Author Topic: Robots and their impact on the future  (Read 203507 times)

Watchmaker

  • Stubble
  • **
  • Posts: 153
Re: Robots and their impact on the future
« Reply #1450 on: June 21, 2017, 12:40:04 PM »
I don't think banning having children would work, for the reasons you mention.  I'm talking about economically disincentivizing having children.

And how, exactly, can you do that without being preferential towards the rich and against the poor?

Progressive taxation?  Doesn't seem any more complicated than a lot of our current taxes.

I don't actually think we're that far apart on this--I agree that improved quality of life is the best birth control around.  You are confident that's all we'll need to keep the population in check.  I hope you're right.   

dougules

  • Pencil Stache
  • ****
  • Posts: 814
  • Location: AL
Re: Robots and their impact on the future
« Reply #1451 on: June 22, 2017, 10:24:15 AM »
I don't think banning having children would work, for the reasons you mention.  I'm talking about economically disincentivizing having children.

And how, exactly, can you do that without being preferential towards the rich and against the poor?

Progressive taxation?  Doesn't seem any more complicated than a lot of our current taxes.

I don't actually think we're that far apart on this--I agree that improved quality of life is the best birth control around.  You are confident that's all we'll need to keep the population in check.  I hope you're right.   

Improved quality of life and education definitely are bringing down birth rates.  If you look at a map of gdp per capita and a map of fertility rates side by side, it's easy to see.  It's not unlikely low birth rates may become the problem in a few decades. 

arebelspy

  • Administrator
  • Senior Mustachian
  • *****
  • Posts: 25245
  • Age: -999
  • Location: Traveling the World
Re: Robots and their impact on the future
« Reply #1452 on: June 22, 2017, 10:31:57 AM »
Improved quality of life and education definitely are bringing down birth rates.  If you look at a map of gdp per capita and a map of fertility rates side by side, it's easy to see.  It's not unlikely low birth rates may become the problem in a few decades.

Yep.

And the thing is, we should do this anyways, to make people's lives better. That it seems like it could help the population problem is a bonus.  But we should be ensuring people are healthy, have food, get access to education, etc. just as a matter of human decency.

So let's try this solution first, and see how population trends due to that, if it matches what has happened over the last century in developed nations.
We are two former teachers who accumulated a bunch of real estate, retired at 29, and now travel the world full time with a kid.
If you want to know more about me, or how we did that, or see lots of pictures, this Business Insider profile tells our story pretty well.
We (occasionally) blog at AdventuringAlong.com.
You can also read my forum "Journal."

Watchmaker

  • Stubble
  • **
  • Posts: 153
Re: Robots and their impact on the future
« Reply #1453 on: June 22, 2017, 12:11:11 PM »
To get back on topic, here is a personal anecdote from my work life:

I work for a manufacturing company with roughly 5,000 employees.  I was recently in Asia visiting a company that makes the same things that we make and had almost exactly the same total sales.  They have 15,000 employees.  The difference is pretty much entirely due to automation at our facilities.

Of course, the Asian company was not without automation.  If they were doing everything manually (as was done in the industry decades ago) they would probably need 5-10 times as many employees.

We're not done automating.  Over the next 10 years, I expect us to double our productivity.  And maybe double it again the 10 years after that.   

Bakari

  • Handlebar Stache
  • *****
  • Posts: 1734
  • Age: 37
  • Location: Oakland, CA
  • Veggie Powered Handyman
    • The Flamboyant Introvert
Re: Robots and their impact on the future
« Reply #1454 on: June 23, 2017, 08:03:45 AM »
oh no, back on topic, just when I noticed the subthread had continued!


That isn't economically disincentivizing kids, it's improving people's lives. They may choose to have less kids after that, but that is not what is being suggested here. That's what I suggested as a better solution than what was suggested by Bakari and Watchmaker.


I'm not sure what you think I was suggesting, but what I wrote was "The transition wouldn't have to be tragically terrible, if we could spread it out a few generations and accomplish it via free sterilizations and birth control and a one-child policy (at least in terms of social expectations, if not law).  We could also, for example, tax each child, instead of providing tax breaks."
Free (not mandatory) birth control, and a social expectation of fewer children, plus not giving tax breaks for having kids.  Taxes incentives tend not to have big impacts on the poor, since they generally pay little or no taxes anyway. 


Although, incidentally, as politically incorrect as it may be to say, I don't think it is actually that terrible to disincentives someone who can't afford to provide for a child to not have that child.  In fact, that may easily make the difference between that person raising out of poverty: average spending per child in the 0-40k income range, (average of only 18k) is 150k over 18 years.  That's almost 1/2 of everything they make. 


More than 1/2 of pregnancies are unintended.  These are dramatically higher among poor women - 5 times as high http://www.nejm.org/doi/full/10.1056/NEJMsa1506575


While people tend to default to genocidal scenarios, the reality is closer to that poverty greatly exacerbates the problem of unwanted pregnancies, which in turn keeps people in poverty.
https://www.theatlantic.com/business/archive/2015/03/unplanned-births-another-outcome-of-economic-inequality/386743/


Free birth control, sterilization, would already go a long way to ensuring each child born was wanted.
And in a very real, direct, and literal way, it IS improving people's lives.


Eliminating tax credits only affect the middle class in any meaningful way anyway - because poor people don't pay taxes anyway, and the rich can easily afford them.  But the highest portion of the population is middle class, so that could still be significant.  There is no particular reason adding to the population should be subsidized.
Any tax penalty could kick in beyond replacement level (2, which is what a majority of people who want kids say they want anyway), at say 4, where it only affects those people who have entire broods.  If everyone else has 2 kids each, then in a few generations Mormons will completely overrun the Earth.  They can at least pay a little more for the social services they'll use as it happens.
« Last Edit: July 03, 2017, 12:35:41 PM by Bakari »
Anything I've said here useful or interesting?  Find a lot more of my thoughts here: http://randomthoughts.fyi

tomsang

  • Pencil Stache
  • ****
  • Posts: 965
Re: Robots and their impact on the future
« Reply #1455 on: June 30, 2017, 02:41:37 PM »
https://www.usatoday.com/story/money/2017/06/29/ai-stealing-human-jobs-isnt-problem-is/412217001/

Good article/video about AI and jobs.  The potential disconnect that I see in this and many articles is that they are using historical expectations on projecting technological advancements. They are conveying that people will adapt.  What I am seeing is that AI/Robots are advancing exponentially every few years.  If that growth continues over the next decade, then what we see in a decade will make today's advanced technology look like a child's toy, twenty years in the future is hard to envision what technology will be capable of doing. 

Those with advanced degrees should be fine for a decade or so, but 70% of the US population does not have a college degree and frankly many don't have the ability to thrive in an academic setting.  I believe that most of the jobs that don't require a job will be replaced by automation, AI, robots, etc.  I would be shocked if stores have cashiers, restaurants have servers, cashiers, cooks, shipping and freight companies use drivers, construction companies use employees, manufacturing companies of all type are human labor dependent in 20 years. I keep seeing tons of lawyers graduating with $250k in student loans desperate to find a job that has experienced significant automation. Gone are the days when you had 40 lawyers in a war room searching through case law.  An AI computer can do what the 40 people took weeks or more in a matter of minutes.  The tools that are being developed for architects, accountants, engineers, lawyers, doctors, writers, and virtually every profession are increasing the productivity to the amount where the work can be done with 1/10 of the staff compared to 20 years ago.  The advancements are increasing on a daily basis and show signs of that the advancements are speeding up.     

