Having lost the skill of manual labour means you will fail to achieve the goal or at least struggle immensely, which leads to nowadays capitalist crisis as the actual work is being degraded and recycled. AI just accellerates the process and promotes the contradictions inherent in the capitalist system.

I agree, I couldn’t find it anymore because it got limited release but I read an article several years ago (before LLMs even came out) that explained basically the more capitalism progresses, the less we teach students. They also used the electrician example, apprentices used to have to learn how a multimeter actually worked, what it measured, how it did, etc. Nowadays, according to the author, it’s sufficient that they know how to read the output. Stick the prongs in the outlet and then you should see X or Y number, that’s all you need to know.

So AI is just one more thing in the long line of such advancements that make earlier methods trend towards obsolescence. I wouldn’t say some work is being made obsolete right now, it’s not there yet, but it’s trending towards it. It’s like the steam machine; the steam machine proletarianized a lot of people and made commodity production possible on a scale never seen before, but if we’d said “let’s destroy the steam machine” then we would still be living in feudalism. I mention artists in the OP because it’s the one group I see most opposed to AI (to the point that in some circles it completely silences other legitimate criticism of AI because it’s all about the illustrators - in the essay I linked, the author also talks about the people who initially categorized and organized images on archives and repos that were used as a basis for AI models to learn how to describe image content), but it’s not unique to that group. Their problem is with capitalism, not that their job is getting proletarianized… or at least they should see it that way. A lot don’t.

On the topic of what to know vs not to know, I don’t know if we can really stop or slow down that trend either. What I mean is there are lots of skills we’ve lost, such as making a fire in the wild, building a timber house, or printing on a Gutenberg press. Some people try to revive these skills and good for them, but they’re also not doing it professionally or in any meaningful capacity beyond providing side income for themselves on the novelty of it or as a hobby. Nowadays books are printed on rolling printers, not on a Gutenberg press. I don’t think it’s unique to capitalism but certainly due to profit motives capitalism accelerates it. So it’s not really a thing AI did, it just made it more apparent. Personally when I use LLMs I still use my brain, it helps me get an idea off the ground or get me started in the parts I’m worse at, and then I handle the rest. It lets me get stuff out the door quickly and focus on the actually interesting parts.

Speaking of art since we’re on it… if we wanted to be ‘purists’, then painters should be mixing their own colors, crafting their own canvas and making their own brushes. A lot do because they want something that’s not commercially available, but a lot are also digital artists where the color mixing is done in software for them without actual pigments. It’s this kind of trend that I’m talking about, it’s not just AI - if we say “this is going too far I don’t like it” then we’re advocating for reactionary socialism (the one described in the manifesto). Even I as a designer wouldn’t know how to use the tools of old, before Photoshop and Adobe.

I could name a ton of stuff I’ve done with AI, but for example we designed the new ProleWiki homepage with AI help in the thinking process. It helped us clarify our audiences, what we needed to have on the homepage, and make sure that we were on the right track with our assumptions.

Using the output of an LLM raw is what tech companies say to do because they want to hype up their product and make it seem like it’s as easy as typing words, but it’s a terrible idea. They say it does everything, but it doesn’t. I doubt their developers are using AI to do “everything”. I usually go in with an idea of what I want and I put the parameters in the prompt. It’s a long prompt, it takes time, but it helps me think about my project too. Then I pick the good from what the LLM generates and discard the nonsense.

And most of the essays/articles I wrote I did with AI help. Not to do the actual writing, because it doesn’t capture my voice, but for the research. I was able to get Perplexity to find more socialist sources (and even vietnamese sources that it translated for me for an essay) and then pick from that. It helped me learn things myself, I know a lot about Vietnam’s independence struggle now and still do.