He / They

  • 4 Posts
  • 575 Comments
Joined 3 years ago
cake
Cake day: June 16th, 2023

help-circle




  • Throughout history we’ve generally moved towards ever more ephemeral storage media. Stone tablets > clay > paper > etched vinyl > magnetic tape > hard disks > solid state memory, and a bunch more I’m not remembering or not aware of (and there are some that have made extremely long lives their purpose, but they’re not generally in common use, see the etched gold records in Voyager I).

    Arguably, what actually survives is only what a civilization values enough to go back and put into its longer lifespan storage media, it’s only really in the modern era of mass-production and mass-media that the desire to preserve everything has popped up.

    I’m very nostalgia-driven, and I still consume a lot of the media I did as a kid, but my Jellyfin and Kavita servers aren’t meant to represent things I believe should be preserved for others, just for myself.

    I’ve visited the Internet Archive several times (I used to live on 3rd and California, so it was a 5 minute walk), but I think their original purpose of especially preserving the “old internet” blogs and forums and whatnot has somewhat lost its purpose now. It doesn’t archive the mediums where most internet interaction takes place now, like Discord, and it hasn’t for a while (even twitter in its heyday wasn’t being backed up unless you manually requested a tweet be saved).

    There are micro-cultures that absolutely are going to disappear entirely from our cultural record once they become defunct, but that’s always been true, and I don’t think the kinds of cultural records this article is about are really part of our larger “culture” in a meaningful way. MTV News? Gawker? How about SubSim.com and NoMutantsAllowed (2 game-specific forums I’m part of)? Does our proverbial Library of Alexandria really need to remember those existed?










  • The more genAI is influencing our ideas, the less they are our ideas

    This makes no sense to assert specifically about LLMs. Everything influences our ideas. If “influenced by external stimulus or input” == “not your ideas”, then people literally have no ideas of their own.

    Just because the problem isn’t new, doesn’t mean it isn’t accelerating. There are tons of problems that were true when Eleanor Rigby was released that are worse today.

    And not one of those is going to be fixed by getting rid of LLMs, nor are they the product of LLMs uniquely, so sell me on why LLMs uniquely need this outsize focus? To me, this feels like adopting the flavor du jour of liberal media faux resistance that denies individuals tools, in pursuit of some supposed moral high ground, while leaving the structural use of that tool by government in place, similar to anti-gun arguments.




  • transformed into a commodity product

    Which is an argument against commercial AI, not open source AI.

    respecting the wishes of a person that spend a good portion of their life honing a skill and creating artworks

    This is just making a copyright argument from a moral paradigm rather than a legal one. Stealing a physical painting from that artist is wrong. Defacing a physical painting of that artist’s is wrong. Making a digital copy of a digital image is not wrong. You are creating a moral argument for ownership of an IDEA, even if it is a very specific and unique idea.

    “Intellectual property” is not anarchist, no matter how much you insist it’s mean not to treat it as such.



  • isolates us from eachother

    This is a question of how people use it, not something it encourages or forces. “Eleanor Rigby” was written 50 years before LLMs existed.

    makes us more predictable

    What do you mean by this?

    GenAi is currently in use by various militaries for the purpose of killing people

    I think there is likely a misunderstanding regarding what is GenAI and what is not.

    Vision models that e.g. do facial recognition, or direct drones, are not GenAI.

    The only generative AI use I could think of in a “kill chain” is Israel using it for creating lists of people to target, but that system (abhorrently and psychopathically named “Where’s Daddy”) just used LLMs to ‘prioritize’ the names that the IDF and Mossad fed into it, it didn’t actually come up with the names or intelligence. It was about speeding up their killing machine, and has nothing to do with how or what LLMs as a technology do.

    There’s an entire discussion (which has been being had for decades by forensics experts and lawyers and privacy advocates) about people using/trusting computers as trusted black boxes, and this is an extension of that; plausible deniability to do the action you already intended to do via “the computer said so” as a shield.