• addie@feddit.uk
    link
    fedilink
    English
    arrow-up
    11
    arrow-down
    1
    ·
    1 year ago

    I’d have to disagree with that. If you don’t have enough trust in your managers to talk to them directly about toxicity, stress, and overload, then how on earth would you trust them to monitor all of your communications to determine the same? I suspect that the actual result would be that all employees would be sure to only discuss sensitive matters in-person or through some non-monitored channel, while looking for another job elsewhere. Also, call me cynical, but I’ve seen enough leadership decisions that are ‘we’ve asked for all these powers, but don’t worry, we promise not to abuse them!’ that they did, in fact, turn out to abuse.

    And after reading all the stories about AI’s copyright-infringing ways, slurping up decades of Twitter and Reddit comments, you’d trust the authors to ‘keep it on site’ and ‘forget everything on demand’?

    • wsippel@kbin.social
      link
      fedilink
      arrow-up
      4
      arrow-down
      1
      ·
      1 year ago

      AIs don’t judge, don’t remember and don’t hold anything against me, so I’d rather have an AI screening my stuff than a human - especially my superiors.

      And yes, I trust an AI I run myself. I know they don’t phone home (because they literally can’t) and don’t remember anything unless I go through the effort to connect something like a Chroma or Weaviate vector database, which I then also host and manage myself. The beauty of open source. I would certainly never accept using GPT-4 or Bard or some other 3rd party cloud solution for something this sensitive.

      • moon_matter@kbin.social
        link
        fedilink
        arrow-up
        3
        ·
        edit-2
        1 year ago

        AIs don’t judge, don’t remember and don’t hold anything against me, so I’d rather have an AI screening my stuff than a human - especially my superiors.

        They do judge, in the sense that managers are going to want statistics and those stats are going be interpreted a certain way. It’s a “numbers don’t lie or show bias, but anything lower than a 7/10 is bad according to humans” situation.

    • Eezyville
      link
      fedilink
      English
      arrow-up
      3
      arrow-down
      1
      ·
      1 year ago

      All your points are valid but you forgot to mention the bias that AI may have. People seem to think that AI is unbiased because it’s a computer but no one thinks about who made and trained that AI. How does it change over time with more input from people? How do you code morality and empathy? How do you account for changing social norms or unrest? How would AI react to people affected by the George Floyd protest of even the War in Ukraine? You can try and train the AI on a company’s culture but every employee has their own life, problems, and history that the AI can’t account for.

      People tend to forget that time Microsoft put some AI on Twitter but had to quickly take it back down.