H2O COLORS

When you're in school, the cognitive tasks are there for the explicit purpose of being brain exercises. It's weightlifting. It is FOR building your mental muscles and making you a stronger thinker and planner. "Offloading the cognitive tasks", then, is just Not Doing The Weightlifting. What happens when you pay for your gym membership and just stand around messing around on your phone? Nothing. Nothing happens. Just money leaving your wallet. Nothing else.

Using AI is a short term pleasure that is going to fuck you over in the long term, and by the time you realize that you didn't build the necessary muscles you need for the cognitive tasks required of your ACTUAL JOB (or, like, adult life in general), it's going to be too late to do anything about it... except going back and doing the real work all over again to get you up to speed.

As you probably already know, AI is just a statistics machine. It's given like billions of example scenarios and it then identifies (like millions of) somewhat random "key" metrics, which it finds patterns in to run predictive analysis off of. You descibe your job as handing over tools that you identify off of memorized knowledge (easy for a computer) and context information from the surgery and knowledge of the doctor (small enough frame of context with clear visual signals, so we'd get extremely accurate). The only part here that would be hard is the handing bit. I imagine having to take the tool from a set location would be nonfeasible for the doctors themselves and so it'd make sense with the current limitations of technology to make an AI drawer system that spits out the right tool and hands it to a more replacable assistant who only needs to hand it to the doctor. In the rare situation the doctor wants something unpredictable there could be a vocal processing component so it spits out the requested tool.

There are still two main limitations to AI. (1) situations in which it makes no sense to be using predictive statistics in the first place (literally most of the current usages of chatgpt), and (2) situations in which you need someone to be liable for failure since you can't meaningful punish (ie. throw under the bus when sued for malpractice) an AI. In your case if the AI augmented assistant hands the doctor the wrong tool the doctor can still be held responsible for using it, so we aren't saved by the hospital system needing a human to strategically fire later.

But surgical sets don’t unpack and organize themselves, patients don’t drape themselves, instruments don’t clean themselves over the course of the surgery, tissues don’t retract themselves, sutures don’t load themselves, OR lights don’t adjust themselves (if you can write a program for that, I’d actually appreciate that, I hate having to readjust the spotlights), patients don’t clean or bandage themselves or transfer themselves from OR table to bed on their own under anesthetic and you still haven’t solved the human component of critical thinking in high paced scenarios where waiting for a vending machine to spit out, say, a hemostatic, can mean the difference between life and death.

Your cabinet idea is, interesting, but I think you vastly underestimate the logistics. The cabinet can know the tools for the case sure but if it’s not sterilely 3d printing these supplies, it would have to be the size of 2 ORs or more just to supply one OR with instruments for one specialty (and that’s if it’s general surgery, if it’s ortho, with reps and vendors involved, bump that supply requirement up to 4 ORs if you want to store everything on-site and still have the drawer dispense it) if it’s going to work the way you describe in the event of something unexpected being needed, and that’s without touching the sterility issues, since you’d have to hire a sterilization tech just to keep one of these things stocked and you’re still not have replaced the surgical assistant who is unpacking what it spits out (btw by spits out are you envisioning a conveyor belt situation to transport because those could compromise the sterility of packaging if anything gets stuck), organizes it, keeps it sterile, and then hands it over.

I've been thinking about how this cabinet would work all morning and at this point of the iteration I was thinking it would be layers of scanned drawers with a monitor, where the computer would track which tool was which and which was where and then on the monitor tell you which drawer to open and visually highlight the correct item on a real-time feed of the drawer so that theoretically you wouldn't need to know which item is which or track where they currently are yourself. In my head it would be watching the doctor to identify which tool they wanted to be handed next and predict that for you, ideally faster than you or the doctor even realise what their next step is. Re-reading your response I think there was some confusion as to the task I was thinking would be replaced here--someone like you would still be supplying the cabinet before each surgery based off a pick ticket. The cabinet would be tracking the tool locations so you don't need to organize them (just clean it sounds like), and tracking the surgery to let you know which needs to be handed over when (and where it went).

You've added on a lot more tasks of your job that a computer very much can't do, but I think a part of this discussion needs to be that those skills aren't in any way fairly valued right now. In many ways, manual labor, say havesting on a farm--should be more valuable than being middle management... but it's not. Farm worker are certainly producing more actual value to society than middle managers (in some ways I think if all middle managers staged a strike and stopped working for a month the world would run better), but we pay them way less. Often criminally less.

Memorizing tools and predicting which tool is next may not be the most valuable part of your job, but it seems like the largest intellectual labor component, and if we take those all away you end up in the blue collar worker class: doing incredibly valuable work without pay or recognition. I don't think AI in a hospital setting will ever reduce the number of staff actually in the building--I fear instead it will reduce the status of many more jobs down to that of the receptionist and custodian. They do extremely valuable work and the hospital would not run without them, but because they are not knowledge workers they are viewed as more replacable and paid as such.

When you're in school, the cognitive tasks are there for the explicit purpose of being brain exercises. It's weightlifting. It is FOR building your mental muscles and making you a stronger thinker and planner. "Offloading the cognitive tasks", then, is just Not Doing The Weightlifting. What happens when you pay for your gym membership and just stand around messing around on your phone? Nothing. Nothing happens. Just money leaving your wallet. Nothing else.

Using AI is a short term pleasure that is going to fuck you over in the long term, and by the time you realize that you didn't build the necessary muscles you need for the cognitive tasks required of your ACTUAL JOB (or, like, adult life in general), it's going to be too late to do anything about it... except going back and doing the real work all over again to get you up to speed.

As you probably already know, AI is just a statistics machine. It's given like billions of example scenarios and it then identifies (like millions of) somewhat random "key" metrics, which it finds patterns in to run predictive analysis off of. You descibe your job as handing over tools that you identify off of memorized knowledge (easy for a computer) and context information from the surgery and knowledge of the doctor (small enough frame of context with clear visual signals, so we'd get extremely accurate). The only part here that would be hard is the handing bit. I imagine having to take the tool from a set location would be nonfeasible for the doctors themselves and so it'd make sense with the current limitations of technology to make an AI drawer system that spits out the right tool and hands it to a more replacable assistant who only needs to hand it to the doctor. In the rare situation the doctor wants something unpredictable there could be a vocal processing component so it spits out the requested tool.

There are still two main limitations to AI. (1) situations in which it makes no sense to be using predictive statistics in the first place (literally most of the current usages of chatgpt), and (2) situations in which you need someone to be liable for failure since you can't meaningful punish (ie. throw under the bus when sued for malpractice) an AI. In your case if the AI augmented assistant hands the doctor the wrong tool the doctor can still be held responsible for using it, so we aren't saved by the hospital system needing a human to strategically fire later.

think of 100 °F as MAX HOT and 0 °F as MAX COLD. typically central US will get close to but not hit both ends every year. if you get closer to the poles or equator (Maine, Florida) you will definitely see temps below 0°F or above 100°F, but just by a bit. humans can tolerate cold better than heat because coats exists for cold, but tearing your skin off does not work for heat, so when you reach the middle you'll note the tolerable weather range has more cold in it than heat. room temp is around 65°F. interestingly 65% is also passing on standard US grading scales which is probably why we can easily think of it as % hot. 65°F is just good enough hot.

-30°F, weather should not do this

-10°F, do not go outside if you can avoid it

10°F, really cold, but if you have a good coat you can wait outside for the bus

30°F, freezing

50°F, totally fine, a bit chilly

70°F, warm

90°F, really hot, but if you hydrate you'll be fine

110°F, do not go outside if you can avoid it