Article on Doctors Using AI in Practice \ stacker news ~AI

pull down to refresh

Article on Doctors Using AI in Practice www.abc.net.au/news/2025-08-06/doctor-ai-artificial-intelligence-scribe-notes-appointment/105615244

331 sats \ 12 comments \ @OT 6 Aug AI

I thought this was an interesting article about GP's using AI for consultation, diagnosis and federal letters

"It's mostly surprisingly accurate,"

"Occasionally, it will mishear the name of something. Occasionally, it will mishear a diagnosis."

"I have had one letter where I think, 'Oh, I don't think they've checked this properly. They've clearly got one of the diagnoses not quite right,'" he said.

"It's like the GPS in the car. You are still the driver, there's suggestions, but you have to check it."

I wonder how long these Doctors have left in their job...

"All data is protected according to ISO 27K and SOC2 requirements, which are the highest enterprise standards that exist."

Dr Kelly said Heidi Health was audited by third parties to "protect our data and ensure the security that we have".

This would be my main concern with my Doctor using AI. Until the day I can host my own LLM that has a high accuracy rate of diagnosis I'll probably continue using the DR the old fashioned way.

view all related items

0 new comment

111 sats \ 0 replies \ @gmd 22h

I wonder how long these Doctors have left in their job...

Longer than most other white collar jobs I will bet :P I know it is coming though...

I will often have chatgpt and grok tabs open to look stuff up I should already know and between my own brain, I have 3 neural nets to triangulate on the truth.. again it has to be stuff you already know / understand because of the hallucination issue, or they will sometimes just omit stuff or get them wrong.

For me at least in the hospital diagnosis is not usually an issue. 90+% of stuff is simple bread and butter and diagnosis is never in question.

0 new comment

100 sats \ 0 replies \ @kepford 22h

I wonder how long these Doctors have left in their job...

That's been the vision for at least a decade. Back in the days when IBM Watson was new, that was one of the main use cases. You don't see a doctor. You see Watson. Institutionally we've been being prepped for this change for decades with the move in language. You aren't asked about your doctor. Its your provider. You often don't see a doc. You see a nurse. Cost cutting.

Honestly, I'm not against it as I don't have nearly as high of an opinion of doctors as most people I know. Like many professions they have their own bell curve. If you find a good doc you know it.

0 new comment

47 sats \ 5 replies \ @optimism 6 Aug

Data safety is a massive concern.

ISO 27k1 is a process standard (like 9001 and countless others) and not an information security standard per-se. It says something about the security organization and its processes, but very much about the security of your data.

0 new comment

100 sats \ 1 reply \ @kepford 22h

Didn't Trump just announce something about opening up the medical industry to AI. I imagine it includes changes to HIPA.

0 new comment

0 sats \ 0 replies \ @optimism 22h

The "AI Action Plan" mentioned ¹

Launch several domain-specific efforts (e.g., in healthcare, energy, and agriculture), led by NIST at DOC, to convene a broad range of public, private, and academic stakeholders to accelerate the development and adoption of national standards for AI systems and to measure how much AI increases productivity at realistic tasks in those domains.

https://www.whitehouse.gov/wp-content/uploads/2025/07/Americas-AI-Action-Plan.pdf page 5 ↩

0 new comment

153 sats \ 1 reply \ @teemupleb 6 Aug

Instead of “trust me bro we will delete your sensitive data”, I would like to see a method that makes it cryptographically impossible for my data to be leaked. Maybe using a secure enclave?

0 new comment

0 sats \ 0 replies \ @optimism 22h

I've been considering this for a moment and I think that a secure enclave could actually work as an architecture for this. But, as long as someone reverse engineers things like efused keys and doesn't do a press release about obtaining these, I'm not sure how you can be 100% sure with a platform like SGX. It doesn't work well in extreme adversarial settings.

0 new comment

5 sats \ 0 replies \ @optimism 6 Aug

but not very much

0 new comment

42 sats \ 0 replies \ @LAXITIVA 6 Aug

I had a doctor use an ai to record me cause he didn’t want to listen or type and I mumble and am confusing already so I’m sure that wrote some gibberish and made him look incompetent

Footnotes