• xthexder@l.sw0.com
    link
    fedilink
    English
    arrow-up
    13
    ·
    9 months ago

    ChatGPT does very well at summarizing existing information. I haven’t seen it change the meaning of some input text yet. But if generating new text, sure it’s an issue then.

    • oce 🐆@jlai.lu
      link
      fedilink
      English
      arrow-up
      7
      ·
      9 months ago

      Summarizing a news article and a scientific article is a bit different, the later needs to give the essential information of the article with perfect accuracy and the language used in scientific literature is very precise.

      • macarthur_park@lemmy.world
        link
        fedilink
        English
        arrow-up
        5
        ·
        9 months ago

        I agree that anything generated by an LLM will need human verification for accuracy, but scientific abstracts are one of the few areas where LLMs can be immediately useful. I’ve played with having ChatGPT create abstracts in subjects that I’m familiar with and it’s been surprisingly good. It even respects word limits. If you have a block of text that needs to be condensed, ChatGPT can probably get you most of the way there.

        For example, I had chatGPT summarize your comment in 5 words:

        “Summarizing scientific articles demands precision.”

    • duckington@lemmy.world
      link
      fedilink
      English
      arrow-up
      4
      ·
      9 months ago

      It misses what details are important most of the time or just does a very inelegant job. Even when using gpt4 I still don’t really use it for direct rewrites when it’s something important