I’ve tried several types of artificial intelligence including Gemini, Microsoft co-pilot, chat GPT. A lot of the times I ask them questions and they get everything wrong. If artificial intelligence doesn’t work why are they trying to make us all use it?

        • Feathercrown@lemmy.world
          link
          fedilink
          English
          arrow-up
          2
          ·
          3 months ago

          I find that a lot of discourse around AI is… “off”. Sensationalized, or simplified, or emotionally charged, or illogical, or simply based on a misunderstanding of how it actually works. I wish I had a rule of thumb to give you about what you can and can’t trust, but honestly I don’t have a good one; the best thing you can do is learn about how the technology actually works, and what it can and can’t do.

          • Kintarian@lemmy.worldOP
            link
            fedilink
            arrow-up
            2
            ·
            3 months ago

            For a while Google said they would revolutionize search with artificial intelligence. That hasn’t been my experience. Someone here mentioned working on the creative side instead. And that seems to be working out better for me.

            • Feathercrown@lemmy.world
              link
              fedilink
              English
              arrow-up
              2
              ·
              edit-2
              3 months ago

              Yeah, it’s much better at “creative” tasks (generation) than it is at providing accurate data. In general it will always be better at tasks that are “fuzzy”, that is, they don’t have a strict scale of success/failure, but are up to interpretation. They will also be better at tasks where the overall output matters more than the precise details. Generating images, text, etc. is a good fit.

              • Kintarian@lemmy.worldOP
                link
                fedilink
                arrow-up
                2
                ·
                3 months ago

                That sounds about right. I heard that the recommendation from AI to put glue on your pizza was from a joke on Reddit about how to keep cheese from falling off the pizza. So obviously the AI doesn’t know what a good source of information is from a bad source of information. But as you say something that’s fuzzy and doesn’t need to be 100% accurate works pretty well apparently. Also my logic is a little fuzzy once in awhile myself.