• 0 Posts
  • 42 Comments
Joined 2 years ago
cake
Cake day: June 12th, 2023

help-circle

  • Worse - easy way to set the settings then gaslight the user to say they asked for it that way.

    How much you wanna bet that it makes those changes in a way that is generally indistinguishable from as if it was done by the user’s own credentials? (Except save perhaps in recall’s own logs)


  • I appreciate the sentiment - in fact it’s been fun watching AI being integrated into home assistant by end users and being given full control, lots of incredibly interesting times.

    But not all AI is the same. Somehow I expect that Microsoft’s implementation will make it ridiculously easy to opt-in to Microsoft services and relaxed privacy settings, but will leave opting out as an exercise left to the user.



  • TeddE@lemmy.worldtoTechnology@lemmy.world*Permanently Deleted*
    link
    fedilink
    English
    arrow-up
    21
    arrow-down
    3
    ·
    2 months ago

    Large language models (LLM) are the product of neural networks, a relatively recent innovation in the field of computer intelligence.

    Since these systems are surprisingly adept at producing natural sounding language, and is good at create answers that sound correct (and sometimes actually happen to be) marketers have seized on this as an innovation, called it AI (a term with a complicated history), and have started slapping it onto every product.
















  • Emphasis on by comparison, as in “molten hot metal is cooler than the surface of the sun, by comparison”.

    TikTok and Temu actively have code in them that would be considered a virus in other contexts. They exploit your system to gain more access than they should, violating the point of sandboxed access.

    By comparison Meta and Google merely take advantage of user ignorance and apathy by making opting out frustrating - but still technically doable.

    Both practices are terrible, but that’s not the same as saying they’re equally bad.