Advertisement - Members don't see this ad
-
CycleTrack and SDN are teaming up to make medical school admissions more transparent and accessible! Read the announcement to learn how we’re supporting the future of CycleTrack’s free application-tracking tools .
Will doctors become obsolete?
Started by pgoyal
Get help with your application
Use all the free resources available to you from SDN: articles, guides, expert advising, forums discussions, and school research.
Maybe, but then there will be probably new professions. Although, if machines did take over everything, I'd be happy to sit on my ass eating junk food while a robot cures my diabetes and stimulates my dopamine system so it's as if I was having an orgasm 24/7
Yes, you could become an engineer. It's going to be quite some time before machines are designed that design themselves.
Also, I don't think if machines took over everything that they're gonna be your slaves. I think it's going to be the reverse. 🙂
Advertisement - Members don't see this ad
To think physicians are going to be replaced by machines anytime soon is wishful thinking. I don't think the epidemiological data necessary to simply use a computer algorithm to diagnose is even collected (e.g., sensitivity/specificity of symptoms in various diseases). Not to mention the incredible liability such a machine (and, presumably, the company manufacturing it) would bear. Technology can't be used if we haven't done the legwork necessary to tell it how to work.
Computers don't need to have a preplanned flowchart for everything. They are capable of "thinking" and making decisions on the fly like people do.
If people don't need all the epidemiological data to make diagnoses, then neither does a computer. It just needs to mirror whatever thinking process the person uses to come up with their decision. The benefit is, a computer will always have more data and knowledge at hand and readily accessible than a person does. Watson can access over 2 million pages of medical literature in 3 seconds to make a decision. (quoted from the article in the OP) Can't find any people who can do that.
Not a very easy thing to do, but not impossible. The technology will surely get there with enough time and effort on the part of the computer scientists/coders/engineers designing it.
Liability is also an issue, though I don't think a huge company like IBM which will be making billions off of these machines would have a problem paying for malpractice insurance on behalf of their machines.
Last edited:
A stripped down version of Watson kicked the two best Jeopardy players' asses which requires far more than just data input, it also requires speech and language understanding.
Medicine isn't trivia in the form of a question.
Sent from my DROID RAZR using SDN Mobile
I think he was referring to the natural language component, not the trivia component.
It is hard to explain how these things work, you'd have to have a background in AI to really understand.
What the engineers are looking to do is not create flowcharts that can be followed based on inputs. What they are actually doing is taking the thought processes of a human and creating systems that can recreate those thought processes and land on the same/similar decisions as the humans would, given the same inputs. This is something that has been quite successful in the past and will likely to see far greater successes in the future.
Some thought processes are a lot harder to model in code than others since they are not based mainly on logic (logic is the bread and butter of computers). Examples would be musicians, writers, artists of any kind, poets, etc. However, medical thinking is mostly logical. It's based on information available, knowledge from past experience, scientific theory, etc. A lot of it is pattern recognition, maybe even the majority of it (basically everything mid-levels do). Computers are excellent at recognizing patterns even in very large (billions and trillions of points) and confusing sets of data. Very little of medicine is out of thin air and for that reason it's not too difficult to model.
Last edited:
I understand that, but Watson is a far cry from what you are describing here. Watson was programmed with text recognition (it may have had voice on it, but it was voice to text at any rate) and basically had the internet downloaded into it and was able to scan all of its databases when presented with a query and reproduce the best response according to a rule set. Watson is much closer to a search engine than a thinking robot or AI, and as such it is only really capable of trivia. I think we are decades away from an AI or VI (that may only be a video game term... not sure haha) being able to competently perform diagnostics.
Yeah, but the problem is the research hasn't even been done yet - unless there's some database with the sensitivity/specificity of symptoms for every disease I'm unaware of. That's the first thing that would need to be done and is obviously a pretty big undertaking.
👍
That is the other thing. Medicine is constantly changing. Does Jeopardy do current events, or are they mostly historical and cultural trivia? Depending on the fluidity of the programming, by the time they roll out the latest version of Dr. Watson and get the bugs ironed out he may be obsolete again.
No these aren't static systems. The point of an AI is that it can train and learn and improve itself as it works. Watson is also connected to thousands of online databases that are constantly updated, like wikipedia. The healthcare version is connected to just about every information resource we use like pubmed, uptodate, etc and it can get the latest information and incorporate it instantaneously.
