Google’s Project Jarvis will be shown off as soon as December, when it releases the next version of its Gemini LLM, reports The Information. It’s the latest in a line of recently announced AI agents that can take limited control of a PC.

  • floofloof@lemmy.ca
    link
    fedilink
    English
    arrow-up
    37
    ·
    edit-2
    19 days ago

    The AI is supposed to automate everyday, web-based tasks by taking screenshots, interpreting the information, then clicking buttons or entering text.

    Like Microsoft’s Recall. What a funny coincidence that all these new AI tools need to keep taking screenshots of what you’re doing on your PC. That could reveal all kinds of things. Thank goodness Google and Microsoft have proven themselves to be utterly trustworthy.

    • DaddleDew@lemmy.world
      link
      fedilink
      English
      arrow-up
      10
      ·
      edit-2
      18 days ago

      I’m imagining anything filling text and clicking buttons automatically and it strikes me as extremely annoying and unsafe.

      Filing you tax report? Sorry now you’ve committed fraud because the AI sent false information on your behalf. Went on your banking website? Looks like the AI sent a bunch of money to some random account. Going on Amazon? How weird, the AI has ordered a bunch of useless crap on your behalf.

      Something similar happened in the early to mid 2000’s already. It was browser hijacking malware.

      • MajorHavoc@programming.dev
        link
        fedilink
        English
        arrow-up
        7
        ·
        18 days ago

        It was browser hijacking malware.

        I don’t like when people talk about my friend, Bonzi Buddy, that way! Lol. It’s completely fair and accurate, but I still don’t like to hear it. He was so cute!