• Is using generative AI ethical?
  • Is contributing to its development ethical?
  • Why does the Hexbear search function return every single post with “ai” as a substring?
  • MovingThrowaway [none/use name]@hexbear.net
    link
    fedilink
    English
    arrow-up
    1
    ·
    2 days ago

    I don’t see how it’s any different than other technology tbh. Most discussions of ethics in this context are committing some sort of scope error, where the implication is that any one individual’s choices in this regard have a meaningful impact. Either that or we’re talking about some fantasy where the working class has any amount of class consciousness and is able to act as an entity in its own class interest. I won’t fault anyone for avoiding unpleasant vibes, but on an individual level none of that is particularly Marxist.

    Wrt the analysis, in addition to the great points already mentioned in this thread:

    The explosion of the internet created a new sort of frontier with untapped resources and unenclosed commons. Billions of people passively and actively creating art and information and data points for decades, most of which freely given or taken by tech corporations. Part of the trick is that this stuff wasn’t really a resource at the time, not in the sense that it is now. The question “how can someone own my conversations, my habits, my preferences and tendencies and opinions and thoughts” mirrors “how can someone own land?”. A social transformation, a dialectical development, a de- and reterritorialization.

    AI models require a tremendous amount of data to train. One of the LLM models, for example, needs about 70 years of input in order to learn a new language (as an aside, compare that to the ~1500 hours it might take a human). The end result is an incredibly useful and valuable machine, capital, imbued with a tremendous amount of dead labor.

    The high barrier for entry means a further concentration of capital in every industry where AI can effectively be utilized (and isn’t just a gimmick). The AI-owning bourgeoisie are incentivized to heighten this barrier for entry, and this will happen in lockstep with how much doing so decreases their flow of cheap data compared to how much data they need. But the energy and tech cost is already high.

    This abstract notion of “data value” is transforming into a concrete one, and with that comes the enclosure that’s characteristic of a property economy.

    Now, this could be a lot more impactful than many online leftists seem willing to admit, but it’s still taking place in a highly abstracted place with tenuous ties to the material mechanisms and primary contradictions of society. Part of the difficulty in analyzing it comes from the spectacle nature of these abstract realms; they can replicate or imitate similar processes that happen in the material world, like enclosure and exploitation, but that doesn’t reveal what their mechanism in the actual material world is. Does the existence of an AI that can speak French change the flow of resources from Africa to Europe?

    I think its impacts will closely align with the scopes it exists in, so more impactful within the abstractions it operates in, like the internet and media, than base material flows, where it’s mostly just a resource sink. From an international perspective, it’s more like a reorganization of the lord’s manor than an actual restructuring of the system at large. A change in how spoils of empire get divvied up is largely immaterial, in this scope.

    As it consumes its abstract frontier, though, many people that subsist there might find themselves proletarianized (or “materialized”, forced out of the digital proletariat into the manual proletariat, echoing the historic flow from countryside to city). That’s where I find myself personally, with the work I’ve done for 10 years quickly disappearing.