Every doctor in my healthcare network uses this and they all side eye me when I say no. I get it, it takes the tedious boring part of the job and basically reduces it by 90%, but fuck you buddy you can never even answer basic questions about the model like if it uses data fed into it for refinement and often don’t even know what that means. You’re a goddamn doctor, read up on this shit for a few hours before blindly accepting it
Fwiw I run a small behavioral health practice. I get emails from people selling this shit like 2-3 times a week. There aren’t that many companies doing it but they’re super aggressive about sales. They’ll have a rep email you over and over and over and when you tell them to fuck off they’ll just have a new rep start.
There are models for this that are better in theory, they require you to type rough notes which are refined which sidesteps the gigantic issue of transmitting such sensitive audio. Some even guarantee that the model used for generating notes is independent of the model used for reinforcement learning and refinement, eg the notes I would feed in don’t get used to improve unless you consent to being used for model research. But even with that I’m not comfortable here because I don’t trust these companies at their word. Trialing systems does show it can be very effective though and I fear this is inevitable as again it reduces administrative (read non billable) burden by a tremendous amount
Also for some reason many of the companies offering systems that audio record sessions, transcribe, and summarize (over just summarizing text notes) are Israeli and that is sketch to me. I don’t trust such intimate data with a product from a country that is hostile, violent, and known for aggressive espionage. This is behavioral health though where EMR solutions are often more barebones for small practices, your local hospital probably uses epic or oracle (who don’t bother with small practice stuff) though I’m not sure if they develop LLM stuff in house or just license it