In order to ensure that this thread continues to meet TL standards and follows the proper guidelines, we will be enforcing the rules in the OP more strictly. Be sure to give them a re-read to refresh your memory! The vast majority of you are contributing in a healthy way, keep it up!
NOTE: When providing a source, explain why you feel it is relevant and what purpose it adds to the discussion if it's not obvious. Also take note that unsubstantiated tweets/posts meant only to rekindle old arguments can result in a mod action.
On July 16 2015 10:38 Belisarius wrote: Just on that, here's a fun little toy that predicts how likely a given job is to be automated. The full document/paper is here.
1. Telemarketers 2. Title examiners, abstracters, and searchers 3. Hand sewers 4. Mathematical technicians 5. Insurance underwriters 6. Watch repairers 7. Cargo and freight agents 8. Tax preparers 9. Photographic process workers and processing machine operators 10. New accounts clerks 11. Library technicians 12. Data entry keyers 13. Timing device assemblers and adjusters 14. Insurance claims processing and policy clerks 15. Brokerage clerks 16. Order clerks 17. Loan officers 18. Insurance appraisers, auto damage 19. Umpires, referees and other sports officials 20. Tellers
The little gadget thing lets you stick any job in and see how it rates. Of course it's just a bunch of researchers' guesses but the results are pretty interesting.
I would think choreographers would be more unlikely?
On July 16 2015 11:01 Belisarius wrote: Yeah it looks quite rigorous. At the end of the day it's still an estimation like any other, but calling it a guess is probably not doing the work justice.
It's gonna be a lot better than my random anecdotal guesses would be, anyway.
I think you're incorrect about programming meaning everything else would be automated as well.
Do you program? In my opinion an AI capable of analyzing problems and coding solutions would at that point be capable of automating almost everything else other than strictly human services (prostitution, counseling, etc.)
On July 16 2015 10:38 Belisarius wrote: Just on that, here's a fun little toy that predicts how likely a given job is to be automated. The full document/paper is here.
1. Telemarketers 2. Title examiners, abstracters, and searchers 3. Hand sewers 4. Mathematical technicians 5. Insurance underwriters 6. Watch repairers 7. Cargo and freight agents 8. Tax preparers 9. Photographic process workers and processing machine operators 10. New accounts clerks 11. Library technicians 12. Data entry keyers 13. Timing device assemblers and adjusters 14. Insurance claims processing and policy clerks 15. Brokerage clerks 16. Order clerks 17. Loan officers 18. Insurance appraisers, auto damage 19. Umpires, referees and other sports officials 20. Tellers
The little gadget thing lets you stick any job in and see how it rates. Of course it's just a bunch of researchers' guesses but the results are pretty interesting.
On July 16 2015 11:01 Belisarius wrote: Yeah it looks quite rigorous. At the end of the day it's still an estimation like any other, but calling it a guess is probably not doing the work justice.
It's gonna be a lot better than my random anecdotal guesses would be, anyway.
I think you're incorrect about programming meaning everything else would be automated as well.
Do you program? In my opinion an AI capable of analyzing problems and coding solutions would at that point be capable of automating almost everything else other than strictly human services (prostitution, counseling, etc.)
I'm a biophysicist (which is luckily pretty high on the "unlikely" scale). I do coding on a pretty regular basis but I would be blown out of the water by an actual CS dude.
Programming is very rules- and function- based, and has a lot of peculiarities that lend it to automation which may or may not apply to other "problem-solving" jobs. I don't believe it's a case of "if AIs can do X they can do anything" because every role has unique challenges.
Often I suspect a given role will turn out to be hard/easy based on some completely random technicality that's not even obvious to people doing it.
On July 16 2015 10:38 Belisarius wrote: Just on that, here's a fun little toy that predicts how likely a given job is to be automated. The full document/paper is here.
