My arthritis is playing up, if I come across as terse forgive me.

Technological development isn’t intrinsically harmful though, and that includes so-called “AI”. It’s the ways in which capitalism exploits new (and old) technologies that is harmful, because of the way capitalists use those technologies to drive down labor costs and eliminate jobs.

I don't want to argue a strawman position (i.e. that technological development is intrinsically harmful) but I do adopt the position that certain technologies lend themselves more readily to good, and conversely bad, ends. I think this is extremely obvious and won't bother explaining further unless asked.

In the field of so called AI I will admit that the underlying technology, the transformer model, is freaking awesome. Huge step forward, unlocked giant leaps in computer vision among many other things. The chatbots and diffusion algorithims? not so much, genuinely I struggle to think of positive social impacts. Most people give me are based on fundamental misunderstandings of how they are made and what they do, e.g. they are an appalling way to learn a topic or improve your writing (see semantic ablation if you need that last one explained https://www.theregister.com/2026/02/16/semantic_ablation_ai_writing/).

Mass manufacturing anti-information not only misleads people but also buries real information in a tide of slop. They are fundamentally designed to seem plausible, aka to deceive a user that information/communication/language/art is happening when it is not.

There are many open source models available.

This is a popular refrain and it based on a misunderstanding of what they are. They are open weight models, the source (the information used to make them) is not shipped. They are more like shareware. In theory you could tweak them, like you could go to neuron 65423455 and tweak the weight but as the models are by nature inscrutable this is not a useful exercise. You cannot study what training went in, analyses sources for bias, replicate the creation, or anything else a truly open source model would be, you can simply run the finished product.

Hobbyists and anarchists can deploy a locally hosted AI to work against capitalism, with none of the downsides. It’s perfectly possible to use AI tools ethically and responsibly.

Can you? The technology is fundamentally nondeterministic, impossible to analyse, extremely convincing which makes disinfo/bullshit (in the Frankfurt sense) much more dangerous. Further because of the compute required to train them and the lack of source these models rapidly become outdated and of diminishing use. I think capital is better undermined by building solidarity, sabotaging these machines/their training, and making sure you retain your ability to operate without tools created by them.

But folks aren’t fighting against them by complaining to us, they are instead directing their energy to fighting with an anarchist instance which is dedicated to fighting against capitalism, including corporate abuses of “AI” tools.

I don't think anyone is trying to harm the instance, simply convince people. Are people ddosing you guys? doxing people? If you believe someone is doing something harmful you basically have a moral duty to try and convince people don't you?

Are we responsible for any of the harms associated with corporate AI? No. And we have never “promoted” AI, at most, we have just defended the way in which we use it, because it isn’t harmful in that context. And yet…

The horde is interesting but I do think that it is dependant on capital as outlined above, and that in general using these things devalues language/art/humanity and harms your brain.

In terms of relative harms, the anti-GenAI crowd on lemmy has done way more actual harm than anyone on this server has by posting an AI image.

There are many devoted leftist that are against this stuff, there are many horrible techbro types that love it. It is not reasonable to smear a position based on the worst people that hold it.