• elliot_crane@lemmy.world
    link
    fedilink
    English
    arrow-up
    1
    ·
    6 months ago

    We’re looking at how we can use local, on-device AI models – i.e., more private – to enhance your browsing experience further. One feature we’re starting with next quarter is AI-generated alt-text for images inserted into PDFs, which makes it more accessible to visually impaired users and people with learning disabilities. The alt text is then processed on your device and saved locally instead of cloud services, ensuring that enhancements like these are done with your privacy in mind.

    IMO if everything’s going to have AI ham fisted into it, this is probably the least shitty way to do so. With Firefox being open source, the code can also be audited to ensure they’re actually keeping their word about it being local-only.

    • PseudorandomNoise@lemmy.world
      link
      fedilink
      English
      arrow-up
      0
      ·
      6 months ago

      Don’t you need specific CPUs for these AI features? If so, how is this going to work on the machines that don’t support it?

      • lemmyvore@feddit.nl
        link
        fedilink
        English
        arrow-up
        1
        ·
        6 months ago

        You only need lots of precessing power to train the models. Using the models can be done on regular hardware.

      • space@lemmy.dbzer0.com
        link
        fedilink
        English
        arrow-up
        1
        ·
        6 months ago

        Running AI models isn’t that resource intensive. Training the models is the difficult part.

      • elliot_crane@lemmy.world
        link
        fedilink
        English
        arrow-up
        1
        ·
        6 months ago

        With it being local it’s probably a small and limited model. I took a couple courses on machine learning years ago (before it got rebranded as “AI”), and you’d be surprised at how well a basic image recognition model can run on the lowest-spec macbook from 2012.

        • ferret@sh.itjust.works
          link
          fedilink
          English
          arrow-up
          1
          ·
          6 months ago

          Tbh the inversion of typical intuition that is LLMs taking orders of magnitudes more memory than computer vision can mess people unfamiliar up on estimates of the hardware required

      • sacredbirdman@kbin.social
        link
        fedilink
        arrow-up
        1
        ·
        6 months ago

        Nope, they can use your NPU, GPU or CPU whatever you have… the performance will vary quite a bit though. Also, the larger the model the more memory it needs to run well.