More articles are touching on Basic Income, which I think is going to be heavily discussed in the next decade as labor is permanently displaced in our society and those that own the companies keep all the gains from productivity increases.  I truly believe that the future is incredibly bright in total.  I am a bit shocked by how the GOP has convinced a good portion of society to vote against their best interest.  Stripping away entitlements to the population while reducing the taxes on the top 1% seems like a hard sell.  Yet, they are doing it very effectively.  As someone in the top 1%, I appreciate the reduction in taxes, but I think it is crazy that people think that trickle down economics is the right answer.

maizeman

  • Handlebar Stache
  • *****
  • Posts: 1158
  • Location: The World of Tomorrow
Re: Robots and their impact on the future
« Reply #1456 on: July 01, 2017, 08:39:54 AM »
Editorial from a few days ago in the NYtimes brought up an issue I hadn't thought of before with the idea of taxing the companies whose profits will shoot up with the adoption of AI/automation to may for a UBI or other services for the displaced and unemployed:

Quote
A.I. is an industry in which strength begets strength: The more data you have, the better your product; the better your product, the more data you can collect; the more data you can collect, the more talent you can attract; the more talent you can attract, the better your product. It’s a virtuous circle, and the United States and China have already amassed the talent, market share and data to set it in motion.

...

This leads to the final and perhaps most consequential challenge of A.I. The Keynesian approach I have sketched out [tax the companies to provide some level of support to the newly unemployable] may be feasible in the United States and China, which will have enough successful A.I. businesses to fund welfare initiatives via taxes. But what about other countries?

"It’s a selective retirement," Richard explained, "a retirement from boring s**t."

My source code & my journal

tomsang

  • Pencil Stache
  • ****
  • Posts: 965
Re: Robots and their impact on the future
« Reply #1457 on: July 01, 2017, 12:04:59 PM »
This leads to the final and perhaps most consequential challenge of A.I. The Keynesian approach I have sketched out [tax the companies to provide some level of support to the newly unemployable] may be feasible in the United States and China, which will have enough successful A.I. businesses to fund welfare initiatives via taxes. But what about other countries?

The GOP tax plans that are being pushed through the process are slated to lower the corporate tax rate from 35% to 15%.  It is realistically expected that it will be set at 28% vs. 15%, but the concept is at a time when corporations are making money at historic levels and income inequality is growing significantly, that there would be a push to lower tax rates on corporations when the US government has an annual deficit is crazy.  In order to fund Universal Basic Income, healthcare and other programs to share the technological wealth with the population you will need corporate taxes to increase.  The fact that the GOP is pushing for significant decreases, and their poor and middle class constituents are supporting this is baffling.

tomsang

  • Pencil Stache
  • ****
  • Posts: 965
Re: Robots and their impact on the future
« Reply #1458 on: July 04, 2017, 11:11:49 AM »
https://www.yahoo.com/finance/news/programmers-having-big-debate-over-182231099.html

Interesting article.  Guy automates his 40 hour job into a 2 hour job, does not tell his employer.  Is it ethical?  Does it give a glimpse of what the future holds even for those educated office jobs?

Kriegsspiel

  • Handlebar Stache
  • *****
  • Posts: 1036
Re: Robots and their impact on the future
« Reply #1459 on: July 04, 2017, 12:15:51 PM »
First they came for our office jobs,
easily automated.
v2.3 came for our souls,
and wrote their roboty songs.
Then they came for our bodies, with their nano-factories, to turn us into paper clips.
To hold together the sheet music.

mozar

  • Handlebar Stache
  • *****
  • Posts: 2352
Embracing the absurd condition of human existence while also defiantly continuing to explore and search for meaning

pdxmonkey

  • Stubble
  • **
  • Posts: 225
Re: Robots and their impact on the future
« Reply #1461 on: July 05, 2017, 01:10:23 AM »
This leads to the final and perhaps most consequential challenge of A.I. The Keynesian approach I have sketched out [tax the companies to provide some level of support to the newly unemployable] may be feasible in the United States and China, which will have enough successful A.I. businesses to fund welfare initiatives via taxes. But what about other countries?

The GOP tax plans that are being pushed through the process are slated to lower the corporate tax rate from 35% to 15%.  It is realistically expected that it will be set at 28% vs. 15%, but the concept is at a time when corporations are making money at historic levels and income inequality is growing significantly, that there would be a push to lower tax rates on corporations when the US government has an annual deficit is crazy.  In order to fund Universal Basic Income, healthcare and other programs to share the technological wealth with the population you will need corporate taxes to increase.  The fact that the GOP is pushing for significant decreases, and their poor and middle class constituents are supporting this is baffling.
If you cut them and cute all the historic fat it gives you room to raise them and add more appropriate modern day like ubi

prognastat

  • Pencil Stache
  • ****
  • Posts: 797
  • Location: Texas
Re: Robots and their impact on the future
« Reply #1462 on: July 05, 2017, 08:41:26 AM »
https://www.yahoo.com/finance/news/programmers-having-big-debate-over-182231099.html

Interesting article.  Guy automates his 40 hour job into a 2 hour job, does not tell his employer.  Is it ethical?  Does it give a glimpse of what the future holds even for those educated office jobs?

I don't think it's unethical as long as he is getting the work agreed upon done to the employer's satisfaction. The only thing I feel is slightly unethical is the introducing of bugs in to his work to make it look more human and actually doing the job less well than it could be done.

sol

  • Walrus Stache
  • *******
  • Posts: 5286
  • Age: 40
  • Location: Pacific Northwest
Re: Robots and their impact on the future
« Reply #1463 on: July 05, 2017, 09:00:16 AM »
https://www.yahoo.com/finance/news/programmers-having-big-debate-over-182231099.html

Interesting article.  Guy automates his 40 hour job into a 2 hour job, does not tell his employer.  Is it ethical?  Does it give a glimpse of what the future holds even for those educated office jobs?

I don't think it's unethical as long as he is getting the work agreed upon done to the employer's satisfaction. The only thing I feel is slightly unethical is the introducing of bugs in to his work to make it look more human and actually doing the job less well than it could be done.

The discussion about that article basically breaks people down into one of two groups.  You either believe that an employment contract obligates the employer to pay a specified amount if the employee meets minimum performance benchmarks, or that it obligates to the employee to perform to a minimum standard in exchange for a specified reward.  Or both, I guess, but I tend to think it is neither.  Employment is almost at always "at-will" on both sides of the contract.  Both sides agree to it, temporarily, because it seems advantageous to their side.

In a bygone age, people joined a firm the way people join the military, with a personal commitment to support the mission and not just an agreement to perform specified services, and in exchange the firm offered them unconditional support (food and housing and future employment guarantees).  In that sort of situation, I agree the employee is at fault for failing to live up to that expectation, by defrauding the employer.  But in the modern world, I think it's fine to meet your minimum obligation.  They don't care about you, so why should you care about them?

