The research from Purdue University, first spotted by news outlet Futurism, was presented earlier this month at the Computer-Human Interaction Conference in Hawaii and looked at 517 programming questions on Stack Overflow that were then fed to ChatGPT.

“Our analysis shows that 52% of ChatGPT answers contain incorrect information and 77% are verbose,” the new study explained. “Nonetheless, our user study participants still preferred ChatGPT answers 35% of the time due to their comprehensiveness and well-articulated language style.”

Disturbingly, programmers in the study didn’t always catch the mistakes being produced by the AI chatbot.

“However, they also overlooked the misinformation in the ChatGPT answers 39% of the time,” according to the study. “This implies the need to counter misinformation in ChatGPT answers to programming questions and raise awareness of the risks associated with seemingly correct answers.”

  • Petter1@lemm.ee
    link
    fedilink
    English
    arrow-up
    12
    arrow-down
    2
    ·
    5 months ago

    I guess it depends on the programming language… With python, I got very fast great results. But python is all about quick and dirty 😂

    • exanime@lemmy.today
      link
      fedilink
      English
      arrow-up
      4
      ·
      5 months ago

      I asked ChatGPT for assistance with JavaScript doing HL7 stuff and it was a joke… After the seventh correction I gave up on it (at least for that task)

    • anlumo@lemmy.world
      link
      fedilink
      English
      arrow-up
      1
      ·
      5 months ago

      In Rust, it’s not great. It can’t do proper memory management in the language, which is pretty essential.

      • Petter1@lemm.ee
        link
        fedilink
        English
        arrow-up
        1
        ·
        5 months ago

        Well, if you use free chatGPT you only have knowledge until 2022, maybe that’s the reason