House bill 238 to give AI prescribing authority

This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.
Status
Not open for further replies.

voxveritatisetlucis

Membership Revoked
Removed
2+ Year Member
Joined
Jun 22, 2021
Messages
4,923
Reaction score
4,497

Expecting apologies from all the people on here who said I was crazy for saying that AI will replace doctors within the next 10 years. Actually will be much sooner! In fact, I predict that anybody entering medical school in the classes of 2030 and later will never work as an attending physician. Too much money to be saved on not paying physician salaries

Members don't see this ad.
 
Last edited:
Members don't see this ad :)
Vox not this roller coaster again bro
Brace Yourself Here We Go GIF by MOODMAN
 

Expecting apologies from all the people on here who said I was crazy for saying that AI will replace doctors within the next 10 years. Actually will be much sooner! In fact, I predict that anybody entering medical school in the classes of 2030 and later will never work as an attending physician. Too much money to be saved on not paying physician salaries
1. Who's able to get sued under this? The hospital? Open AI? Somehow I doubt that the Altman's of the world really want to wade their way into medical tort law.
2. Even if we entertain your idea that ChatGPT will replace physicians in prescriptive authority, do you genuinely believe or think that it will be able to do complex neurosurgery or transplant surgery cases when ChatGPT exists entirely within the internet????
3. So you think all med schools will close up shop and that this year is that last cycle anyone will ever be able to earn an MD????

Be for real man. Are you even a med student or just a troll?
 
Politicians propose stupid bills all the time that never make it out of committee.

But let’s pretend it does pass—the bill just authorizes states to allow ChatGPT to prescribe if authorized by the states and cleared by the FDA.

No state will want to touch this with a 12 foot pole.

Have you ever worked with the FDA? Do you know how long it takes them to approve things? Do you have any idea how thorough they are? My dad worked with them his whole life on the industry side, submitting the data to them to review. Nothing clears the FDA without good quality data, strong evidence. They are thorough, sometimes too thorough. And we have what, zero RCTs to support AI’s medical diagnostic abilities?
 

Expecting apologies from all the people on here who said I was crazy for saying that AI will replace doctors within the next 10 years. Actually will be much sooner! In fact, I predict that anybody entering medical school in the classes of 2030 and later will never work as an attending physician. Too much money to be saved on not paying physician salaries
Drop out now and go back to commerce, Vox.

Bill hasn't even made out of the House. Senate has to agree and President has to sign it into law.

So right now, yet again, you're whining about something that isn't real.
 
As the White Coat Investor says half-jokingly, sometimes the best retirement plan is an AK-47 and a bunch of canned dog food.

Look, maybe the sky will fall and we’ll all be replaced. I doubt it. But we can’t live as if that’s going to happen. What else am I going to do? The business world is no better off.

If the day comes where for some reason I am not able to practice medicine, the skills that medical training has refined in me (absurd work ethic and the ability to learn anything very quickly) will allow me to succeed in a different industry.

There aren’t many people my age willing to work a 36 hour shift. Worst comes to worst I’ll just outwork everyone
 
What I can’t figure out is why @voxveritatisetlucis is in medical school at all if they really believe AI is going to take over all medical jobs in 20 years. Assuming they graduate within the next 2 years and then only do a 3 year residency and no fellowship, that leaves 15 years to break even on the real cost and the opportunity cost of medical training, Not enough time honestly. If Vox is as financially and technologically savvy as they believe, this is a terrible life choice.
 
I don’t think the sky is falling but i think letting AI write some scripts isn’t a terrible idea. We already let NPs in minute clinics do it. Many other countries let pharmacists have some prescribing authority for simple things.

In truth, a well trained AI will probably be better at abx stewardship for viral URIs than most midlevels. And there’s probably a range of simple low acuity conditions and fairly safe meds that an AI could do quite well and simply refer anything more complex to a real doctor.

I think vox is partly right - AI will almost certainly replace a number of midlevels. It will (already has) change the practice of medicine immensely, but MDs won’t be obsolete until much of society is obsolete.
 
