I understand the consensus at large is no, but what if they could be used for, say, administrative purposes, or some other task which, though not unimportant, might be considered trivial compared to, say, surgery or assessments?

Just a thought.

Also, in addition to whether such technology should be used, do you believe it to be inevitable in any case?

To be sure, I’m not advocating for or against it, merely curious what other people’s opinions are and why they hold such opinions on this potentially inevitable subject

  • dgdft@lemmy.worldEnglish
    8·
    15 hours ago

    I work in the non-LLM biomed space, and there’s a lot of cool stuff happening! Off the top of my head:

    • Vision assistance for pathology: Lots of time is spent by trained professionals scrolling a microscope lens over a slide looking for a needle in a haystack. Having tools that can act as a second set of eyes to flag needles a human might miss is a gamechanger.
    • Segmentation for neurosurgery, radiology, RadOnc: Lots of time is spent measuring the size and shape of organs on CTs and MRIs, to track changes, to plan where to cut during surgery, and to map out tumor targets for radiation therapy. We now have tools that work kinda like the magic wand in photoshop, where you wave your mouse around the margins of an anatomical structure on a 2D image slice and the computer can map out the full shape in 3D.
    • Automated bone age scans: these studies are commonly carried out to assess for developmental + endocrine abnormalities, and in some legal cases involving identity disputes. There’s solid commercial work towards fully automating these studies in a reliable manner.