There are still some old-timers out there who think that modern employment should be more like military service.  I expect that those are the people who will be offended by the notion of a guy who automates away his own job.

tomsang

  • Pencil Stache
  • ****
  • Posts: 965
Re: Robots and their impact on the future
« Reply #1464 on: July 05, 2017, 10:08:05 AM »
The discussion about that article basically breaks people down into one of two groups.  You either believe that an employment contract obligates the employer to pay a specified amount if the employee meets minimum performance benchmarks, or that it obligates to the employee to perform to a minimum standard in exchange for a specified reward. Or both, I guess, but I tend to think it is neither.  Employment is almost at always "at-will" on both sides of the contract.  Both sides agree to it, temporarily, because it seems advantageous to their side.

I think you are hitting on it, with the addition that the minimum standard is usually defined as working 40+ hours vs. accomplishing a specific list of tasks for salaried employees. Most people "work" significantly less than 40 hours but are at work for more than 40 hours.  That is deemed to be acceptable as they are at their employers place of business vs. sitting at home playing video games or posting on MMM.

Schaefer Light

  • Pencil Stache
  • ****
  • Posts: 765
Re: Robots and their impact on the future
« Reply #1465 on: July 05, 2017, 10:31:55 AM »
https://www.yahoo.com/finance/news/programmers-having-big-debate-over-182231099.html

Interesting article.  Guy automates his 40 hour job into a 2 hour job, does not tell his employer.  Is it ethical?  Does it give a glimpse of what the future holds even for those educated office jobs?

To me, the whole debate boils down to whether to pay people by the hour or to pay them for the results they produce.

Bakari

  • Handlebar Stache
  • *****
  • Posts: 1734
  • Age: 37
  • Location: Oakland, CA
  • Veggie Powered Handyman
    • The Flamboyant Introvert
Re: Robots and their impact on the future
« Reply #1466 on: July 05, 2017, 11:18:00 AM »
I doubt any employer would find it preferable to have non-productive employees sitting around for its own sake compared to purely performance based pay, its just that not all jobs have an easily identifiable task list or set of completeable objectives, or any way to measure productivity objectively, or it may just be impossible to guess exactly when service may be needed.


A security guard, a cashier, a hotel desk person, a firefighter, on a slow day they might only be doing productive work for an hour in an 8 hour shift, but they need to be on call at all times, because there is no way to know exactly when the service may be needed.


Other jobs it may just be more complex to try to figure out the fair or appropriate value to give to each unit of productivity - in a factory that makes metal parts of dozens of shapes and sizes and materials, how do you weight the commission price for a small square part with a hole on the side compared to a large round part with a hole in the middle and a twisty thing on the edge, such that the person assigned to each is both fairly and proportionately compensated for their time?
Anything I've said here useful or interesting?  Find a lot more of my thoughts here: http://randomthoughts.fyi

sol

  • Walrus Stache
  • *******
  • Posts: 5286
  • Age: 40
  • Location: Pacific Northwest
Re: Robots and their impact on the future
« Reply #1467 on: July 05, 2017, 11:33:17 AM »
Other jobs it may just be more complex to try to figure out the fair or appropriate value to give to each unit of productivity

Fairly valuing productivity is one of the hardest parts of labour economies, and it's not just manufacturing jobs.  I work in an office where some people who work twice as hard as I do make half as much money, and some people who hardly work at all make twice as much as I do.  What determines the fair market rate for wiggling your fingers on a keyboard?

In my case, it mostly has to do with the status of the people you wiggle them at, and the educational background required to be allowed to do that wiggling, but it's certainly not in any way correlated with the amount of effort you put in.  And yet we are all paid by the hour, as if we were ditch diggers who could measure our outputs in linear feet of ditches dug, correlated with hours spent digging.

This warped system has created perverse incentives for people to move up the food chain regardless of their aptitudes or abilities.  If you are bad at your job, but want to make more money, you just need to figure out how to wiggle those fingers at a higher level of people within the organization.  People with talent or diligence don't get promoted, because we need them to do the (low paid) work.  People who suck at the actual work rise up through the ranks until they get paid gobs of money to do nothing except oversee and report on the work other people have done.

Optimiser

  • Bristles
  • ***
  • Posts: 277
  • Age: 34
  • Location: PNW

tomsang

  • Pencil Stache
  • ****
  • Posts: 965
Re: Robots and their impact on the future
« Reply #1469 on: July 19, 2017, 02:20:19 PM »
http://www.cnbc.com/2017/07/17/elon-musk-robots-will-be-able-to-do-everything-better-than-us.html

Interesting story.  It will be interesting if rules, processes, etc. are put in place before AI's impact is fully in effect. It seems like it would be beneficial to be in front of this vs. wrestling control, taxes, out of the hands of those owning the tech or when/if we get to the point where nothing can control AI. 

tomsang

  • Pencil Stache
  • ****
  • Posts: 965
Re: Robots and their impact on the future
« Reply #1470 on: July 19, 2017, 07:31:55 PM »
http://www.weeklystandard.com/cursed-be-the-machines-for-they-shall-inherit-the-earth/article/2008899

But the principle that everyone is replaceable does bring some comfort. For just the other day, I came across a heartwarming piece in Computerworld. It explored the research of Evans Data Corp, which surveyed 550 software developers, asking them about the most worrisome aspect of their careers. Ranking second and third, respectively, was that the platform they were working on would become obsolete (23 percent) or wouldn’t catch on (14 percent). But what kept a plurality of them up at night (29 percent) is that they and their development efforts would be replaced by artificial intelligence. In other words, the people who are designing the machines to replace us live in abject fear of being replaced by the machines.

bender

  • Pencil Stache
  • ****
  • Posts: 844
Re: Robots and their impact on the future
« Reply #1471 on: July 19, 2017, 08:18:38 PM »
That's funny!

It's a valid concern in software development if your job is the software equivalent of burger flipper.  We're not going to see software algorithms develop themselves anytime soon, but there is a lot of automation going on in software that improves productivity.  This allows companies to do more with fewer developers.  A lot of effort goes into test/QA for instance, and this area is the focus of many productivity enhancing tools.

brooklynmoney

  • Bristles
  • ***
  • Posts: 309
Re: Robots and their impact on the future
« Reply #1472 on: July 19, 2017, 08:24:45 PM »
I love machine learning and AI and robots but we gotta long way to go yet: https://www.google.com/amp/amp.usatoday.com/story/491227001/

tomsang

  • Pencil Stache
  • ****
  • Posts: 965
Re: Robots and their impact on the future
« Reply #1473 on: July 19, 2017, 09:18:23 PM »
I love machine learning and AI and robots but we gotta long way to go yet: https://www.google.com/amp/amp.usatoday.com/story/491227001/

Funny stuff! 

jordanread

  • Walrus Stache
  • *******
  • Posts: 5875
  • Age: 32
  • Location: Colorado Springs
  • Live Long, Live Free, Drop Dead
    • Frugal FIRE Show
Re: Robots and their impact on the future
« Reply #1474 on: July 31, 2017, 04:04:56 PM »
Holy crap. This just showed up in my inbox.

https://www.glassdoor.com/blog/9-jobs-that-the-robots-cannot-take/

Personally, I think they are way off base, but that's just me.
Join the cycling challenge!
Get in shape in 2017!
Frugal FIRE - Episode 2

"Mustachians rarely sit back and let things happen to them. Mustachians go out and happen to things."

Camp Mustache Canada - Sold Out Already!!

lost_in_the_endless_aisle

  • Bristles
  • ***
  • Posts: 475
Re: Robots and their impact on the future
« Reply #1475 on: July 31, 2017, 05:44:27 PM »
Holy crap. This just showed up in my inbox.

https://www.glassdoor.com/blog/9-jobs-that-the-robots-cannot-take/

Personally, I think they are way off base, but that's just me.
Never is a long time.

