Top executives at leading Western AI companies are increasingly warning about the safety and national security risks posed by Chinese open-weight frontier models. What they tend not to mention is that these models are improving rapidly and, because they are freely available, pose a serious threat to Western labs’ business models.

The most powerful models from U.S. labs such as OpenAI, Anthropic, and Google DeepMind are closed, meaning the parameters that shape their outputs are kept secret.

  • midribbon_action@lemmy.blahaj.zone
    link
    fedilink
    arrow-up
    12
    ·
    9 hours ago

    [Open weight] models are improving rapidly

    Aka: not profitable. Show me the profit! I’m sick of the open/closed debate, this technology is not ready in any form. No ‘improving rapidly’, no ‘growing revenue’, fucking profit. We’re deep into an ai bubble without any profitable ai company, open or closed. We are years in without any proven productivity increases, no software improvements, no hardware improvements. Everything has gotten worse: software is more brittle than ever, SaaS offerings are less reliable and functional, and every new consumer product is a shitty new way to farm more data. Oh and the internet is ruined.

    The solution to corpo-llms being shoved down our throats isn’t to run a bunch of open llms instead. This tech is not nearly as useful as anyone says it is, if at all.

    Also, it’s weird to avoid talking about France and mistral in an article about open weight models.

    • Hegar@fedia.io
      link
      fedilink
      arrow-up
      1
      ·
      7 hours ago

      Show me the profit

      not nearly as useful as anyone says it is, if at all.

      It’s not about profit, it’s about control.

      Yes, for many users ai is just an inaccurate summary generator. Or if your job involves creating pointless bullshit reports (as most do) you can save some time - big fucking deal, for sure.

      But as a tool of power and control they are brutally useful. Llms can replace the search engine and the site - google, wikipedia and reddit controlled by the same gatekeeper. Want information disappeared? Way easier.

      They’re also the best way elites have to weaponize vast resources in the information age. Being able to instantly create and automate thousands of fake social media accounts to argue down your opponents is now a power the rich have - they don’t have to buy those shills and keep them happy and active the old fashioned way.

      Saying llms are not useful is like telling a lord their castle is a waste of stone and labour. It’s literally power, and they know it.