Chapotraphouse

I'm not against small scale, decentralized or even local, specialized models that use little training data and don't hurt the environment too much. But we must have a strategy for how to fight AI in the imperial core. And there are many reasons, the last and most pressing of which, no one even talks about:

• AI (meaning current LLMs) is born out of the failed attempts to stop deindustrialization. Since Obama, immense state funding attempted in vain to bribe capitalists to stop deindustrialization. Instead they poured it all into the most capital intensive technology ever that, best case, will lower the profit rate even more, further increasing deindustrialization, but will much rather create a massive, destructive crisis, fueling monopolization and fascism (but also potentially raising consciousness).

• The main motivation fueling this investment, is the desire to create a weapon against workers. AI is meant to destroy wages and employment, break solidarity and weaken labors negotiating position. To the degree, that AI isn't soly a scam, to the degree that it offers us as marxists any kind of strength, these same capabilities will be wielded as a weapon against us thousandfold. Because we don't yet control it and we won't until the revolution. We have to face that reality. This alone makes sabotage necessary, even if we also harm our own access to this technology in the process.

• Even if AI were a complete scam, devoid of any practical use (it isn't, but it's very much overhyped). Capital will use it's grip on the government to enforce AI's intended effects anyway: by bailing them out when the bubble hits, by making us suffer for it, by forcing AI on us anyway, even if it doesn't increase productivity, simply to save their investment.

• AI is massively destructive on the environment. (I don't think I need to elaborate)

• AI has a massive, almost completely untapped potential for capitalist propaganda. This is the thing no one talks about. As marxists, we often falsely flatter ourselves to be immune to capitalists propaganda. Western Marxism and compatible leftism, consciously build up by the CIA over decades proof how wrong that assumption is. Yet it was done by a few targeted investments, sponsorships and endorsements. AI is on a completely different scale and I'm not just talking about biased training data. It can create a profile on you, link it with data from intelligence agencies and data miners, subtly nudge you to share your deepest fears, anxieties and traumas without even noticing. You will provide immediate feedback through your prompts, offering clues about your mood and thoughts, enabling learning against you. It will first be used to sell you junk, to create desires in you with hidden advertising. But sooner rather than later, the CIA and other intelligence agencies will be handed control over western AI and buy or force control of much of non western AI. You know they will. Don't wait for the reveal in fifty years. They will spend billions on it. And they will absolutely use it against communists. That's their whole deal. They will use it to find us, to turn us, to crush our hope and solidarity, to make us despair and turn towards wrong strategies that are doomed to fail. By subtly engineering answers and learning from feedback over months and years. And that's just marxists. Imagine what they can do to average workers. Fascism could be turned up on a literal dial.

But we can still win the fight for control over AI. And we can use this struggle to win other fights. Class consciousness around AI is incredibly high and rising. We need to agitate around struggles against data centers. And the bubble will burst sooner or later, we can use this to our advantage. AI won't disappear, so we need to gain leverage with strikes, force transparency and local models without hidden pre-prompts, restrict which professions introduce AI and on which timeline and wage struggles around wether workers or capitalists get to reap the benefits and pay for the inevitable shortcomings and failures of AI. Sabotage and rejecting AI as whole are at this point a valid strategy in this fight to win time and gain ground.