1. Telemarketers 2. Title examiners, abstracters, and searchers 3. Hand sewers 4. Mathematical technicians 5. Insurance underwriters 6. Watch repairers 7. Cargo and freight agents 8. Tax preparers 9. Photographic process workers and processing machine operators 10. New accounts clerks 11. Library technicians 12. Data entry keyers 13. Timing device assemblers and adjusters 14. Insurance claims processing and policy clerks 15. Brokerage clerks 16. Order clerks 17. Loan officers 18. Insurance appraisers, auto damage 19. Umpires, referees and other sports officials 20. Tellers
The little gadget thing lets you stick any job in and see how it rates. Of course it's just a bunch of researchers' guesses but the results are pretty interesting.
I would think choreographers would be more unlikely?
13th is still very unlikely. They ranked something like 500+ individual jobs, so even the difference between the top 20 and 40 is probably a bit arbitrary.
On July 16 2015 11:01 Belisarius wrote: Yeah it looks quite rigorous. At the end of the day it's still an estimation like any other, but calling it a guess is probably not doing the work justice.
It's gonna be a lot better than my random anecdotal guesses would be, anyway.
I think you're incorrect about programming meaning everything else would be automated as well.
Do you program? In my opinion an AI capable of analyzing problems and coding solutions would at that point be capable of automating almost everything else other than strictly human services (prostitution, counseling, etc.)
Any job that isn't maintaining humans or computers (or animals) can likely be done by computers. And we're talking hands on treatment; diagnostics could definitely done by computers.
For example, surgeons are likely going to be pretty hard to replace, but most physicians can easily be replaced by a smart database and some diagnostic hardware.
I wonder how governments are going to handle the masses of jobs lost when low level positions (like tellers), and the transportation industry get replaced by computers. We simply aren't going to be able to follow the "everyone works to earn their keep" model for much longer. My guess is on them restricting the hell out of the technology so human workers can compete.
It's going to be a long time before computers completely take over for computer programmers. It would take a true problem solving computer with some creativity to do that. I don't think it will happen in my lifetime.
However, computer programming jobs are already being lost because programming is so much less time intensive than it used to be. Much of the work is already done and waiting to be connected together by modern programmers. Jobs are also being lost because more and more people are literate in programming and can do small things for themselves without relying on a professional.
I too wonder what will happen if/when governments finally accept that 100% employment is not feasible. In the future, even 75% employment may be very difficult to achieve.
edit: If you don't think surgeons can be replaced, check out Robot-Assisted Surgery. http://www.mayoclinic.org/tests-procedures/robotic-surgery/basics/definition/prc-20013988 It won't completely replace surgeons for the foreseeable future, but it will continue to improve and reduce the time a surgeon spends in surgery with a single patient. That means a single surgeon can service more patients and thus there would be less need for surgeons per capita. It will also slowly help to phase out many of the helpers that are usually involved in the surgery. That will further reduce jobs in higher-end nursing fields.
On July 16 2015 11:01 Belisarius wrote: Yeah it looks quite rigorous. At the end of the day it's still an estimation like any other, but calling it a guess is probably not doing the work justice.
It's gonna be a lot better than my random anecdotal guesses would be, anyway.
I think you're incorrect about programming meaning everything else would be automated as well.
Do you program? In my opinion an AI capable of analyzing problems and coding solutions would at that point be capable of automating almost everything else other than strictly human services (prostitution, counseling, etc.)
Any job that isn't maintaining humans or computers (or animals) can likely be done by computers. And we're talking hands on treatment; diagnostics could definitely done by computers.
For example, surgeons are likely going to be pretty hard to replace, but most physicians can easily be replaced by a smart database and some diagnostic hardware.
I wonder how governments are going to handle the masses of jobs lost when low level positions (like tellers), and the transportation industry get replaced by computers. We simply aren't going to be able to follow the "everyone works to earn their keep" model for much longer. My guess is on them restricting the hell out of the technology so human workers can compete.
Medicine, which I work with medical devices on the legal side, is going to be automated very early on. Not because its easy to automate, but because its safer (a huge % of our costs are complications from unsanitary practices), there are a lot of doctors who are not "interactive", and because standardization is the best way to reduce liability costs. The guy who created RFID sponges basically reduced the cost of surgical procedures on the back end by some ridiculous percentage. Surgeon-controlled robots are going to be the new-normal soon, and the only reason that that will remain a 200k+ job instead of being given to talented and trained (but not 4 useless college years + 2 semi-useful med school years +2 years of semi-useful rotations + residency) artisans making 50, 60, 70k is because the AMA has monopoly and state power.
