I’m not a troll, I just call other people out for their BS.
I think most of lemmy.today is bad faith astroturfing.
This is not the first time !PapaSkwat@lemmy.today has preemptively banned me from a community I’m pretty sure I’ve never interacted with.
This is a warning for others. Do not assume posts on Lemmy.today are real or in good faith.
Here is a related post with similar circumstances.
Why are you (presumably) subscribing to AI communities when you hate AI?
I didn’t, I just saw this in my mod log. I’ve never interacted with this community to my knowledge.
PS, AI is a very broad term. I wouldn’t say I hate AI. But I do hate generative AI (genAI) which includes LLMs and models that generate images, videos, and audio. I especially hate the resources used for its compute, the way it poisons our environment and that it is being sold at massive losses like a drug to get people hooked. Also I think genAI is a tool to surveil, control, and isolate its users. Most of what genAI does means other humans are taken out of the loop which I think is extremely detrimental to human society. We should be coming together more than we should be trying to do things alone what a stolen facsimile of a human.
This is off-topic, and I don’t want to get too far into the weeds here, but not all LLM use depends on large data centers or Big Brother Corporations.
Smaller open-source models can be downloaded and run locally on your personal computer. No connection to internet needed to use them. No corporate oversight. No surveilance, no overt control. They can also give users much more control over their own data, software, and hardware. The same resources use you are using just using your computer. These are the LLMs I use and advocate for.
Most of your issues about large-scale corporate AI don’t apply to this local LLMs use. Local LLMs are just another tool that people can choose to use or not use. I share a lot of your concerns about massive BS corporate AI infrastructure, but these small, locally run models are a totally different thing.
It’s just a tool. Some people use it for good. Some for bad. Like any other tool out there. You can build a house with a hammer. But you can also kill someone with a hammer. Humans are the real problem, not the tools.
I agree that locally hosted FOSS llms and genAI models would be a better option than something like Anthropic’s Claude.
But actually my largeest concerns are not alleviated by local hosting.
And this is assuming that this facsimile of stolen human intelligence (all open source models were still trained on unauthorized human content) can have a low rate of error.
Even when fine tuning or using RAG, the rates of model collapse and catastrophic failure are very high. And those rates compound the more these “tools” are used in the same context windows.
Plainly put, it would be better to ask for help from a human than from a genAI.
I agree with your idea, but reality is that many humans are shitty people. Or just ignorant. They have been long before AI, and will continue to be long after AI. Even in this very thread, comments have been removed just today and the mods here banned him from this community for his ugly transphobic behavior. And the poster admitted he just created the username today just to say that shit here. I’d take AI over that guy any fucking day.
I’m not arguing that you should like AI. Feel free to hate it. Feel free to boycott it. I’m just giving you some background about the nuance of it. Also, this is a pro-AI instance (concerning local llms, not coporate ones) that we’re having this discussion on.