I think Doug Hofstadter wrote a while back (in his SciAm contribution compilation Metamagical Themas maybe?) that he didn't think it would be possible to automate language translation for a variety of good reasons. The last several years have seen massive gains in machine translation, and it's now expected current methods will displace nearly all human translation work with the next 5-10 years. Having said that, the deep learning the translation methods are based off of is not anything like general artificial intelligence that could do the sort of jobs listed in the article (yet). Additional foundational theoretical breakthroughs are likely needed before we have AGI. I think it can and will happen eventually and that 100% of jobs are at risk in the long run (20-100 years[?] based on surveys of experts in AI).
« Last Edit: July 31, 2017, 05:50:58 PM by lost_in_the_endless_aisle »

sol

  • Walrus Stache
  • *******
  • Posts: 5286
  • Age: 40
  • Location: Pacific Northwest
Re: Robots and their impact on the future
« Reply #1476 on: July 31, 2017, 06:40:59 PM »
Having said that, the deep learning the translation methods are based off of is not anything like general artificial intelligence that could do the sort of jobs listed in the article (yet).

Isn't that just a skills problem, though?  I have good general intelligence, but I'm mostly useless as a language translator.  It's a skill my intelligence hasn't learned.

I think the advances here are in the methodologies deployed to learn each new skill.  A generally intelligent AI won't have to know everything, it will have to know a great many things and (more importantly) it will have to know how to train itself to learn new things.

And the scary (or great?) part of AI is that computer data transmission makes transfering that skill to other AIs basically instantaneous.  Right now, language translation or car driving or poetry writing are skills that some people have and other don't, but with AIs as soon as one AI learns it all connected AIs have instant access to that skill.  That's where the huge exponential growth in capabilities comes that so rapidly and threateningly outstrips human capabilities.  Even an AI that isn't as smart as a dumb person can potentially have more skills than every smart person put together.

lost_in_the_endless_aisle

  • Bristles
  • ***
  • Posts: 475
Re: Robots and their impact on the future
« Reply #1477 on: July 31, 2017, 07:01:12 PM »
Having said that, the deep learning the translation methods are based off of is not anything like general artificial intelligence that could do the sort of jobs listed in the article (yet).

Isn't that just a skills problem, though?  I have good general intelligence, but I'm mostly useless as a language translator.  It's a skill my intelligence hasn't learned.

I think the advances here are in the methodologies deployed to learn each new skill.  A generally intelligent AI won't have to know everything, it will have to know a great many things and (more importantly) it will have to know how to train itself to learn new things.

And the scary (or great?) part of AI is that computer data transmission makes transfering that skill to other AIs basically instantaneous.  Right now, language translation or car driving or poetry writing are skills that some people have and other don't, but with AIs as soon as one AI learns it all connected AIs have instant access to that skill.  That's where the huge exponential growth in capabilities comes that so rapidly and threateningly outstrips human capabilities.  Even an AI that isn't as smart as a dumb person can potentially have more skills than every smart person put together.
That is a good point and I should probably read up more on Fodor's Modularity of Mind concept as background on this. Are generally smart systems merely a connection of dumb narrowly smart domains of intelligence or does general AI require a deeper & qualitatively distinct connection between the disparate elements of narrow, domain-specific knowledge?

There was some chatter a while ago around autism spectrum individuals and how so-called "weak central coherence" can account for some observations of the abilities of individuals on the spectrum: in that theory, autism is (in part) a problem of global information synthesis that results (in cases of high-functioning autism) in very high domain-specific abilities, though such individuals cannot synthesize information as effectively across domains. If the highest levels of creativity depend on interconnections between disparate domains then those links would need to be accounted for and implemented in a system that successfully exhibits intelligence at that level (which makes the "wire a bunch of domain specific AIs together" idea seem much more questionable if we have no way of accounting for what it takes to achieve central coherence).

Bakari

  • Handlebar Stache
  • *****
  • Posts: 1734
  • Age: 37
  • Location: Oakland, CA
  • Veggie Powered Handyman
    • The Flamboyant Introvert
Re: Robots and their impact on the future
« Reply #1478 on: August 01, 2017, 08:44:17 AM »
I can't think of a circumstance where any one intelligent AI that can replace jobs would need access to every possible (human) skill.  The universal translator doesn't need, or even have any use, for driving a truck in traffic or determining probability of guilt given criminal evidence.  It doesn't need a robot body and the algorithms to walk around over rough ground.  An author bot may need to synthesize many different areas of history, philosophy, and elements of story telling, but it needs no physical skills.  A construction bot needs only physical skills. 



So as far as employment becoming obsolete, a "general" intelligence wouldn't need to be as general as actual human beings are - which means the step of central coherence may not even be necessary.


Incidentally, a lot of research does seem to indicate that we have a bunch of more or less independent mind "modules", and that our unified sense of self is mostly an illusion (which is why brain damage tends to cause the loss of specific skills or abilities or memory, while often not affecting anything else)

Anything I've said here useful or interesting?  Find a lot more of my thoughts here: http://randomthoughts.fyi

prognastat

  • Pencil Stache
  • ****
  • Posts: 797
  • Location: Texas
Re: Robots and their impact on the future
« Reply #1479 on: August 01, 2017, 09:01:58 AM »
Holy crap. This just showed up in my inbox.

https://www.glassdoor.com/blog/9-jobs-that-the-robots-cannot-take/

Personally, I think they are way off base, but that's just me.

It cracks me up how clueless the writer is about actual progress in robots and AI. From their examples it comes across as them thinking of robots as the stuff in a factory and not really any advanced AI. They also seem to not realize that robots/AI wouldn't have to take over every single task to make most jobs in the fields non existent. It also seems that a lot of them might not automate as soon as for example driving, but eventually they likely will be anyway.

Teacher: I don't see how a virtual classroom couldn't be run by a sufficiently advanced AI. Better yet we would have an unlimited supply of teachers allowing every student to have a 1 on 1 teacher that tailors the lessons directly to them instead of a class as a whole.

Nurse: Not sure why an AI with advanced sensors would be incapable of performing many of the nursing tasks.

Surgeon: They are already working on developing surgery robots. No reason this couldn't be done.

CEO: Imagine an AI CEO with perfect knowledge of a company's data and performance analyzing it and making choices based on these in depth analyses.

Police Officer: Much of the paperwork related side of the police department could be automated reducing the labour force significantly and if robotics advances enough having actual robotic patrols doesn't seem impossible to imagine.

Lawyer: A lot of lawyer work is already being automated. Currently mostly on the discovery portion which makes up a large portion of lawyer's work hours significantly decreasing the amount of lawyers needed to work a case.

Hair Stylist: Not sure why they think this couldn't be done by a robot. I don't even think it would have to be a significantly smart robot in the AI side of things. I could see it making a 3D scan of your head and then using that both to show you possible hairstyles and then once you choose using that same 3D scanning capability to determine how to cut your hair.

Maintenance: I don't see why they believe this can't be automated ever.

Physical Therapy: Pretty much the same reasoning as the Nurse.

