A lot of us here hate AI because of how it was built: training data gathered without the creators’ explicit, opt-in consent, data centers that negatively affect communities’ access to clean water and energy, a technology design that is inherently prone to hallucinations, etc. At least, those are the main reasons why I hate it.

I think I might actually want to support an AI project if I thought it was being done right. Maybe we could get more people away from exploitative models if there was a non-exploitative alternative.

So what would it take to build AI ethically, in your opinion? And do you know of anyone trying to build AI without these issues?

  • Jared White ✌️ [HWC]@humansare.socialEnglish
    4·
    3 天前

    I don’t agree summarization is an OK use case as a product. Maybe as a one off thing that a user explicitly requests of their own data within an office suite or whatever.

    Meanwhile, I’ve been looking at the mincemeat “Google AI mode” makes of my essays, completely changing the meaning and giving people false conclusions which misrepresent my position. It’s shockingly awful.

    • ladybugs@lemmy.worldOP
      2·
      22 小时前

      Yes, Google’s AI features are horrifying for so many reasons!

      I can’t fathom all of the hallucinated information spread by Google alone, often to people who weren’t even trying to use AI but got an AI overview at the top of their results anyway. Google AI mode is just creating more BS that most people will never notice because they won’t check the original source of the information.

    • cloudy1999@sh.itjust.works
      3·
      3 天前

      Yes, they cannot reason at all, despite clever marketing names like ‘reasoning models’. A responsible operator must verify all output, something humans can’t collectively be trusted to do. Even when verification is performed, we must ask ourselves if ‘old-fashioned’ thinking wouldn’t have given a just as good or better result. IMO, it’s hard to find anything positive about this technology.

      Something related I’ve been thinking about: they’re unable to produce truth or lies, only output.

      • ladybugs@lemmy.worldOP
        2·
        22 小时前

        You hit the nail on the head. They produce output that mimics the appearance of a thoughtful response, but isn’t that at all. LLMs do not actually think and do not have any concept of truth.

        This is probably why things like ClickUp naming their AI tool “Brain” annoys me so much. It’s designed partially as a way for organizations to get aggregated access to the major LLMs. So yeah, my former coworkers are getting LLM output from “Clickup Brain.” What a marketing scam.

        I’ve been wondering how people’s attitudes toward LLMs would shift if society collectively changed the language we used about them to be more accurate. Maybe there wouldn’t be so many people claiming “AI is great for research” and whatnot. Even then, though, I doubt people would fully get past the human tendency to trust confident-sounding language.