If they release models then they will lose control over what people can use the models for, they will face competition from other companies that can host the same models better, and they will lose the ability to increase subscription prices to make the line go up. It's incompatible with America's toxic investment culture, which is why the American government will eventually outlaw foreign models or nationalize the losses of the American AI companies.

  • source
  • parent
  • context
  • [–] 11 points 1 month ago* (1 child)

    They also knew that, ultimately, destroying the economic foundation for humans making content while also making it easy to generate slop by the bucketful, would wreck the internet—and, thus, their own content supply chain. Hecht warned that AI had started a “doom loop” that would “threaten the economic foundations of its essential suppliers” and eventually make the internet infinitely worse (a process for which the media critic Cory Doctorow has created the term “enshittification,” and that you may recognize from searching for… pretty much anything on the internet).

    This is not at all what enshittification means. Doctorow used enshittification to describe the mechanisms by which tech platforms establish monopoly with a good product and then extract value from that monopoly with a shittier product until they die. This was already before modern LLMs rolled around

    But I will say enshittification would be a fantastic word to describe the process above so I can see why the author got it wrong

  • source
  • [–] 23 points 1 month ago*

    i was downvoted recently because i said the ai issue is not about copyright, but labour. a lot of people parrot what seem nintendo talking points just because of ai but the issue is not that they used content without attribution, but about the expropriation of labour this is connected with. deepseek also infringed copyrights to train their models, but theirs is open source, so that labour theft becomes released to the masses. on the other side, western companies infringe copyrights and lock their software.

  • source
  • [–] 24 points 1 month ago

    TLDR: wait for 2028

    The turning point for supply may arrive in 2028. A TrendForce projection presented at the 2026 Future of Memory and Storage conference indicated that DRAM supply could exceed demand by about 2% that year.

    Until then, AI servers will continue competing with conventional products for a limited pool of DRAM wafers.

  • source
  • [–] 10 points 1 month ago (2 children)

    By trying everything they could to force dominance they have pushed many many people away from US AI.

    Funny enough if they had let capitalism go normally and let true competition go they might have ended up truly dominant.

  • source
  • [–] [S] 2 points 1 month ago

    I'm saying that they might've just made a fresh repo and didn't carry over commits they did locally while fleshing it out. I've done that plenty of times before. And sure, it's a good idea not to rely on fresh off the press projects, but I never suggested otherwise here. I just thought the project was worth a look.

  • source
  • parent
  • context
  • [–] 3 points 1 month ago (1 child)

    ?

    Publish date has nothing to do with commit timestamps

    Look, all I’m saying is, normal cybersecurity rules still apply. Don’t go running off to download and run code that’s hot off the press, with no track record, no established user base, developed by a rando who is visibly moving at a breakneck pace. Especially with any category of software that’s notoriously difficult to secure.

  • source
  • parent
  • context
  • [–] 5 points 1 month ago*

    So if your phone is hot it could be:

    • working as intended,
    • working as intended but bad at it,
    • or working as intended but working really hard right now.

    How is this, in any way, useful information?

    To quote a famous line from Billy Madison: "Everyone is this room is now dumber for having listened to it."

  • source
  • view more: ‹ prev next ›