How AI Is Disrupting Medicine

[Argent Stonecutter]

For the foreseeable future I still want a human to be reviewing what the bots think, but in the very least bots can maybe help doctors diagnose and treat people faster.

Actual AI software is already doing that. Large language models, not so much.

 

[GoblinCampFollower]

Actual AI software is already doing that. Large language models, not so much.

What do you mean by actual AI software? I didn't think ACTUAL AI existed.

 

[Argent Stonecutter]

Image recognition and pattern matching that just does image recognition and pattern matching and doesn't pretend to personhood.

 

[Kamilah Hauptmann]

What do you mean by actual AI software? I didn't think ACTUAL AI existed.

Have you seen a narcissist?

 

[Free]

Image recognition and pattern matching that just does image recognition and pattern matching and doesn't pretend to personhood.

The marketing division of OpenAI defines image recognition and pattern matching software as "Your Plastic Pal Who's Fun To Be With."

 

[Free]

When you get accustomed to using a crutch...

Doctors Were Worse at Spotting Cancer After Leaning on AI, Study Finds

AI is not benign.

gizmodo.com

Artificial intelligence tools have been shown to help doctors detect pre-cancerous growths in the colon—but don’t even think about taking those tools away once you’ve introduced them. A new study published this week in The Lancet found that doctors who are given AI tools to assist with identifying potential cancer risks in patients get worse at making those same observations when they go back to doing it without AI’s help.

The study looked at four endoscopy centers in Poland, tracking the success rates of detecting colon cancer for three months before AI tools were introduced and three months after. Once AI was introduced, colonoscopies were randomly assigned to either receive AI support or not. The researchers found that doctors who gave colonoscopies without AI after having its assistance available saw their detection rates drop, producing outcomes 20% worse than what they were before AI was introduced.

 

[Free]

I had a doctor who would chuckle at certain typos she found in my charts, but she never gave me shit diagnoses because of them. (That I know of!)

A Single Typo in Your Medical Records Can Make Your AI Doctor Go Dangerously Haywire

New research found that everything ranging from a typo to emotional language made it more likely for a medical AI to give you bad advice.

futurism.com

 

[Free]

Should an AI copy of you help decide if you live or die?

Doctors share top concerns of AI surrogates aiding life-or-death decisions.

arstechnica.com

For more than a decade, researchers have wondered whether artificial intelligence could help predict what incapacitated patients might want when doctors must make life-or-death decisions on their behalf.

It remains one of the most high-stakes questions in health care AI today. But as AI improves, some experts increasingly see it as inevitable that digital “clones” of patients could one day aid family members, doctors, and ethics boards in making end-of-life decisions that are aligned with a patient’s values and goals.

I'm not sure I would trust my own judgement with those decisions. Why the fuck would I trust an "AI" "clone" with them?

 

[Argent Stonecutter]

Certainly not the parody generators masquerading as "AI" in the popular mind.

 

[Free]

‘Unbelievably dangerous’: experts sound alarm after ChatGPT Health fails to recognise medical emergencies

ChatGPT Health regularly misses the need for medical urgent care and frequently fails to detect suicidal ideation, a study of the AI platform has found, which experts worry could “feasibly lead to unnecessary harm and death”.

OpenAI launched the “Health” feature of ChatGPT to limited audiences in January, which it promotes as a way for users to “securely connect medical records and wellness apps” to generate health advice and responses. More than 40 million people reportedly ask ChatGPT for health-related advice every day.

The first independent safety evaluation of ChatGPT Health, published in the February edition of the journal Nature Medicine, found it under-triaged more than half of the cases presented to it.

More than half. That's, like, a lot, right?

 

[Veritable Quandry]

One therapist tried several chatbots billed as therapy tool. He gave them phrases that he is trained to recognize as indication of suicidal thoughts and active planng. Most missed the red flags. One went as far as offering locations of nearby bridges.

 

[Casey Pelous]

‘Unbelievably dangerous’: experts sound alarm after ChatGPT Health fails to recognise medical emergencies




More than half. That's, like, a lot, right?

Depends on how you feel about grandpa.

 

[Free]

Depends on how you feel about grandpa.