Maybe the writer is actually a robot/AI and trying to trick us in to feeling safe:
https://www.wired.com/2017/02/robots-wrote-this-story/

robartsd

  • Pencil Stache
  • ****
  • Posts: 917
Re: Robots and their impact on the future
« Reply #1480 on: August 01, 2017, 09:08:28 AM »
Having said that, the deep learning the translation methods are based off of is not anything like general artificial intelligence that could do the sort of jobs listed in the article (yet).

Isn't that just a skills problem, though?  I have good general intelligence, but I'm mostly useless as a language translator.  It's a skill my intelligence hasn't learned.
They're getting pretty good at the skill of deep learning for language translation, that skill does not nessisarily translate to deep learning in vastly different domains. I am quite skilled at learning technical things, but not very skilled at learning human languages. I have specific knowledge in some technical things, but find it relativily easy to learn other technical things as needed.

I think the advances here are in the methodologies deployed to learn each new skill.  A generally intelligent AI won't have to know everything, it will have to know a great many things and (more importantly) it will have to know how to train itself to learn new things.

And the scary (or great?) part of AI is that computer data transmission makes transfering that skill to other AIs basically instantaneous.  Right now, language translation or car driving or poetry writing are skills that some people have and other don't, but with AIs as soon as one AI learns it all connected AIs have instant access to that skill.  That's where the huge exponential growth in capabilities comes that so rapidly and threateningly outstrips human capabilities.  Even an AI that isn't as smart as a dumb person can potentially have more skills than every smart person put together.
Before I started driving cars, I had been a passenger in many different models. Even as a beginner driver, I gained experience in a variety of vehicles (the driver's training school car, my parent's sedan, my parent's van) before becoming a licensed driver. I have since driven many more cars. I transfer the skill of driving a car to each new arrangement of displays, controls, and vehicle size/shape. I have yet to drive a driver-on-right car, so I'm sure that would take some adjusting, but I'm also sure I could do it. As far as I know, current AIs are learning to drive particular cars (or at least particular sensor sets for self-driving cars). To the AI, a different sensor arrangement may be more difficult to adjust to than operating a driver-on-right car would be for me.

We certainly are able to make AIs that can master specific domains. How close to a general Artificial Intelligence are we getting?

Bakari

  • Handlebar Stache
  • *****
  • Posts: 1734
  • Age: 37
  • Location: Oakland, CA
  • Veggie Powered Handyman
    • The Flamboyant Introvert
Re: Robots and their impact on the future
« Reply #1481 on: August 01, 2017, 04:50:43 PM »
One of the tasks in last years DARPA robot challenge was driving a car(t) designed for human control.  I.e. with no sensors or feedback other than what they have built in. 
Someone posted a link in this thread to a robot that flies a small airplane (a regular one, built for human control).
So there is def. nothing inherent to the software that requires being purpose built to a specific vehicle.


To the extent that it is most often done that way, the reason goes back to my last point: there is simply no particular benefit to building a robot that can drive cars, as opposed to just building a car with AI built in that can drive itself. 


I've read many modern drones are not fully programmed with how to deal with different weather conditions, they are just given a set of goals and they figure it out through real time feedback, getting better at piloting with experience.
I've also seen video of a monkey with electrodes in its head, learning to operate a robot arm with its thoughts (and they same for paraplegic human with a mouse cursor, but monkeys controlling robot arms just seems a lot cooler...)
The two don't seem that different to me, and theres no reason the same adaptive software couldn't apply to figuring out how different sensors are mapped in different cars. 
Heck, my ultraguage can adapt to different sensors in different cars, and it isn't even AI
Anything I've said here useful or interesting?  Find a lot more of my thoughts here: http://randomthoughts.fyi

Bakari

  • Handlebar Stache
  • *****
  • Posts: 1734
  • Age: 37
  • Location: Oakland, CA
  • Veggie Powered Handyman
    • The Flamboyant Introvert
Re: Robots and their impact on the future
« Reply #1482 on: August 01, 2017, 05:05:20 PM »
It cracks me up how clueless the writer is about actual progress in robots and AI. From their examples it comes across as them thinking of robots as the stuff in a factory and not really any advanced AI. They also seem to not realize that robots/AI wouldn't have to take over every single task to make most jobs in the fields non existent.


For the most part I agree with you, but I think the key point for nurses, police, lawyers, even CEOs, is human relation skills, which might reasonably be expected to not be mastered by AI.


...at least until perhaps there is a truly fully general intelligence that synthesizes ALL areas of human knowledge - basically when/if AI becomes fully sentient, possibly with emotions and all.  Of course, if that did happen, they might lose all the benefits of being AI, and have emotional break downs, or its own goals, demand higher pay, or just start behaving as irrationally as humans do!


After all, the most rational and efficient among humans tend to not be great at human relations!


As to maintenance / handyman - well, the DARPA robot challenge showed just the tip of the iceberg in terms of how complex that can be and how far we are from it.  Not saying it will never happen, but the amount of completely different technical skills needed and the variety of implementations of every one of them makes for a much larger skill set necessary than most jobs.  And of the many, varied tasks, there isn't really even individual subroutines that could be efficiently automated with a more specialized bot.
Its not so much the knowledge part, but the dexterity.  Maybe when we have robots good enough to run an American Ninja Warrior course, I'll be worried about my job.  Hope to be retired long before it comes to that!

Anything I've said here useful or interesting?  Find a lot more of my thoughts here: http://randomthoughts.fyi

tomsang

  • Pencil Stache
  • ****
  • Posts: 965
Re: Robots and their impact on the future
« Reply #1483 on: August 01, 2017, 05:18:15 PM »
Maybe when we have robots good enough to run an American Ninja Warrior course, I'll be worried about my job.  Hope to be retired long before it comes to that!

They could probably do that within 24 months if someone footed the bill to build it.  What parts of the course are you believing would be overly challenging to overcome?

Bakari

  • Handlebar Stache
  • *****
  • Posts: 1734
  • Age: 37
  • Location: Oakland, CA
  • Veggie Powered Handyman
    • The Flamboyant Introvert
Re: Robots and their impact on the future
« Reply #1484 on: August 01, 2017, 06:40:20 PM »
The same as with the handyman stuff - its not so much the AI as the robotics.  Every portion is so different than the rest, sure it might be easy to build a robot that can do any one obstacle perfectly, but todays technology could def. not do all of them.

I've been assuming everyone has seen this, but here:
https://www.youtube.com/watch?v=g0TaYhjpOfo


This is the current state of the art.  They had three years to develop, test, and perfect.  Some big budgets.  Built by the world's top universities, government agencies (NASA), and businesses (google). 
Walking on anything other than perfectly smooth pavement is still a challenge for a robot designed to do anything more complex than walking.
Opening a door, walking up stairs, getting out of a cart, walking over uneven ground, the whole course would take a normal human a few minutes.  The best robot took 45. 
The type of skills on ANW have to be performed in real time.

And that's without considering that nobody knows what the course actually is until they day they show up, so you can't even practice anything specific to the course.

Getting a robot to run an obstacle course that 90% of athletes can't finish is a lot more than 24 months away.
« Last Edit: August 01, 2017, 06:52:14 PM by Bakari »
Anything I've said here useful or interesting?  Find a lot more of my thoughts here: http://randomthoughts.fyi

lost_in_the_endless_aisle

  • Bristles
  • ***
  • Posts: 475
Re: Robots and their impact on the future
« Reply #1485 on: August 01, 2017, 07:15:36 PM »
I can't think of a circumstance where any one intelligent AI that can replace jobs would need access to every possible (human) skill.  The universal translator doesn't need, or even have any use, for driving a truck in traffic or determining probability of guilt given criminal evidence.  It doesn't need a robot body and the algorithms to walk around over rough ground.  An author bot may need to synthesize many different areas of history, philosophy, and elements of story telling, but it needs no physical skills.  A construction bot needs only physical skills. 