I don’t think the sky is falling but i think letting AI write some scripts isn’t a terrible idea. We already let NPs in minute clinics do it. Many other countries let pharmacists have some prescribing authority for simple things.
This would kill the job market for PCPs. At least half of all visits to your local FM/peds Doctor are for things like pneumonia,UTI,adhd,anxiety etc. without these visits, I doubt primary care clinics would be able to stay afloat
 
Members don't see this ad :)
What I can’t figure out is why @voxveritatisetlucis is in medical school at all if they really believe AI is going to take over all medical jobs in 20 years. Assuming they graduate within the next 2 years and then only do a 3 year residency and no fellowship, that leaves 15 years to break even on the real cost and the opportunity cost of medical training, Not enough time honestly. If Vox is as financially and technologically savvy as they believe, this is a terrible life choice.

I hope that I match surgery, which despite getting all honors except for 1 and having decent research, still may not be likely due to the surge in competitiveness (as more people look towards procedural fields, in my class alone there are 50 planning to apply surgical field or anesthesia compared to only 20 who applied this year)

If I don’t match, then I’ll probably just do EM or something and likely be replaced. But by that time, most jobs won’t exist anyway. The smarter thing to do would just be to ban AI since it’s going to lead to economic chaos not seen since the Great Depression
 
how do you think it'll change physician jobs

also mods how long do I have the ankle tag
Lots of good and some bad.

Already it’s scribing for us, managing inbox messages for some. It writes all my insurance appeals now - just copy paste from gpt. I use it for decision support and evidence review especially for things that are rare. It will soon get good at coding and billing too and for a time docs will make more money.

At least until cms devalues codes as docs are able to bill better since most currently under bill. Some docs like me are coding gurus and we make a lot more than our peers for the same hours of work. With AI, those differences will eventually vanish. So there will be a few years where we’re all killing it, but then CMS will revalue the codes and things will even out again.
 
This would kill the job market for PCPs. At least half of all visits to your local FM/peds Doctor are for things like pneumonia,UTI,adhd,anxiety etc. without these visits, I doubt primary care clinics would be able to stay afloat
I doubt it simply because we already have minute clinics in most pharmacies and yet PCPs seem pretty darn busy to me. I had a new patient who needed one and the handful i recommended couldn’t see a new patient until October - almost 9 months from now. So they’re clearly not sitting around wondering where all the patients went.

You are partly correct though in that it will/already has changed what sort of patients PCPs see. Where I am, very few PCPs can squeeze in same day sick visits, so those folks all hit up the doc in a box (usually an NP) or a minute clinic (always an NP). So the MDs will probably find themselves managing more complex patients, more chronic issues, and possibly doing more procedures. Thankfully, I think the sicker and more complex patient panels will hit as AI scribe and coding software hits its stride, so the burden on the docs won’t be so bad.

At least in my city, there’s not usually much wait for the urgent care or minute clinics, so I suspect those markets may be approaching saturation and I’m in a pretty underserved area. I doubt adding AIs in place of the CVS NPs will change things for my pcp MD friends any more than if 100 new urgent cares opened here.

No, it will make work a bit tougher because you won’t have as many simple visits to pad the day and let you catch up, but it won’t eliminate their jobs entirely.
 
I hope that I match surgery, which despite getting all honors except for 1 and having decent research, still may not be likely due to the surge in competitiveness (as more people look towards procedural fields, in my class alone there are 50 planning to apply surgical field or anesthesia compared to only 20 who applied this year)

If I don’t match, then I’ll probably just do EM or something and likely be replaced. But by that time, most jobs won’t exist anyway. The smarter thing to do would just be to ban AI since it’s going to lead to economic chaos not seen since the Great Depression
Once again I think that you're comically overrating gen surg. competitiveness. And while we're at it, for these "non-procedural fields" (especially thinking psychiatry and the non-surgical parts of OBGYN) do you really think for a second that an AI is going to replace those people? And how will AI conduct any of the procedures an EM doc is expected to do? I don't see Claude or Gemini placing tubes or running a code any time soon, maybe ever
 
also mods how long do I have the ankle tag
2 weeks from the warning.
This would kill the job market for PCPs. At least half of all visits to your local FM/peds Doctor are for things like pneumonia,UTI,adhd,anxiety etc. without these visits, I doubt primary care clinics would be able to stay afloat
vox, you have no idea what you're talking about, but also this bill isn't getting passed. I'm really confused what exactly you're expecting an "apology" for since there is literally nothing here.
 