It won't be out of date unless all the sources it relies on are out of date. In that case, people would be out of date too, because we also rely on those sources.
There is no need for a database of sensitivity or specificity. That's not how this technology works. It works the way you work. It's built to simulate a human brain. Whatever doctors are using right now to do their jobs is enough for a computer too.
I think you've misunderstood how Watson, and other AIs actually function.
IBM is actually working on yet another project which will simulate the human brain far more accurately. It will supposedly simulate > 100 billion neurons.
http://idealab.talkingpointsmemo.co...er-in-two-years-artificial-human-brain-in.php
http://techland.time.com/2011/10/05/ibm-aims-to-build-artificial-human-brain-within-10-years/
http://idealab.talkingpointsmemo.co...er-in-two-years-artificial-human-brain-in.php
http://techland.time.com/2011/10/05/ibm-aims-to-build-artificial-human-brain-within-10-years/
Advertisement - Members don't see this ad
No they do actually function that way now, they just aren't at the level of a human brain yet. They're part of the way there already. I have no doubt that they will eventually make it all the way or at least come very close. The technology is developing very rapidly.
Last edited:
I'm pretty sure every profession will be obsolete eventually but I don't think doctors will be the first to go. I'd imagine accounting would go more quickly; the technology is already halfway there with TurboTax, you just need some way for all of the relevant information to be downloaded rather than using W-2 forms and so on like we do now.
Except for the fact that you can't use a piece of software to find loopholes or ways to cheat on your taxes. Accountants don't just plug stuff into turbotax, you know... at least not the good ones 😛
We hope that doctors aren't serving a similar purpose.
The first professions to affected will likely be the ones where the most money is to be made, the professions that cost the most to society.
If you build a piece of software that can do the job of someone making $30,000, that's the most you can expect anyone to pay for it (in reality it'll be much less). If you build one that can do the job of someone making $300,000... well that's 10x more lucrative for you. If you ran a huge tech company where would you invest? You also need to consider the demand. If there's a huge shortage in the field, anything you make can potentially sell.
Everyone's got their eyes on medicine- the tech companies, the nurses/midlevels, everyone- because there's a lot of guaranteed money to be made. It's a big pie and everyone wants a slice.
Not to mention that the government is dying to do anything that will bring healthcare costs down. Also the tech companies get a lot of good press from investing in things that "save lives".
Wasn't there a gigantic argument on here about the exact same thing recently...
Yeah, I'd rather not get back into that.
4
45408
The article is kind of silly:
Pilots have physical limitations as to what kind of maneuvers they can pull, whereas a robot can withstand much higher G-loads. Comparing a pilot to a surgeon is actually quite dissimilar. Regardless, surgeons aren't going to be replaced by autonomous robots in my career.
Aren't subspecialists also doctors? 🙄
I can agree with that, but then that leads back to a person wanting advice. A computer doesn't really give advice and apply it to your situation.
I wonder if anyone has told Lockheed Martin or the JCOS this, because I'd be interested to see the JSF program closed out after they spent a trillion dollars on it.
Pilots have physical limitations as to what kind of maneuvers they can pull, whereas a robot can withstand much higher G-loads. Comparing a pilot to a surgeon is actually quite dissimilar. Regardless, surgeons aren't going to be replaced by autonomous robots in my career.
4
45408
Yes, you could.
I guess technically you could, but it would be a lot harder and easier to catch.
Just like it's a lot easier to "cheat" with your patient notes when you're using paper charts than when you're using an EMR that has everything logged down to the second and knows when you edit stuff.
What I really meant was that the software isn't going to be designed for the purpose of cheating on your taxes. That would be illegal. You'd still need someone with the know-how to modify it to do so.
Last edited:
We already had this argument, so I'm not going to get back into this, but this just isn't true. Maybe you haven't used software that does this, but that doesn't mean it doesn't exist. It does, and there's more and more sophisticated stuff coming out that can tailor the advice to be far more applicable to a person's situation.
I have a friend who actually works in this field and I have tested out many of his prototype suggestion algorithms which are very good at synthesizing a lot of information about you and coming up with very customized and applicable suggestions.
