By: Emily M. Bender and Decca Muldowney At a recent appointment, Emily’s physical therapist (who knows some about her research) said, “Before we get started,...

At a recent appointment, Emily’s physical therapist (who knows some about her research) said, “Before we get started, there’s something I want to ask you about.” The something was an automatic “scribing” system their office is trialling for two weeks and deciding whether to purchase. These systems take in a (presumably audio-only) recording of the patient encounter and then output a draft patient note for the chart.

So what’s the big deal with “AI” charting? Here are nine reasons why we recommend refusing to consent to the use of scribing tools in healthcare settings:

  • chicken@lemmy.dbzer0.comEnglish
    2·
    2 months ago

    Summarization is known for being one of the things weaker models can handle competently, and there are LLMs with very low system requirements:

    239 tok/s decode on AMD CPU, 82 tok/s on mobile NPU. Runs under 1GB of memory

    I don’t know exactly what the minimum performance a model would have to have to do this specific task, but it seems very likely it could be done by something that can run on a phone, or other low powered hardware.