You know what OP? **** you. I'm a software engineer at google in Pittsburgh offices and i work on machine learning and have a masters in AI theory from Carnegie Mellon University. I'm going to make the ****ing medicinal AI just to **** you over. You're ****ing unemployed kiddo.

You can easily prove the percentages to yourself by looking up doctor error rates, malpractice information...

Your response will probably be that its the claimer responsibility to provide the evidence. The challenge to that is that I simply dont care what you think. Im relaying the factual data that has been clearly stated in case studies. Have you read any of them? ...You arent actually interested in the technology and the truth. You are simply interested in "Yea, but" arguments...Someone provides a rebuttal to you incorrect position and then you simply slide two inches to the left..You dont actually have any interest in the topic. If you genuinely had interest in the topic, I would have no issues looking up dozens of data points for you. Do you think Im going to spend my time providing stats when you arent going to look at them and simply come from another position (that has no basis) and then ask me to provide more data - I'll pass.

You dont have to believe me. The worlds experts are telling you its coming. They are telling you its going to disrupt everything we know and do. They are discussing safe guards. They are discussing the financial impacts (Universal income)....Yet here you on an internet board going - "NOPE - not gunna happen. Gunna be forever before it happens and it probably never will happen as the experts say it will".....But the problem is that its already happened. You can buy AI in a box on-line today. Industries are already taking cost reductions from it. People are already feeling the impact (losing jobs).

You can continue to point out why you dont believe it will happen in the medical community. The disconnect is that you think Im "arguing" that AI is going to revolutionize medicine and while you try to poke holes in it. Thats not actually what is occurring. Im not trying to convince you AI will completely disrupt medicine. Im simply telling you its already underway. Hospitals, insurance companies, research facilities have it running today. They are already looking for ways to use it to revolutionize the industry....If you dont believe those facts, its ok. There were people that thought the idea of a personal computer in your house was laughable because "no one needs a computer in their house" and "computers take up entire rooms"...

There are some really good questions in this post...I will pick out the main topics and address.

Privacy - Lets discuss how vertical specific AI is built at the very highest level for medical. A technology company will partner with a major medical institution that does research. A data scientist will work with the institute to build the appropriate AI queries/code that will run on their massive data store. The doctors will determine what data is of value and the technology company will code AI algorithms to search for these data points. The data scientist will point this infrastructure at the data store it will take months for the systems to find/learn patterns, analyze and suggest automation. The research teams will jointly validate the data...NOW you have a product that provides value. The technology company takes that product to a hospital or insurance agency and lets them know they can reduce their cost by 50% by pointing their data at a set of AI querries. The hospital/insurance agency runs a proof of concept. They apply algorithm to their own data store and they receive massive value. The hospitals data store never left the hospital. No PII data ever left the hospital. No HIPAA laws were violated.

The larger the data store the more value, however. Research universities and drug companies will inevitably want larger stores for data. Its easy to see a day in which hospitals/colleges join together in a big data effort, remove PII data and let AI search massive data stores. These types of efforts would have no PII so I doubt it would impact privacy concerns.


There will definitely be involvement from the Governments of the world to set up rules for AI in regards to public safety (Medical/autonomous driving/flight). Data manipulation is just like accounting manipulation. You can make the "numbers" say anything you want. The goal of AI is to provide the highest percentage of probability and automating the response to this probability. If bad people influence the data that AI provides probability predictions against, AI WILL give you bad data. Im not a huge fan of government regulation but this is an area where we will need Hipaa type guidelines but instead of privacy for data integrity.

What "doctor error" rates? Nearly all errors shown are things like improper or late drug administration or minor errors (or nursing errors) that are largely irrelevant. The patient population is already very sick to begin with and has a poor prognosis from the start.

The probability of a healthy person suffering from actual physician medical error is exceedingly low.

Look i have no doubt its coming. And i know tech is being worked on but i know they havent figured out consciousness FOR US. Then i brought up specific things that are needed, and there are MANY more and all anybody had to say was that "technology improved". Not giving any fuken thought as to how serious this problem is and with nothing CLOSE to a solution. Oh and btw, moore's law isnt always true. The thought being that once you reduce something small enough you cant go any further.
The nvidia ceo or whatever has been saying this and why the gpu's have incrementally improved lately. So its almost like this framework that applied to a certain period. No DOUBT though that massive tech breakthoughs COMPOUND productivity, that is not what im saying

Theoretically, consciousness may not be needed; it also may be needed. But then through that im saying youre changing what it means to be "intelligent". If they are not conscious then they are most likely not autonomous and will have to rely on a programmer/ humanThis is precisely what we're, at least i am (and i believe a few others) are talking about:This in reply to your quantum suggestion