I don’t think the sky is falling but i think letting AI write some scripts isn’t a terrible idea. We already let NPs in minute clinics do it. Many other countries let pharmacists have some prescribing authority for simple things.

In truth, a well trained AI will probably be better at abx stewardship for viral URIs than most midlevels. And there’s probably a range of simple low acuity conditions and fairly safe meds that an AI could do quite well and simply refer anything more complex to a real doctor.

I think vox is partly right - AI will almost certainly replace a number of midlevels. It will (already has) change the practice of medicine immensely, but MDs won’t be obsolete until much of society is obsolete.

I was thinking the same thing. The most important thing to note is I don't think that AI will replace doctors, at least not anytime soon, but I do think it can be used to augment healthcare. I would be less worried about AI taking over and more worried about health insurance companies continuing to slash reimbursements and dictate how we treat patients.
 
2 weeks from the warning.

vox, you have no idea what you're talking about, but also this bill isn't getting passed. I'm really confused what exactly you're expecting an "apology" for since there is literally nothing here.
True, this iteration won’t pass but you have to ask yourself who is pushing for these type of bills in the first place. I guarantee you people like Musk and other tech oligarchs who bought the presidency will push for this.

 
Why do we have to ask ourselves this question, exactly?
I mean you, personally, don’t have to ask these questions. Presumably from your posts you’ve been an attending for 20+ years and therefore have likely reaped the financial benefits that came with the career for so long. Even if AI replaced all doctors tomorrow, you would likely still be okay financially. However, the rads/im/fm/peds PGY1 or new attending with 500k of debt (that will not get any forgiveness or interest subsidy under trump btw) should definitely be asking these questions and striking while they still have power to guarantee protections against AI like longshore workers, teamsters etc. are doing. Other workers understand that you to obtain protections against AI while you still have bargaining power.

For example, if all Kaiser PCPs went on strike today and demanded pay for 15 years in the event they are replaced by AI, they would likely get it because as of today the system can’t function without them. If they wait 5 years to take action, it will be too late and mid levels + AI or AI alone would have already replaced them. Then they have no bargaining power


Doctors should be more like the workers in the article above.
 
Last edited:
Once again I think that you're comically overrating gen surg. competitiveness. And while we're at it, for these "non-procedural fields" (especially thinking psychiatry and the non-surgical parts of OBGYN) do you really think for a second that an AI is going to replace those people? And how will AI conduct any of the procedures an EM doc is expected to do? I don't see Claude or Gemini placing tubes or running a code any time soon, maybe ever
I mean maybe it’s an anomaly, but many of the people in my class who were set on rads and even some who came in day 1 wanting FM have changed to surgery solely because of AI. In fact, it seems more and more like the only ones who want non procedural specialties are those with trust funds/rich parents since it doesn’t matter if they’re replaced.

At my schools hospitals’, trauma surg, rather than EM, run all codes and do chest tubes and what not. Maybe it’s different at smaller hospitals? But then again now many codes is a 20 bed hospital getting. The EM residents here barely get any experience doing chest tubes or intubations
 
Last edited:
People who think AI is going to take over medicine anytime soon clearly have never worked in an underfunded, low(er) resource setting where we're lucky to have a functioning EMR at times lol. EMRs have existed since the 1970s and yet paper charting still happens. Plenty of my patients can't figure out how to log in to a telemedicine visit. The amount of time I've had to spend explaining how to use a signature pad to sign an electronic consent form is baffling. I have a lot of low literacy patients who can barely use a cell phone. it's going to be a long long time before we cut human doctors entirely out of the process.

At my schools hospitals’, trauma surg, rather than EM, run all codes and do chest tubes and what not. Maybe it’s different at smaller hospitals? But then again now many codes is a 20 bed hospital getting. The EM residents here barely get any experience doing chest tubes or intubations

Trauma runs your medical codes? yikes please no
Trauma vs. EM for running trauma codes is very institutional and has less to do with large vs small. My med school and residency programs do it differently despite otherwise being very similar institutions/both large trauma centers.
 
