• Gladaed@feddit.org
    link
    fedilink
    arrow-up
    39
    arrow-down
    2
    ·
    26 days ago

    Using computers and simulations is good, actually.

    Vast parameter spaces cannot be robustly explored by experimental means. Medicine has been using ML for a while now and for good reason. AI and ML is much more than LLMs or whatever the current poster child is.

    • UnderpantsWeevil@lemmy.world
      link
      fedilink
      English
      arrow-up
      1
      arrow-down
      18
      ·
      edit-2
      25 days ago

      Using computers and simulations is good, actually.

      Using narrow models and simulations without adequately testing in live humans or monitoring against side effects or investigating problems is how you get Phen Fen and Oxycotin.

      Vast parameter spaces cannot be robustly explored by experimental means.

      Leaving us with “cutting edge” pharmaceuticals with all sorts of unexpected (or deliberately omitted) hazards.

      You still have to do clinical trials, even after you’ve built up a backlog of possibilities.