What I learned from the last argument on this topic, and why I'm not continuing this one, is that it's impossible to discuss this topic with people who don't have a background in computer science or understanding of computer technology and its capabilities. There is a lot more power to it than you think.
Last edited:
Definately will play a larger role but the subtle forms of interpersonal communication and the nuiasances of the clinical exam will not make doctors obsolete. As well as the interpretation of the algorithim's results in the context of these things. We still dont even have telephone operator machines that can interpret speech that well and thats just dealing with accents/speech patterns. Let alone one that can pick up on nonverbal cues. As well as understand and integrate -relevant cultural/social/economic factors
The day when Windows Help is actually more helpful than a google search I'll believe that algorithms will be better than human doctors.
Building 100 billion fake neurons is great and all. But if they aren't connected exactly correct then they are useless. There are plenty of ******ed people with all neurons intact but can't do much.
I hate this topic. It surfaces every 1 month on SDN. Just when every on this board finishes talking about a ridiculous idea it surfaces again.
As someone with an extensive computer science background and whose career has been technology oriented (and intensive) for the last decade plus, I can very VERY safely agree with prowler. It is more important to understand healthcare at projecting this than the computer science. We have had EMRs since the 1970s (40 years ago) and despite all the OBVIOUS benefits of it, we still don't even have the majority of clinicians using it. While Obama's healthcare plan is going to start forcing more and more to use an EMR, it is still going to be some time before it is standard. Another example, look at the Da Vinci robots, Hansen robots etc. Introduction of these things is very very slow, not because the technology doesn't exist, but because of money, people, distrust etc.
As for thinking computers. Do you remember when Deeper Blue beat Kasparov? Yes, our computer based logic and reasoning has improved substantially, but don't kid yourself. When you get into the real world, you will realize that medical care is far too complex to be replicated by current machines and algorithms. Further, when technology DOES advance to that point, it will be decades if not much MUCH longer before it will be effectively implemented for a variety of non-technology reasons.
As for thinking computers. Do you remember when Deeper Blue beat Kasparov? Yes, our computer based logic and reasoning has improved substantially, but don't kid yourself. When you get into the real world, you will realize that medical care is far too complex to be replicated by current machines and algorithms. Further, when technology DOES advance to that point, it will be decades if not much MUCH longer before it will be effectively implemented for a variety of non-technology reasons.
4
45408
It is now, and it will be for some time to come. When computers can give medical advice applicable to an individual, taking into consideration family dynamics, socioeconomic issues, religious concerns, and the patient's past/present concerns, they will also be able to give dating advice, and a robot will come along and steal your girlfriend.
What I learned from the last argument is that it's impossible to have this discussion with someone who still has a weak grasp on medicine and especially surgery. You truly do not understand what you're talking about.
I'm not in computer science, but I've been avid computer user for almost 20 years, and I build my own computers. Computers have made a lot of progress and are a powerful tool, but they will remain the tool and not the user for the rest of my career.
Advertisement - Members don't see this ad
I do not see what is wrong with having a nurse(or some educated biology or comp science grad) work together with a machine that basically utilizes our textbooks to query for symptoms and to make the differential diagnosis. Sure they will make mistakes. But no more than a doctor. We pride ourselves on our "knowledge" but even those who get 260+ on step1 would be unable to do as well on that multiple choice exam several years later, much less come up with various "signs of disease". "Liability" cannot be used just to justify your high doctor incomes. If you provide less value than other professionals then nobody cares about this technicality.
This assumes medicine is a cookbook.
It isn't.
If it was a cookbook then training wouldn't be 7-10 years.
4
45408
Yes, it can, and it is. It's not the only factor though. The expensive schooling, extremely long training path, grueling hours, being on call, and the stress of trying to save someone's life after they came in with a ruptured AAA are all reasons for a high salary.
Plus, there's more demand than supply.
As someone who (on occasion) beat chess masters in tournaments I can say that the strategies involved in playing chess are much more complicated than medicine. To teach a computer to play chess it is much more than teaching it to solve problems or memorize common moves. I am certain that a nurse with an open access to your textbooks (even if on iphone) could easily get 260+ on step1 or any other medical exam. Because it is all about memorization. But give me access to all the chess strategy books and even let me have a 2nd chess board so I could practice various chess positions and I still wouldn't have a chance against Kasparov. Just like technology would not help me become an artist or a mathematician. By the way if Kasparov did not have a nervous breakdown he would not lose to that computer. His overall game as evaluated by other grandmasters was deemed superior.
