• einkorn@feddit.org
      link
      fedilink
      English
      arrow-up
      0
      ·
      1 month ago

      Probably the same type of image recognition that volunteers have been training for years on data sets ranging from exoplanets to tumor cells.

        • filcuk@feddit.uk
          link
          fedilink
          English
          arrow-up
          0
          ·
          1 month ago

          That’s like people being worse at riding horses today. We don’t need the skill because we have tools that replaced it and generally do a better job.

          • XLE@piefed.social
            link
            fedilink
            English
            arrow-up
            0
            ·
            1 month ago

            It’s a good thing that we don’t need any doctors to train the AI or verify its results afterwards, because otherwise you’d be dangerously wrong with this comparison.

            • filcuk@feddit.uk
              link
              fedilink
              English
              arrow-up
              0
              ·
              1 month ago

              That’s fair. I’m probably not thinking it through. I can certainly see how this could quickly go downhill. I think we’ll see the reality of it with today’s school generations getting jobs in a few decades.

    • Zarxrax@lemmy.world
      link
      fedilink
      English
      arrow-up
      0
      ·
      1 month ago

      It says it in the article, it just takes some data from the patients chart and calculates a score for the person’s risk of getting an infection.