Comment on ChatGPT plays doctor with 72% success

CyberCatBytes@kbin.social ⁨1⁩ ⁨year⁩ ago

• hydroel@lemmy.world ⁨1⁩ ⁨year⁩ ago

  Unfortunately, if the data is biased, the model is biased.

  source

  • Contramuffin@lemmy.world ⁨1⁩ ⁨year⁩ ago

    Yes, that’s something that’s constantly emphasized in scientific research. You might have the most infallible algorithm, but… garbage in, garbage out. You’ll still get garbage data if what you enter into the algorithm is garbage

    source
• Domille@sh.itjust.works ⁨1⁩ ⁨year⁩ ago

  I was about to say…

  Wonder what the success rate of doctors is. I’d be surprised if it is above 70% lol

  source
• stopthatgirl7@kbin.social ⁨1⁩ ⁨year⁩ ago

  Or if it’s better at diagnosing minorities, too.

  source

  • cooopsspace@infosec.pub ⁨1⁩ ⁨year⁩ ago

    It almost certainly won’t, but it’s nice to hope.

    It might remove the face to face human bias of a GP but it doesn’t make up for the decades of preconceived or absent research about women or minorities.

    source