4
45408
No, it's not, and it's even funnier to hear this from someone who hasn't even taken it yet.
You certainly do not need to be a doctor to have good interpersonal communication skills. Businessmen on average have better communication skills...
Yes, medicine is all memorization. Remind me how much exposure to medicine you've had exactly? Sorry, but this is just plain ignorant.
Everyone agrees that mcat is much less about memorization that step1. Well I pwned mcat and I feel it had everything to do with memorization. If I had the work ethic to memorize the bible like many people in 19th century, I certainly would not have any problems studying for step1. It's not such a big accomplishment.
You reached this conclusion based on?
Every time I open my pathology text I am thinking about how easily I could be replaced by a computer. Anyway, computer is low on batteries, so back to studying.
So scoring high on step 1 = being a good doctor? Did I miss something?
Okay, this makes sense. You actually know nothing about medicine, but feel like you can accurately compare chess to medicine. Sorry to burst your bubble, but medicine isn't plug and chug. MS1 Pathology and Step 1 are less like practicing medicine than being a lawyer. Alright, I'm done with this. If you don't know anything about the topic being discussed, you should probably refrain from arguing and maybe learn something first.
Doctors aren't the only profession in the world that has long hours and stress. There are many others whose pay isn't even close. I dunno why you all keep using these things as arguments for why you deserve many times more pay.
Sent from my DROIDX using SDN Mobile
If medicine isn't plug and chug then how come I see midlevels plugging and chugging every day? Studies show they handle at least 80% of cases as well as mds do.
Sent from my DROIDX using SDN Mobile
Advertisement - Members don't see this ad
Actually step 1 is mostly recall of facts. The questions are a bit convoluted but they all are based on remembering some set of facts. I know I've taken and passed it.
Sent from my DROIDX using SDN Mobile
You guys seem to be under the false impression that doctors are some kind of superhuman geniuses who handle the most complex issues on earth. Unfortunately that's not even close to the truth. You don't need to be a genius to be a doctor. You just need to have a somewhat greater than average intelligence and be a very hard worker. The second part is even more important than the first. Before you tell me "how do you know"- a doctor who has been in practice for over 40 years told me that. Besides, I've seen with my own eyes people graduating med school who are not the brightest bulbs in the bunch.
The fact is, the vast majority of issues that the average doctor deals with are not very complicated at all. They're actually fairly straightforward. That's why midlevels even exist.
Not like I think doctors aren't smart or important. I'm in med school, I obviously do. But I think we need to be a bit more humble and stop acting like we are Gods ourselves.
The fact is, the vast majority of issues that the average doctor deals with are not very complicated at all. They're actually fairly straightforward. That's why midlevels even exist.
Not like I think doctors aren't smart or important. I'm in med school, I obviously do. But I think we need to be a bit more humble and stop acting like we are Gods ourselves.
Duh. It's the people who are providing the answers. If I have some tech problem with windows, it's humans talking about it on a forum that has the answer. Not window's millions of dollars invested in an absolutely useless Help feature.
You can't compare windows help, which is basically a book in digital form with an extremely simple search function, to an Artificially Intelligent Supercomputer.
Your argument that "windows help sucks so an artificially intelligent supercomputer must also suck" is basically the same as saying a nurse sucks at doing surgery so a surgeon must also suck at it. They are two completely unrelated things.
You think microsoft invested millions of dollars to make Windows Help? That's the dumbest thing I've ever heard. I can't believe anyone can be that stupid. It's just a book in digital form. It doesn't have a brain. It's not artificially intelligent. It's running on your $500 laptop not a multi-million dollar supercomputer. It's supposed to be extremely simple. Do you expect your pathology textbook to diagnose you too?
By the way, how did google find that search you did out of billions of pages on the internet in a few milliseconds. Can you do that?
Also, "windows" isn't an entity, it's the name of a product. The maker is Microsoft. If you're gonna bash something at least get the name right.
Last edited:
They already tried that.... nurses had a sub 50% PASS rate. Best to not be certain about things you are wrong about 👎
Sent from my DROID RAZR using SDN Mobile