I mean maybe it’s an anomaly, but many of the people in my class who were set on rads and even some who came in day 1 wanting FM have changed to surgery solely because of AI. In fact, it seems more and more like the only ones who want non procedural specialties are those with trust funds/rich parents since it doesn’t matter if they’re replaced.
How is AI supposed to do a physical exam, again?
 
People who think AI is going to take over medicine anytime soon clearly have never worked in an underfunded, low(er) resource setting where we're lucky to have a functioning EMR at times lol. EMRs have existed since the 1970s and yet paper charting still happens. Plenty of my patients can't figure out how to log in to a telemedicine visit. The amount of time I've had to spend explaining how to use a signature pad to sign an electronic consent form is baffling. I have a lot of low literacy patients who can barely use a cell phone. it's going to be a long long time before we cut human doctors entirely out of the process.



Trauma runs your medical codes? yikes please no
Trauma vs. EM for running trauma codes is very institutional and has less to do with large vs small. My med school and residency programs do it differently despite otherwise being very similar institutions/both large trauma centers.
I suspect trauma does not run “all codes.” Vox is limited by his understanding of different types of hospitals and by different presentations of problems in different areas of the hospital. Trauma is not running codes for STEMI or non-trauma related respiratory failure. As a general surgery resident, a surg res responded to all codes, but to see if they needed a line placed. If it wasn’t a surgical issue and if the medicine team running the code wanted to place their own line or didn’t feel they needed one, we then returned to whatever we were doing when the code pager went off. Additionally trauma codes (which ARE often run by the trauma team at a Level 1 center) are usually a different announcement overhead than someone admitted to the floor. We did run our own codes in the trauma ICU, and did not page these out overhead, because we had all the necessary people right there and because the reason for the code was trauma related. But medical ICU/SICU codes were run by PCCM or anesthesia. Floor codes same.

At a level 2 or 3 or non-trauma rated center, medicine or EM is running all the codes usually. EM is definitely running the ED STEMI/respiratory/non-trauma ED codes.
 
Last edited:
I suspect trauma does not run “all codes.” Vox is limited by his understanding of different types of hospitals and by different presentations of problems in different areas of the hospital. Trauma is not running codes for STEMI or non-trauma related respiratory failure. As a general surgery resident, a surg res responded to all codes, but to see if they needed a line placed. If it wasn’t a surgical issue and if the medicine team running the code wanted to place their own line or didn’t feel they needed one, we then returned to whatever we were doing when the code pager went off. Additionally trauma codes (which ARE often run by the trauma team at a Level 1 center) are usually a different announcement overhead than someone admitted to the floor. We did run our own codes in the trauma ICU, and did not page these out overhead, because we had all the necessary people right there and because the reason for the code was trauma related. But medical ICU/SICU codes were run by PCCM or anesthesia. Floor codes same.

At a level 2 or 3 or non-trauma rated center, medicine or EM is running all the codes usually. EM is definitely running the ED STEMI/respiratory/non-trauma ED codes.
yeah this is more what I'm used to. Where I am for residency (level 1 trauma center) EM leads most of the code 2 or 3 traumas, trauma surg takes over more often for the code 1s/traumatic arrests (though sometimes EM leads these in the beginning too, mostly seems to depend on how much notice the trauma team gets to come downstairs). Where I did med school (also level 1) trauma surgery lead everything trauma related (and the trauma bay was an entirely separate building from the main ED)

AI can't do a bedside thoracotomy, thats for sure
 
True, this iteration won’t pass but you have to ask yourself who is pushing for these type of bills in the first place. I guarantee you people like Musk and other tech oligarchs who bought the presidency will push for this.

But it doesn't matter, because the country will always need physicians.

To wit, once upon a time high res imaging like CT/MRI wasn't readily available across the country, and its availability has radically altered how we manage patients. But it has not replaced having a physician at the bedside and calling the shots on the patient. Doubtless AI will similarly alter our practice. But it isn't a substitute for a physician.
 
I mean maybe it’s an anomaly, but many of the people in my class who were set on rads and even some who came in day 1 wanting FM have changed to surgery solely because of AI. In fact, it seems more and more like the only ones who want non procedural specialties are those with trust funds/rich parents since it doesn’t matter if they’re replaced.