That sounds fair to me. Can the federal government mandate that a minimum wage be instated that provides roughly equivalent living conditions country-wide? In other words: enough to pay for a home, food, medical expenses, education and a minimal level of comfort (washing machine, computer, clothes, that kinda stuff)? Or would such a law be impossible to draft? I for one cannot think of how it can be phrased.
Because if it simply up to the states, it seems likely that nothing is going to happen in some of the states it is needed most (because lets face it: Oregon is not the state that needed a minimum wage most).
Again, this is why I think it makes more sense to expand the EITC rather than deal with skyrocketing cost of living and inflation as costs are passed down to the consumer. A program like this can probably better compensate for regional cost of living expenses etc.
Oregon as a whole might not need a minimum wage, but Portland- one of the most gentrified cities in the US sure as fucking hell does lol. (Just not too high... we don't want everyone moving here )
On July 16 2015 11:01 Belisarius wrote: Yeah it looks quite rigorous. At the end of the day it's still an estimation like any other, but calling it a guess is probably not doing the work justice.
It's gonna be a lot better than my random anecdotal guesses would be, anyway.
I think you're incorrect about programming meaning everything else would be automated as well.
Do you program? In my opinion an AI capable of analyzing problems and coding solutions would at that point be capable of automating almost everything else other than strictly human services (prostitution, counseling, etc.)
Any job that isn't maintaining humans or computers (or animals) can likely be done by computers. And we're talking hands on treatment; diagnostics could definitely done by computers.
For example, surgeons are likely going to be pretty hard to replace, but most physicians can easily be replaced by a smart database and some diagnostic hardware.
I wonder how governments are going to handle the masses of jobs lost when low level positions (like tellers), and the transportation industry get replaced by computers. We simply aren't going to be able to follow the "everyone works to earn their keep" model for much longer. My guess is on them restricting the hell out of the technology so human workers can compete.
Actually, it'd probably be the other way around. Surgeons would probably be easier to replace than physicians (although I highly doubt either will be replaced). Diagnosing/treating is nothing like a math equation; there are so many things to take into account and there are so many different possible answers from the same set of signs and symptoms that a computer just couldn't do it. Too much of a (quality) physician's job relies on judgments based on interactions with the patient as a human being.
I don't see why diagnostics can't work somewhat like Akinator (the 20 questions bot) where it learns from the answers to constantly improve its accuracy. The more human experience it gets the better it is at identifying patterns that lead it to the correct conclusion. Yes, humans don't always immediately tell you the right symptoms to diagnose something but a computer program doesn't distinguish saying "I heard a loud crack and now I can see my shin bone" from "I was hesitant to come in because it's probably nothing". It's not incapable of intuition because it doesn't have a base level of communication which excludes things it would intuit, if that makes sense. Everything is valid for diagnostics and with a decent sample size it ought to be able to draw conclusions mimicking intuition.
If anything it ought to outperform doctors, it can't miss cues and it has every other case it has ever seen (orders of magnitude more than any doctor) to compare against with perfect recall. You'd need human assistance, especially at first while you teach it, but that kind of pattern recognition and database comparison is exactly what computers beat humans at.
On July 16 2015 13:00 KwarK wrote: I don't see why diagnostics can't work somewhat like Akinator (the 20 questions bot) where it learns from the answers to constantly improve its accuracy. The more human experience it gets the better it is at identifying patterns that lead it to the correct conclusion. Yes, humans don't always immediately tell you the right symptoms to diagnose something but a computer program doesn't distinguish saying "I heard a loud crack and now I can see my shin bone" from "I was hesitant to come in because it's probably nothing". It's not incapable of intuition because it doesn't have a base level of communication which excludes things it would intuit, if that makes sense. Everything is valid for diagnostics and with a decent sample size it ought to be able to draw conclusions mimicking intuition.