As a radiology brah, all I can say is that I hope I'm replaced sooner rather than later.

lol every doctor thinks they're House

we gonna have scifi autodocs in no time and you won't have to sue them for malpractice because they're thinking about golf or banging nurses while they amputate the wrong leg

Complex issues can be handled by AI. In fact, AI was developed to handle complex issues. For AI it doesn't matter how much medicine someone is taking. AI would be able to cut down 30 meds to 17. For half the price. Just because the memory is larger and a computer can run through everything at night, or any time it is not seeing a patient.
.
And a robot can lift heavier weight than a person. Turning a dekubitus patient every 30 minutes. No need for coffee break or cigarette break

lol @ thinking 99% of doctors aren't 70 hour/week grinders and don't even know how to golf.

Didn't read thread but M.D.s are about to be a thing of the past. Majority of clinics use PAs or NPs. Fuk, even CRNAs administer majority anesthesia in the U.S.

hmm what would the repercussions be if you eliminate such a sought after degree for colleges? med school loses all those students for the med. field and med school tuition revenue

i would imagine research would take a hit too. with the only hope being ai is complete enough to be vastly superior; and we've already highlighted how they prob only have a fraction of that currently

You're right about the 2nd part. The thing is, there are still endless jobs for all doctors and the salaries keep spiking every single year. Primary care is booming hard, even in saturated cities. Whereas PAs/NPs have a very saturated market.

A run of the mill family doctor can make close to 300k if they pursue it (220k in academia, >350k in busy community practice). And they can work anywhere. An NP/PA is making 85-110k usually and now has a very saturated job market.

With that said, everyone knows how big of a problem midlevels are. It's just that the market reality in 2020 is more jobs for doctors and higher pay. All projections point to even more job growth and further pay increase.

Have read the position papers etc easily avail to search online regarding position of AI in Australasian healthcare setting and the message seems pretty clear

AI isn’t being evolved to take over from DOCTORS. it is being implemented to improve mostly the administration side of medicine and also augment clinician based decision.

every paper had the message to some degree that AI should NOT be used as an ALTERNATIVE to clinician based healthcare and that it can one day be used to augment those decisions once data required for the AI to learn becomes globally available (still a long and arduous process)

Basically

AI will be to today’s doctors what current computing tech was to doctors of the 1930’s

To all the doomsday MISC prophets claiming doctors will be bundled out - sorry, your still gonna have to get your rectal exam performed by a person, not an AI-not!

Even if we had this AI today in some isolated markets, it would take years and years before it even sped up our work flow.

For 90% of patients, using AI would just slow us down and drastically decrease efficiency. I already have to input a bunch of crap into the computer to document stuff, now I have an extra step of using AI - for what?

Machines have been replacing jobs for hundreds if not thousands of years in one way or another. Every time, they just increase productivity and create more jobs.

For instance, building cars was one of the biggest jobs in the US right after WWII. Robots came in and by the late 90's most of the assembly line is robots, YET UNEMPLOYMENT IS IN THE 3%'s. People feared computers would eliminate jobs but it turned out that those computers created more jobs than they could have ever imagined.

That's true. But what I'm saying is that most realistic models of AI would drastically slow us down and destroy efficiency if it was actually implemented into routine medicine today. It becomes an additional step that offers nothing for >90% of patients.

Hence there's 2 major issues with the AI people. One, it doesn't actually exist right now in any true form. And two, even if it went through the insanely long decades process of getting to the market; it would drastically slow doctors down for >90% of patients.

Cliffs on 17 fukkin pages?

Pro-AI: Robots will replace all doctors in the next 5 years
Voice of reason: True AI doesn't even exist yet and if it actually did, would take a generation to implement it

Re:Privacy- that sounds slightly less intrusive though you skipped the bit about potentially requiring patients to agree to some kind of medical nanny/spy.

And when someone starts talking about AI wanting "all the data" my mind jumps to general data that is already being collected and from what I've read isn't actually anonymous.

https://www.google.com/search?q=your...hrome&ie=UTF-8

As far as public safety goes I guess it is a matter of who do you trust? I'm not sure how any outside authority could truly ensure that a company that creates this proposed AI/machine learning entity hasn't given themselves a back door in that can be exploited for profit at a later date. And while some governments may be more benevolent-ish others don't seem to be at all. There is a documentary on Netflix right now about how China treated their own populace to enforce their old One Child policy.Those are areas where I can see technology actually being helpful in the medical field. Doesn't require something quite as sophisticated as what has been conjured ITT. Robots to assist with the heavy lifting type physical requirements of patients and some kind of medically optimized voice to text transcription 'bot both seem like good ideas.

Cool story bro... try not to break a bone