So as far as employment becoming obsolete, a "general" intelligence wouldn't need to be as general as actual human beings are - which means the step of central coherence may not even be necessary.


Incidentally, a lot of research does seem to indicate that we have a bunch of more or less independent mind "modules", and that our unified sense of self is mostly an illusion (which is why brain damage tends to cause the loss of specific skills or abilities or memory, while often not affecting anything else)
Though I think many specialized jobs are vulnerable to narrow AI without getting too fancy, I'm still skeptical that reducing human-level intelligence to merely the behavior of brain modules and marginalizing the potential significance of central coherence is warranted.

One line of argument I can think of is the importance of meta-cognition in the development of good judgements and (crucially) estimates on the likelihood of something being so given the content of all of the rest of your knowledge. Consider, for instance, crackpot conspiracy theorists like those who believe the earth is flat. Why do most people not believe the earth is flat? How many of us have been in space or replicated Eratosthenes' experiment? The reason to believe in roundness (or in Newton's laws, or that Africa exists) without direct experience or experimentation is to understand how incredibly well the prevailing non-crackpot theories of the world collectively build up a concept of the world that is both hard to falsify and difficult to vary. I've always thought of crossword puzzles as an analogy where only certain facts and explanations fit given a disparate set of other facts and explanations. Intelligence is about weaving those disparate strands of information and belief about the world together into a non-contradictory whole. Cognition at this level seems to require incorporation of as much as we can about every other belief we hold. For those who struggle with this ability, well, there's a group in Fort Collins that I'm sure will be happy to have you!

Maybe meta-cognition is itself just a brain module though (I'm not familiar with arguments to this effect however), in which case deriving its benefits is reducible to a sub-task that is more limited in scope and complexity. I'm not convinced either way but wanted to point out an argument that cast doubt on the swarm-of-narrow-AI "solution" to general AI.

Interestingly, there is a corollary to the argument above as follows: if central coherence is necessary for a true high-functioning general-AI, then that would have the effect of constraining the parameter space of all likely minds. The reason is a strongly centrally coherent AI can't become too lopsided in its skills and beliefs because gaining incremental skill in one particular direction might require a higher baseline of knowledge in a variety of disparate areas. This would tend to lessen the dangers of general AI because it would mean any general AI that was viable would be constrained in any extreme coupling of tendencies and abilities by its overall epistemological horizon. The parameter space of minds is still huge even under this assumption, so it's not entirely reassuring, even if true. This discussion between Sam Harris and David Deutsch is where I got this idea and why I think the two can't reconcile their differing opinions in the course of that discussion (Harris believes any arbitrary AI mind is plausible, while Deutsch believes they will necessarily have much in common with humans because of the unpredictable yet constraining nature of broad knowledge across domains: e.g. a super-AI that is truly good at building weapons will necessarily have some skill at questioning the morality of killing because the requisite intelligence for both has a common ground).

WhiteTrashCash

  • Pencil Stache
  • ****
  • Posts: 553
Re: Robots and their impact on the future
« Reply #1486 on: August 01, 2017, 08:18:44 PM »
I just learned that people are paying $14 per drink to have their drinks mixed by robots at a bar called Tipsy Robot in Las Vegas. They are literally paying extra for this. People are stupid.

applegrapepie

  • 5 O'Clock Shadow
  • *
  • Posts: 11
Re: Robots and their impact on the future
« Reply #1487 on: August 01, 2017, 10:45:34 PM »
Robot is already dominating people I think. And in the future, I think there will be whole separate world like movie I Robot

tomsang

  • Pencil Stache
  • ****
  • Posts: 965
Re: Robots and their impact on the future
« Reply #1488 on: August 01, 2017, 11:43:56 PM »
This is the current state of the art.  They had three years to develop, test, and perfect.  Some big budgets. 

This competition is two years old.  This is old technology with university budgets for the most part.  You throw a billion around and give them a few years, you will see some amazing things can happen.  They could conquer the challenges.  The challenges are not too bad if you know what the course is going to be.  If it is totally random them it will be more challenging.  The shows that we see are based on challenges that are reused with a few that are new.

Darpa's self driving vehicles were a failure.  A few years later, we have self driving cars going down the street.   

Leisured

  • Bristles
  • ***
  • Posts: 391
  • Age: 72
  • Location: South east Australia, in country
  • Retired, and loving it.
Re: Robots and their impact on the future
« Reply #1489 on: August 02, 2017, 06:56:45 AM »
I suspect that many people, faced with widespread automation, will agitate politically to restrict automation, partly through Government intervention.

Late 2003 I was training to be a financial planner at a major bank in Australia. Our instructor introduced us to the software which would help us organising a financial plan for a client. I am familiar with computer programming, and I realised that the software handled the boring part of recording the client’s name, address and phone number, and listed their assets and liabilities, and added assets and liabilities to form totals.

When it came to the less boring part of crafting a financial plans for the client, automation withdrew into the background. It seemed to be that someone in Head Office had said to the programmers, ‘thus far and no further.’

A classmate raised the matter with me. He had done some Fortran programming at university, and I remember the look of dismay on his face when he realised that there had been subtle obstruction to full automation.

Our instructor told the class in general, ‘some of you may think there is scope for further automation, but I can tell you now, do not bother raising the matter, because nothing will happen.’ It seemed that he too knew what was going on.

A few years later, I was working at a large winery, which had recently installed a huge and impressive mechanized wine bottling line. You may have noticed that in cardboard cartons of wine, there are thin cardboard dividers to stop the bottles clinking when you lift up the carton. These cardboard dividers come packed flat, and there is an ingenious machine which expands these dividers and drop them over the bottles in the open carton, as it moves past on the production line. A short patch of the production line vibrates to shake the cardboard divider down between the bottles.

The machine which expands the cardboard dividers went through its motions but failed to fully expand the dividers. The machine relied on rubber suction cups to grip the cardboard, and the cups seemed brittle. Nothing was done, so workers were hired to take the imperfectly expanded dividers and drop them into the wine cartons, as they passed, about one per second.

Nobody said anything. I assumed, perhaps wrongly, that the lack of maintenance of the rubber suction cups was deliberate, to provide employment. There seemed to be a rule: machines which did heavy lifting were allowed to do their job, but lighter work was left to human workers, if the machinery for that work could be disabled.

I have scientific training, and felt dismayed that machinery would be deliberately nobbled to allow work for people. My classmate in the financial planning class was also dismayed.

I suspect that many people would not be dismayed by such obstruction of some aspects of automation, and would see it as common sense to obstruct such automation. 200 years ago, Ned Ludd and his associates, unemployed weavers, broke into mechanized mills and laid about them with sledge hammers. As people say nowadays, the Luddites were not wrong, just 200 years to early. Luddite behaviour now seems to be obstruction rather than destruction.

