Dashing the hopes of liberal arts professors and teachers, OpenAI decommissions its AI content detector due to poor accuracy.
  • einsteinx2@programming.dev
    411·
    1 year ago

    I’m glad they shut it down. An inaccurate tool is worse than no tool, especially if teachers are using it to check student essays and punishing students for false positives…

    Even just more generally, people were trusting these detection tools not realizing how inaccurate they were, which causes huge problems both due to false positives and false negatives. Better to remove the useless tools now and work on a better solution, if one is even possible which I’m not sure it is.

    • ExperimentalGuy@programming.dev
      11·
      1 year ago

      I read the article and was like huh, I’m kinda sad that they shut it down, but I really appreciate it point. I’ve been wondering a lot ab how this will effect academics and frankly, like most things nowadays, just seems futile.

  • Peregrinus@lemmy.world
    211·
    1 year ago

    positive outcome, it was misleading at best and causing people to be given poor grades or employment issues at worst. I’m sure others here tried to use it, I found it couldn’t detect raw chatgpt 3.5 copy and pasted. it was worse when I asked it to use British English etc.