At my schools hospitals’, trauma surg, rather than EM, run all codes and do chest tubes and what not. Maybe it’s different at smaller hospitals? But then again now many codes is a 20 bed hospital getting. The EM residents here barely get any experience doing chest tubes or intubations
I switched from ENT/Ortho level surg sub (have the grades/research/school pedigree to match next cycle) to Rads lmao, think you are way over-hyping this.
 
Another point: Just because AI can prescribe, doesn't mean it's the start of taking over. If done correctly, AI will help us do our jobs, not take them over.

If I was in a chart and say: This patient has hypertension and is failing on metoprolol. Here are their other symptoms and medications. Can you please prescribe to CVS on anystreet in specifictown a three month supply of the next best choice in antihypertensive.

That would be a big help. Plus there would be a third check in place to make sure we are making a good choice to avoid medication interactions.

We can approach this AI thing in several ways, but three main ways as I see it:
1. Denial
2. Doom and gloom
3. Embrace it and apply it in useful positive ways to make things better.

Let's as a group go for #3, ok?

Will it be all AI down the line? I'm not going to say it's not. And yes, there is exponential acceleration of technology as a general rule. But I think AI doing a lot of what we do is way off. I could be wrong, but I don't think anyone knows the answer for sure.

As alluded to earlier, the regulatory component is going to take a lot to get through, especially in the United States. We have to figure out the HIPAA part and the legal part.
 
How is AI supposed to do a physical exam, again?
Be honest, how often do physical exam findings actually inform management of a patient? At least in the outpatient setting, I feel like most of it is performative. For example, a 25 yo comes in with a sore throat and low grade fever and mild cough for 3-4 days, no exudates, no LAD, I feel like 9/10 attendings are still ordering rapid step test even though it’s not indicated based on exam findings.


Same with inpatient. A 40yo with a 2/6 systolic murmur and no symptoms of CHF is likely still getting an echo even though the murmur is benign. At least in my school’s ED, they basically do a CT on everybody with abdominal pain even if the diagnosis is obvious based on clinical findings
 
Be honest, how often do physical exam findings actually inform management of a patient? At least in the outpatient setting, I feel like most of it is performative. For example, a 25 yo comes in with a sore throat and low grade fever and mild cough for 3-4 days, no exudates, no LAD, I feel like 9/10 attendings are still ordering rapid step test even though it’s not indicated based on exam findings.


Same with inpatient. A 40yo with a 2/6 systolic murmur and no symptoms of CHF is likely still getting an echo even though the murmur is benign. At least in my school’s ED, they basically do a CT on everybody with abdominal pain even if the diagnosis is obvious based on clinical findings
You did not answer my question.
 
Be honest, how often do physical exam findings actually inform management of a patient? At least in the outpatient setting, I feel like most of it is performative. For example, a 25 yo comes in with a sore throat and low grade fever and mild cough for 3-4 days, no exudates, no LAD, I feel like 9/10 attendings are still ordering rapid step test even though it’s not indicated based on exam findings.


Same with inpatient. A 40yo with a 2/6 systolic murmur and no symptoms of CHF is likely still getting an echo even though the murmur is benign. At least in my school’s ED, they basically do a CT on everybody with abdominal pain even if the diagnosis is obvious based on clinical findings
So based on your suggestions med schools should stop teaching students how to teach physical exams???? Jfc.

How, exactly, is ChatGPT, which has no human arms or legs, nor ears with which to hear, going to be able to hold a stethoscope to record the murmur for that inpatient? How is it going to be able feel for pedal edema or displacement of the PMI and look for JVP if it has no eyes or fingers? How is it going to be able to hear crackles on inspiration, or palpate for hepatosplenomegaly?
 
So based on your suggestions med schools should stop teaching students how to teach physical exams???? Jfc.