I suspect that in the far future, most people in the world will live comfortable lives, but there will be two groups of nations; the A group, which welcomes full automation, and whose members will live leisured lives, supported by machines; and the B group, whose members will live in economies like the ones rich countries lived in say, thirty years ago, where there were jobs for nearly everyone.


prognastat

  • Pencil Stache
  • ****
  • Posts: 797
  • Location: Texas
Re: Robots and their impact on the future
« Reply #1490 on: August 02, 2017, 07:29:33 AM »
It cracks me up how clueless the writer is about actual progress in robots and AI. From their examples it comes across as them thinking of robots as the stuff in a factory and not really any advanced AI. They also seem to not realize that robots/AI wouldn't have to take over every single task to make most jobs in the fields non existent.


For the most part I agree with you, but I think the key point for nurses, police, lawyers, even CEOs, is human relation skills, which might reasonably be expected to not be mastered by AI.


...at least until perhaps there is a truly fully general intelligence that synthesizes ALL areas of human knowledge - basically when/if AI becomes fully sentient, possibly with emotions and all.  Of course, if that did happen, they might lose all the benefits of being AI, and have emotional break downs, or its own goals, demand higher pay, or just start behaving as irrationally as humans do!


After all, the most rational and efficient among humans tend to not be great at human relations!

Does it really need full sentience/AI to be able to achieve this sufficiently for most people? I suspect you could convince a large portion of people by simply making it good enough at faking it.

What if it can completely convince people 90% of the time on social skills, perform the actual functional part of the job with 99.9% success rate and at less than 10% of the cost? Would people still rather go see a human?

Also from my perspective for a lot of younger people less human interaction is a plus not a negative. And lastly this is why I pointed out that robots wouldn't have to automate all of your tasks to effectively make most people obsolete. If a lawyer AI can do 90% or more of the actual work hours at a fraction of the cost and time this would mean 9 out of 10 lawyers could be fired and instead the last 1 could simply cover the work that is left. Doesn't mean the whole job is automated, but would mean the vast majority of them would be out of a job.
« Last Edit: August 02, 2017, 07:40:32 AM by prognastat »

Bakari

  • Handlebar Stache
  • *****
  • Posts: 1734
  • Age: 37
  • Location: Oakland, CA
  • Veggie Powered Handyman
    • The Flamboyant Introvert
Re: Robots and their impact on the future
« Reply #1491 on: August 02, 2017, 08:33:03 AM »
Though I think many specialized jobs are vulnerable to narrow AI without getting too fancy, I'm still skeptical that reducing human-level intelligence to merely the behavior of brain modules and marginalizing the potential significance of central coherence is warranted.


I agree entirely.  My point is relatively few jobs actually require human-level intelligence.  Really, I doubt there is any one job that uses every single skill that a normal human has.

Quote
One line of argument I can think of is the importance of meta-cognition in the development of good judgements and (crucially) estimates on the likelihood of something being so given the content of all of the rest of your knowledge.
Perhaps so, but we are actually not all that great at determining probabilities and prediction, which is the best test of that synthesis of information.  Is religion any more logical than a flat Earth, given the sum total of human knowledge?  Would AI be as likely as us to develop market bubbles and stock crashes, to gamble at Vegas, or fail to plan for retirement?  Place the risk of dying from terrorist attack higher than dying in a car crash?


Quote
I've always thought of crossword puzzles as an analogy where only certain facts and explanations fit given a disparate set of other facts and explanations. Intelligence is about weaving those disparate strands of information and belief about the world together into a non-contradictory whole.
Pretty sure Watson could solve cross word puzzles faster than a human already, if it was just slightly tuned. 
 
 
Quote
Cognition at this level seems to require incorporation of as much as we can about every other belief we hold.
True, but that is still just one out of many modules, completely independent of things like sight, language, or walking.

Quote
I'm not convinced either way but wanted to point out an argument that cast doubt on the swarm-of-narrow-AI "solution" to general AI.
me neither.  Which is why I'm more worried about job loss than about a robot revolution.

Quote
Interestingly, there is a corollary to the argument above as follows: if central coherence is necessary for a true high-functioning general-AI, then that would have the effect of constraining the parameter space of all likely minds. The reason is a strongly centrally coherent AI can't become too lopsided in its skills and beliefs because gaining incremental skill in one particular direction might require a higher baseline of knowledge in a variety of disparate areas. This would tend to lessen the dangers of general AI because it would mean any general AI that was viable would be constrained in any extreme coupling of tendencies and abilities by its overall epistemological horizon. The parameter space of minds is still huge even under this assumption, so it's not entirely reassuring, even if true. This discussion between Sam Harris and David Deutsch is where I got this idea and why I think the two can't reconcile their differing opinions in the course of that discussion (Harris believes any arbitrary AI mind is plausible, while Deutsch believes they will necessarily have much in common with humans because of the unpredictable yet constraining nature of broad knowledge across domains: e.g. a super-AI that is truly good at building weapons will necessarily have some skill at questioning the morality of killing because the requisite intelligence for both has a common ground).
Interesting indeed.  Goes to my suggestion that perhaps if they ever were as generally "intelligent" as us, they might lose some advantages, by becoming emotional or impulsive or otherwise irrational, just like us

« Last Edit: August 02, 2017, 08:35:23 AM by Bakari »
Anything I've said here useful or interesting?  Find a lot more of my thoughts here: http://randomthoughts.fyi

Bakari

  • Handlebar Stache
  • *****
  • Posts: 1734
  • Age: 37
  • Location: Oakland, CA
  • Veggie Powered Handyman
    • The Flamboyant Introvert
Re: Robots and their impact on the future
« Reply #1492 on: August 02, 2017, 09:05:48 AM »
This is the current state of the art.  They had three years to develop, test, and perfect.  Some big budgets. 

This competition is two years old.  This is old technology with university budgets for the most part.  You throw a billion around and give them a few years, you will see some amazing things can happen.  They could conquer the challenges.  The challenges are not too bad if you know what the course is going to be.  If it is totally random them it will be more challenging.  The shows that we see are based on challenges that are reused with a few that are new.

Darpa's self driving vehicles were a failure.  A few years later, we have self driving cars going down the street.   


a) granted tech moves fast, but I haven't seen any thing any more likely to be capable of running that very simple course in the past 2 years.  What developments make this "old technology"?


b) universities for the most part, but also NASA, Lockheed Martin, Google, Boston Dynamics...


c) no one is throwing a billion around, so what might hypothetically exist if they did is beside the point.  Besides, all summed, more than a billion has been spent on robotics and AI research.


d) whether you are talking the real world challenges of a robot handyman, or the challenges in ANW, the point is you DON'T know what the course is going to be.  You don't even know the elements of the course (except, in ANW you know it will begin with some sort of steps and end with a warped wall).  It is different every time, and the details are a secret. Even the ones that may appear "reused" are different.  (For example, when I ran the course, the steps were much further apart, and the height varied from step to step, while the wall was another foot higher than the year before, while almost everything else was brand new). 
For doing repair work, there are infinite possible configurations of plumbing, electrical, carpentry, HVAC, wall coverings, etc, in different buildings, and infinite different things that can go wrong with them.


e) it was 13 years between DARPAs first grand challenge and today.  If 2 years in old technology, then 13 is more than a "few".  Also, there are still zero fully autonomous cars on the road with no human backup.  Everyone is still saying "within a few years" and "sooner than you think".  What we have is "semi-autonomous" cars, which do the most simple part of driving.

