing information
       
 (DIR) Post #B54Ho7N6Le4EbFZAfo by pojntfx@mastodon.social
       0 likes, 0 repeats
       
       @lumi @wezm A fix for this would more so be to properly fund the commons, whether code or other art IMHO, not to outlaw learning by machines
       
 (DIR) Post #B54Htf5yOKlUdkCLjc by lumi@snug.moe
       0 likes, 0 repeats
       
       @pojntfx @wezm i mean, i'm not going to disagree on copyright being ridiculous. copyright is ridiculousi think, more fundamentally, a fix would be to ditch capitalism. but that is not something that can be done in the short or even medium term x)
       
 (DIR) Post #B54IBeKNz3o7TKW6V6 by pojntfx@mastodon.social
       0 likes, 0 repeats
       
       @lumi @wezm I know that LLMs aren't "intelligent" or "learn" the same way humans would, I'm saying they are useful in their current form AND that their current form needs free access to information the same way humans do unless we want a cartel of OpenAI and Anthropic to be the only ones who can buy content and train their models on it
       
 (DIR) Post #B54IBeWRGFRs4i9jHs by lumi@snug.moe
       0 likes, 0 repeats
       
       @pojntfx @wezm i don't think them being useful resolves the moral issuesi'm all for free access to information, though :3
       
 (DIR) Post #B54KPb96oJpVZA1MsS by pojntfx@mastodon.social
       1 likes, 0 repeats
       
       @lumi @wezm I mean fair, if it's just a different ethical framework then that explains it!