How, exactly, is ChatGPT, which has no human arms or legs, nor ears with which to hear, going to be able to hold a stethoscope to record the murmur for that inpatient? How is it going to be able feel for pedal edema or displacement of the PMI and look for JVP if it has no eyes or fingers? How is it going to be able to hear crackles on inspiration, or palpate for hepatosplenomegaly?
auscultation and palpation, I’ll give you, but it can view images. I’d say it’s on par with most dermatologists when it comes to skin pathology. In regards to murmurs, I feel like most IM attendings can’t even reliably differentiate murmurs. Cardiologists maybe, but what is the difference if most end up getting an echo anyways
 
I like Rads more, the option for procedures is there if I want to do it too via ESIR. And my school is a leader in AI in radiology (they have a whole center for AI in imaging even before all this ChatGPT doom and gloom) and the radiologists working here aren't fearful at all.
They aren’t fearful because they’ve spent the past 25 years making 700k per year and therefore likely have close to 10m saved. they could be fired tomorrow and not notice any change in life quality.
 
Be honest, how often do physical exam findings actually inform management of a patient?

extremely often
I don't need AI to tell me how to manage a patient in status asthmaticus
I do need my eyes and my ears to tell me how hard a patient is working to breath and how bad they are wheezing/much their lungs have opened up and if they need more albuterol or are turning the corner and ready to be spaced out.
I regularly decide whether or not to order imaging based on my physical exam. Quite often not. or if I do order it, it helps me decide which imaging to order
looking in ears helps me diagnose AOM (or choose not to prescribe antibiotics for a perfectly normal eardrum) and also sometimes you find playdough in there that you need to remove
diuretic management is heavily based on my physical exam
"sick or not sick" determinations are so often based on the vibes you get when you walk in the room and lay hands on a patient (the fancy people would call that clinical gestalt, i call it vibes based medicine)

there are 100% parts of the physical exam you're taught in med school that are pointless (I don't remember the last time i actually checked tactile fremitus or whispered pectoriloquy) but there is definitely still an art and purpose to the exam.
 
why do you feel gen surg is so competitive. iirc weren't you ages ago in the t20 rat race for md. people in that washing machine would look at gen surg and put it median competitiveness. legit just get like 5-10 research items and a good step 2 score and connect well
Yes, I was and got into multiple the first time I applied. Obviously, if I ended up at a school like Yale, then I wouldn’t be so worried but from my school specialties like gen surg, ob, anes are solid matches and definitely not guaranteed. You definitely can’t just coast and match these specialties. Most do end up matching in normal years but if 25 this cycle vs 12 in a normal year are applying surgery, then I would assume it will be a lot tougher. Specialties like ENT, plastics, optho, neurosurg are notoriously difficult from my school and are generally reserved for people with perfect everything plus research year plus personal connections and luck
 
Be honest, how often do physical exam findings actually inform management of a patient? At least in the outpatient setting, I feel like most of it is performative. For example, a 25 yo comes in with a sore throat and low grade fever and mild cough for 3-4 days, no exudates, no LAD, I feel like 9/10 attendings are still ordering rapid step test even though it’s not indicated based on exam findings.


Same with inpatient. A 40yo with a 2/6 systolic murmur and no symptoms of CHF is likely still getting an echo even though the murmur is benign. At least in my school’s ED, they basically do a CT on everybody with abdominal pain even if the diagnosis is obvious based on clinical findings
Every surgical patient.
 
auscultation and palpation, I’ll give you, but it can view images. I’d say it’s on par with most dermatologists when it comes to skin pathology. In regards to murmurs, I feel like most IM attendings can’t even reliably differentiate murmurs. Cardiologists maybe, but what is the difference if most end up getting an echo anyways
How would AI know there is a murmur to evaluate?
 
auscultation and palpation, I’ll give you, but it can view images. I’d say it’s on par with most dermatologists when it comes to skin pathology. In regards to murmurs, I feel like most IM attendings can’t even reliably differentiate murmurs. Cardiologists maybe, but what is the difference if most end up getting an echo anyways

Have you worked with real physicians?

Have you worked with patients? Do you have any idea how relieved patients are when the physician actually “takes a look at them?”

You say you’re a med student but I don’t know a single physician who says the physical exam is worthless. Even psychiatry finds value in their exam (mostly an observational one in an outpatient setting, but an exam nonetheless).

