Are these doctors paying for their own locally hosted instances? If not there's a pretty big risk of peoples medical records being injested by the LLMs and spat back out randomly with little or no concern for privacy.

I was offered a permission form at my last trip to the GP. "Do you consent to Ai based note-taking assistants?"

I didn't. I don't because, in addition to Ai being unreliable, it requires we surrender our data sovereignty immediately, since Ai compute space is all in America.

Literally just came from the doctor's office (Sydney) and saw a poster about AI transcription usage on the wall.

Pros I can see:

• More time back for doctors
• This may in turn lead to better patient outcomes.

Cons I can see:

• Higher rates of adverse medical outcomes due to inaccuracies & hallucinations in patient notes
• Higher chance of personal data breaches due to third parties holding and handling private materials.

There are so many issues here. The fact GPs are in a position of thinking this tradeoff is worth it is caused by a cluster of problems. The fact big companies are convincing them this is acceptable is another layer. The history of big tech companies selling such data off to special interest groups (anti-abortion, real estate, etc) a third.

LLM companies are desperate for people to buy their products because nothing is profitable in the AI industry (other than selling the shovels like Nvidia does).

My veterinarian's office started using this recently. I approve for that purpose but definitely not for my own health notes. My PCP has her MA take notes.