I used to look forward to emails from Oxford about new textbooks coming out and emerging fields of research.

Now they send me this fucking attrocity of an email about an event for AI “innovation” simulation in Healthcare, generating studies.

It took me a mere 148ms to unsubscribe, the only reason the date isn’t censored is because I hope somebody attends and makes a ruckus. FUCK YOU OUP.

  • Caveman@lemmy.world
    33·
    2 days ago

    I don’t know the specifics but my general opinion is that people should use neural networks to crunch data around probability of side-effects of medication or to diagnose tumors, classify sleep cycles and so on.

    Asking an AI to diagnose a patient with an LLM is very risky. The LLM will predict the most probably cause but will it pick up on nuances in patients appearance to rule out a less common but more serious disease?

    I’m very skeptical about integrating LLMs into the medical field but AlphaFold has moved medicine forward in massive way for example. I want cancer diagnosis and body scans to be filled to the brim with AI but keep it away from my doctor.

    • FiniteBanjo@feddit.onlineOPEnglish
      13·
      2 days ago

      I think we can use precise enough wording to not allow any confusion on this matter. Talking about the few and far between good or effective “AI” uses does nothing positive, it just normalizes the horrible shit going on under the same moniker.