Brand MU Day
    • Categories
    • Recent
    • Tags
    • Popular
    • Users
    • Groups
    • Register
    • Login

    Learning to Code - A Sprite Break

    Scheduled Pinned Locked Moved Helping Hands
    85 Posts 28 Posters 6.9k Views
    Loading More Posts
    • Oldest to Newest
    • Newest to Oldest
    • Most Votes
    Reply
    • Reply as topic
    Log in to reply
    This topic has been deleted. Only users with topic management privileges can see it.
    • D
      dvoraen @Faraday
      last edited by

      @Faraday said in Learning to Code - A Sprite Break:

      @Apos I’m not a big fan of the tone of SO either. I’m just reporting that the consensus among the programming community is that the answers coming out of ChatGPT are not good for any but the most basic software questions. Want to know how to do a for loop in Ruby or a bubble sort? Well-established, well-documented things? (because after all it’s just stealing the documentation from elsewhere) Sure, go for it. But for anything complicated? You’re just asking to be tripped up by an algorithm that thinks it knows what it’s doing but really doesn’t. Just my 2 cents.

      Kind of a tangent, but for English Comp we are to do persuasive essays where the topic is relevant to the degree being pursued (mine is computer science). I chose overreliance on AI as my topic for exactly this (bolded) reason.

      1 Reply Last reply Reply Quote 1
      • PolkP
        Polk @Faraday
        last edited by

        @Faraday I find SO is as likely as not to give you a 5 year old wrong answer. At work I tell people actively to avoid it.

        FaradayF 1 Reply Last reply Reply Quote 0
        • FaradayF
          Faraday @Polk
          last edited by

          @Polk Again, I’m not advocating in favor of SO. I’m just saying that even by the standards of a site that’s got its own set of issues, ChatGPT was generating such trash answers that it got banned.

          1 Reply Last reply Reply Quote 1
          • FaradayF
            Faraday
            last edited by Faraday

            This is an interesting thread assessing an AI-generated image of a lady in a kimono dress.

            For my “knows next to nothing about kimonos” self, I look at it and say: “Oh, hey, yeah that looks pretty good.”

            But as the kimono expert summarizes:

            So you have a white-faced woman wrapped in fabric scrap with odd hair accessories & funeral undergarments; if I ran into her in Kyoto’s Gion at night I would probably freak out! The subtleness makes it almost nightmarish.

            That’s basically how ChatGPT is with code (and a lot of other things). To the novice/uninitiated it may look good on the surface. But when you dig deeper, it’s just wrong on so many levels.

            hellfrogH 1 Reply Last reply Reply Quote 2
            • hellfrogH
              hellfrog @Faraday
              last edited by

              @Faraday I’m not a kimono expert or even novice but I can tell at a moderate glance that’s AI, too. The sleeves, and the way the folds don’t match above and below the belt. That’s the first thing I look at, making sure the borders of things make sense/line up. AI is not good at it.

              fr fr
              (she/her)

              1 Reply Last reply Reply Quote 1
              • First post
                Last post