When AI can reliably replicate the physical exam, it’s not just physicians out of work, but by then also all lawyers, politicians, bankers, pilots, insurance agents, salesmen, engineers, etc.

80%+ of people would be out of work with the future you predict. It doesn’t matter if you are in a field AI can’t do—no one is going to want/be able to afford a surgeon (or electrician, etc) when they don’t even have money for food and shelter.

If you really believe this is our future then forget medicine—you should really be training for apocalyptic warfare (Mad Max style), and if you’re lucky enough to make it through that, how to farm/ranch/raise sheep, etc.

I don’t believe that’s where we’re headed, but there is no possible world that doesn’t become apocalyptic if physicians are out of jobs within 15-20 years. Because everyone else is as well and society cannot transition the majority of its people to different jobs that fast.
 
Let's not forget one of the biggest hurdles for AI replacing docs - how does one bill for their services?

Even if AI could replace me today, my hospital wouldn't do it because I generate millions of dollars of direct revenue not to mention ancillary revenue from surgeries and consults. Sure, they'd save a few bucks on my salary, but they're already making a nice profit on me as it is. Maybe they could get a PA with AI to do it, but that PA will still have to be paid and if they're taking on all that work and liability, my guess is they're going to want a lot more money.

The smart thing - and what we're actually doing - is rolling out AI services to MDs so I can be more efficient. Once I have a solid AI scribe that I like, I'll probably extend my clinic hours a bit and still get home earlier. They won't replace radiologists, because you need that MD to read the scan to bill for it, so why on earth would a hospital spend money to replace a huge source of revenue with one that can't bill? But they will and are rolling out products that help with dictations and probably will soon be helping with reads as well.

For the prescribing authority, my guess is that the people driving that are large pharmacy companies. They could drop their minute clinics and have AIs for simple acute visits like that. Remember that AI isn't free, either to develop or to use, so people deploying it like this will need to see profit potential. The pharmacies can still get self-pay revenue to use the AI, and then they also get revenue from selling them whatever meds the AI prescribes.
 
If AI completely replaces physicians, then it's going to replace everything. So the world will be so different, it would be impossible to predict what one should do.

AI is going to create a structural change in the economy in general. Like widespread use of computers, the internet, cell phones, etc, it's going to have a huge impact on everything. Physicians will still be needed - but your work will be different. Hopefully better. We hope that AI will address some of the tasks which are time consuming and unpleasant. Note writing, addressing screening gaps, answering simple questions for patients, etc. The impact may be larger in some fields than others -- radiology for example, I expect AI will get quite good at reading images. I expect radiologists will need to over-read these, but if much of the work is already done for you (i.e. all the malignant lesions have all been measured and documented) they likely will be able to read more studies. Perhaps that means we will need less radiologists -- although I expect the volume of studies to continue to increase so I don't think that will be an issue.

Going on strike to prevent the advancement of technology never works in the long run. Just ask the Elevator Operator Union.
 
auscultation and palpation, I’ll give you, but it can view images. I’d say it’s on par with most dermatologists when it comes to skin pathology. In regards to murmurs, I feel like most IM attendings can’t even reliably differentiate murmurs. Cardiologists maybe, but what is the difference if most end up getting an echo anyways
I always read your posts and find these anecdotes of sweeping generalizations based on your experiences as an M3- not even as a resident or attending- based on your "experience" (as a student) with .01% of attendings in the country.

Nobody at my school even remotely cares about AI other than using it to rephrase our lectures and we certainly aren't freaking out about it taking our future jobs.
 
Be honest, how often do physical exam findings actually inform management of a patient? At least in the outpatient setting, I feel like most of it is performative. For example, a 25 yo comes in with a sore throat and low grade fever and mild cough for 3-4 days, no exudates, no LAD, I feel like 9/10 attendings are still ordering rapid step test even though it’s not indicated based on exam findings.


Same with inpatient. A 40yo with a 2/6 systolic murmur and no symptoms of CHF is likely still getting an echo even though the murmur is benign. At least in my school’s ED, they basically do a CT on everybody with abdominal pain even if the diagnosis is obvious based on clinical findings
pancreatic cancer presents with painless jaundice. you don't even need to lay hands.
 
Status
Not open for further replies.
Top