If anything it ought to outperform doctors, it can't miss cues and it has every other case it has ever seen (orders of magnitude more than any doctor) to compare against with perfect recall. You'd need human assistance, especially at first while you teach it, but that kind of pattern recognition and database comparison is exactly what computers beat humans at.
Many diagnostics programs already outperform physicians when just given the nurse's "report". These sorts of things are improving all the time, although they are currently mostly experimental and use more constrained conditions than real life often has (basically, patients lie a lot, and/or don't really know what their symptoms are). The medical industry (if government regulation doesn't prevent this) will likely migrate towards automation + 1-2 middle class employees (nurses/NPs) + lower-middle class employees replacing what is currently done by the same people + a doctor.
A lot of providers use CDSS (clinical decision support systems) which help with a whole range of medical things.
I think a lot of primary care doctors are at risk of becoming obsolete. Specialists are probably a little safer, though they will likely need to adapt to using technology in all sorts of ways when they practice.
Computers are used in diagnosis and correlation work in healthcare already. I think IBM Watson is the strongest example of this. A few videos regarding this. Didn't find the one I watched a few months ago where it showed it being used in a hospital.
It's a question of opinion I've asked before, but as its apropos again, I'll pose it: How many hours of minimum wage work should it take to support a family of 4? How many hours for a single person to support themselves?
I'd say at most 60 hours to support a family. And by support i don't mean four people living in a single small room eating only rice every day as if it were 1903.
This already means that two persons need to work, or one needs to work so much he/she can barely do anything else.
A single person would thus end up at around 20-30 hours to support themselves, which sounds fair.
On July 16 2015 10:38 Belisarius wrote: Just on that, here's a fun little toy that predicts how likely a given job is to be automated. The full document/paper is here.
1. Telemarketers 2. Title examiners, abstracters, and searchers 3. Hand sewers 4. Mathematical technicians 5. Insurance underwriters 6. Watch repairers 7. Cargo and freight agents 8. Tax preparers 9. Photographic process workers and processing machine operators 10. New accounts clerks 11. Library technicians 12. Data entry keyers 13. Timing device assemblers and adjusters 14. Insurance claims processing and policy clerks 15. Brokerage clerks 16. Order clerks 17. Loan officers 18. Insurance appraisers, auto damage 19. Umpires, referees and other sports officials 20. Tellers
The little gadget thing lets you stick any job in and see how it rates. Of course it's just a bunch of researchers' guesses but the results are pretty interesting.
I was about to go off the deep end on how there's literally no way that teachers are ever going to be replaced (especially in primary and secondary grades) because, while you could hypothetically watch instructional videos and possibly create an AI that can respond to student-specific questions and misunderstandings, any teacher will tell you the hardest part about teaching- by far- is the classroom management aspect. Whether it's getting kids to put away their phones, waking them up, stopping a fight, getting them quiet (or getting them to talk and collaborate), or (if they're younger) even dealing with crying and screaming and food/ bathroom issues. You can't just use a screen or program to enforce many of these (in a way that has a positive outcome).
But then I looked up the numbers and high school teachers have a 0.8% chance of being automated, elementary school teachers have a 0.4% chance of being automated, and pre-school teachers have a 0.7% chance, and those are certainly fine numbers to describe the massive improbability of them being replaced. Turns out there are plenty of <1% chance of replacement jobs lol, so that's fine.
But then it says that middle school teachers have a whopping 17.4% chance of being replaced. What the hell? That's incredibly inconsistent, and the classroom management for 7th and 8th graders has plenty of issues as well.
The main point people overlook when the argue certain jobs can't be automated (physicians etc) is that it is not necessary for the job to be COMPLETELY automated for it to change fundamentally. If the job of a physician in the future is interviewing a patient, providing human contact, entering data into the PHYSICIAN 9000 network, and interpreting the results, they will be payed closer to what a nurse makes than what a doctor today makes.
There might be plenty of non-automatable jobs left after the machines take over, but most of them will pay shitty.
EPT
)