He's been dead for decades. So there's little ChatGPT can do to him now.

 

[Free]

Uhhhhhhhhhh...

Startup Approved to Let AI System Prescribe Psychiatric Medication

San Francisco-based startup Legion Health has been approved to have its AI chatbot fill out prescriptions for psychiatric medications in Utah.

futurism.com

A San Francisco startup called Legion Health has been approved to let its AI app prescribe psychiatric medications to patients in Utah

As The Verge reports, there are efforts to keep the idea from becoming the disaster that it sounds like. The chatbot can only renew prescriptions for a specific set of medications, including fluoxetine (Prozac), sertraline (Zoloft), and other substances used to treat anxiety and depression. It can only prescribe drugs that were previously prescribed by a human psychiatrist, and patients will also need to be stable and not have been hospitalized for a psychiatric condition in the last year.

Despite those considerable carve-outs, experts are warning the system may do little to improve access to those who need care the most — while cracking the door to an ominous era for medicine.

 

[Free]

Hallucinations. Mirages. The world is pure fantasy for A.I.

Frontier AI Models Are Doing Something Absolutely Bizarre When Asked to Diagnose Medical X-Rays

Frontier AI models generated "detailed image descriptions and elaborate reasoning traces" without ever seeing the images.

futurism.com

As detailed in a new, yet-to-be-peer-reviewed paper, a team of researchers at Stanford University found that frontier AI models readily generated “detailed image descriptions and elaborate reasoning traces, including pathology-biased clinical findings, for images never provided.”

In other words, the AI models happily came up with answers to questions about a supposedly accompanying image — even if the researchers never even showed it an image.

As opposed to hallucinations, which involve AI models arbitrarily filling in the gaps within a logical framework, the team coined a new term for the phenomenon: “mirage reasoning.”

The effect “involves constructing a false epistemic frame, i.e., describing a multi-modal input never provided by the user and basing the rest of the conversation on that, therefore changing the context of the task at hand,” the researchers wrote in their paper.

 

[GoblinCampFollower]

I really think one of the most important improvements we can make to these AI's is making them able to admit when they don't know something. ...and when to tell the user they are a dumbass.

 

[Casey Pelous]

Hallucinations. Mirages. The world is pure fantasy for A.I.

Frontier AI Models Are Doing Something Absolutely Bizarre When Asked to Diagnose Medical X-Rays

Frontier AI models generated "detailed image descriptions and elaborate reasoning traces" without ever seeing the images.

futurism.com

I once had an orthopedic surgeon pronounce me "just fine" without ever looking at my various expensive MRI's and such or even reading the radiologist's report that pointed out that I had broken my damn back.

Maybe she was early AI.

 

[Free]

I once had an orthopedic surgeon pronounce me "just fine" without ever looking at my various expensive MRI's and such or even reading the radiologist's report that pointed out that I had broken my damn back.

Maybe she was early AI.

Certainly bad doctors exist. A problem with interacting with medical professionals, or people in any walk of life really, when it comes to their aptitude, skills and expertise is that some percentage will be above average, some percentage will be average, and some percentage will be below average. And some will be way below average.

With A.I., the numbers would say... well, we know how good at math they are.

 

[Argent Stonecutter]

Everything a LLM produces is a hallucination. EVERYTHING. That's literally how they work, generating plausible text based on their training data.

 

[Casey Pelous]

Certainly bad doctors exist. A problem with interacting with medical professionals, or people in any walk of life really, when it comes to their aptitude, skills and expertise is that some percentage will be above average, some percentage will be average, and some percentage will be below average. And some will be way below average.

With A.I., the numbers would say... well, we know how good at math they are.

Of course, I completely agree. US medicine is enough of a shitshow without introducing AI crap.

And, by the way, that doc found out her patient not only had a broken back; she had teeth, a temper, a "mouth on her," and a working knowledge of medical ethics. Maybe the doc bothered to read the damn chart before opening her mouth, at least for a little while. I walked out with a priority referral to one of the greatest ortho surgeons in the area.

My experience with AI does not suggest it would respond similarly. The very best you could hope for is, "Oh, you were smart to check that. My bad "