The parts that AI can't consistently handle well are things that any 16 year old can do with a few weeks training and practice.  The robots only need to be as good as an ordinary human, and can't fully do that yet.  In order to complete a ANW course, which 90% of athletes fail at, it has to be much more capable than the majority of humans.





Anything I've said here useful or interesting?  Find a lot more of my thoughts here: http://randomthoughts.fyi

Bakari

  • Handlebar Stache
  • *****
  • Posts: 1734
  • Age: 37
  • Location: Oakland, CA
  • Veggie Powered Handyman
    • The Flamboyant Introvert
Re: Robots and their impact on the future
« Reply #1493 on: August 02, 2017, 09:18:23 AM »
For the most part I agree with you, but I think the key point for nurses, police, lawyers, even CEOs, is human relation skills, which might reasonably be expected to not be mastered by AI.

Does it really need full sentience/AI to be able to achieve this sufficiently for most people? I suspect you could convince a large portion of people by simply making it good enough at faking it.

What if it can completely convince people 90% of the time on social skills, perform the actual functional part of the job with 99.9% success rate and at less than 10% of the cost? Would people still rather go see a human?
I could see AI being close enough in most cases, maybe with the exception of police, involving diffusing high stress situations and conflicts peacefully in real time.  I imagine human therapists and counselors being preferred (even if just on principal)

Quote
And lastly this is why I pointed out that robots wouldn't have to automate all of your tasks to effectively make most people obsolete. If a lawyer AI can do 90% or more of the actual work hours at a fraction of the cost and time this would mean 9 out of 10 lawyers could be fired and instead the last 1 could simply cover the work that is left. Doesn't mean the whole job is automated, but would mean the vast majority of them would be out of a job.
Yes, agree 100%!  I think we have seen this already, to a huge extent, going back all the way to the industrial revolution.  Its the reason the average work week is half the number of hours it was pre-industrial revolution, and the average number of working years is 50% less - even with an expanding economy, there is less need for human labor. And even with the lifetime work hours per person around 1/3 of what it once was, and the ever growing consumerism, unemployment has been gradually trending up for over 100 years.
As the rate of tech growth accelerates, no doubt unemployment will too
Anything I've said here useful or interesting?  Find a lot more of my thoughts here: http://randomthoughts.fyi

sol

  • Walrus Stache
  • *******
  • Posts: 5286
  • Age: 40
  • Location: Pacific Northwest
Re: Robots and their impact on the future
« Reply #1494 on: August 02, 2017, 10:23:26 AM »
In order to complete a ANW course, which 90% of athletes fail at, it has to be much more capable than the majority of humans.

I'm sorry, Bakari, but I think this statement belies a lack of imagination.  Why should a robot have to complete the same arbitrary tasks in the same arbitrary way to be considered "as good"?  Shouldn't there be more objective completion measures?  For example, an autonomous quad could fly the ANW course and deliver itself to the finish line button fairly easily.  Requiring it to run the whole way is like saying computers are better at math than humans because they can multiple faster.  It's an arbitrary metric only vaguely related to the real success we're trying to achieve.

And so I think it is with most AI tasks.  Computers don't have to be as good as humans at doing things the same way humans do them, they just have to as good as humans by any metric.  Robot plumbers don't have to recognize and problem solve every home repair job in order to displace human plumbers, if they can plumb an entire new apartment building in a day.  Robot drivers don't need to parallel park in a school zone to put long haul truckers out of business.

We went through this same shift with every other robot technology.  Mechanized looms are robots that didn't exactly make weavers obsolete, but virtually all of our weaving is now done by robots despite their obvious shortcomings compared to human weavers (artistry, pattern matching, repair work, creativity, material selection, etc.).  The machine isn't trying to be as good as a person, just cheaper and faster at some adequately similar version of the job. 

I think most people haven't yet recognized that what seems obvious with mechanical work like weaving will soon happen with intellectual work in many other fields.  It doesn't have to be as good or as smart as you to put you out of work.

Bakari

  • Handlebar Stache
  • *****
  • Posts: 1734
  • Age: 37
  • Location: Oakland, CA
  • Veggie Powered Handyman
    • The Flamboyant Introvert
Re: Robots and their impact on the future
« Reply #1495 on: August 02, 2017, 11:28:11 AM »
I've been acknowledging all along that many specific subtasks of complex work have been automated and many specific subtasks of complex not yet automated will be.  And yes, having less total work to do will mean less jobs.


However, a plumber doesn't just need to plumb a new building.  They also need to troubleshoot problems in 100 year old houses, and in 5 year old houses, of varying sizes and materials and codes and some that never met code to begin with.  And the job in question wasn't plumber, it was stationary engineer / building maintenance, and they do need to know every type of repair, from concrete to welding to wiring.  Since you don't know what is going to break, or when or where or how, it would be hard to be efficient to replace one person with dozens of extremely capable robots each with one specialty.


Your ANW analogy doesn't work - a person could easily run around the course, or climb along the rafters, and get a better time, but that isn't actually the task at hand (Usher tried it, as did a monkey, once.  Neither received the prize).  A robot wouldn't have to do it in exactly the "same way" as a human, but they would have to follow the same rules.  The task isn't just to hit the buzzer.  The task is to swing climb run over each individual obstacle.  Flying over would be like making an autonomous tank that can roll over other cars and claiming you perfected urban self-driving because it reached the destination.


There are two parallel discussions happening here.
One is about the ability of robots and AI to take the place of human labor.
The other is about robots and AI reaching human levels of ability.


I think the former is/will happen(ing) faster than most realize, but the second is much father away then many (esp. in this thread) believe.


The article about relatively safe jobs is looking for those which require human levels of ability (at least for some parts of the job description), which, although it may have not picked entirely accurately, is a legitimate differentiation. 
Yes, even within those jobs there will be some job loss as some parts are automated, but the job will still probably exist in 20 years.
Anything I've said here useful or interesting?  Find a lot more of my thoughts here: http://randomthoughts.fyi

prognastat

  • Pencil Stache
  • ****
  • Posts: 797
  • Location: Texas
Re: Robots and their impact on the future
« Reply #1496 on: August 02, 2017, 11:35:21 AM »
I've been acknowledging all along that many specific subtasks of complex work have been automated and many specific subtasks of complex not yet automated will be.  And yes, having less total work to do will mean less jobs.


However, a plumber doesn't just need to plumb a new building.  They also need to troubleshoot problems in 100 year old houses, and in 5 year old houses, of varying sizes and materials and codes and some that never met code to begin with.  And the job in question wasn't plumber, it was stationary engineer / building maintenance, and they do need to know every type of repair, from concrete to welding to wiring.  Since you don't know what is going to break, or when or where or how, it would be hard to be efficient to replace one person with dozens of extremely capable robots each with one specialty.


Your ANW analogy doesn't work - a person could easily run around the course, or climb along the rafters, and get a better time, but that isn't actually the task at hand (Usher tried it, as did a monkey, once.  Neither received the prize).  A robot wouldn't have to do it in exactly the "same way" as a human, but they would have to follow the same rules.  The task isn't just to hit the buzzer.  The task is to swing climb run over each individual obstacle.  Flying over would be like making an autonomous tank that can roll over other cars and claiming you perfected urban self-driving because it reached the destination.


There are two parallel discussions happening here.
One is about the ability of robots and AI to take the place of human labor.
The other is about robots and AI reaching human levels of ability.


I think the former is/will happen(ing) faster than most realize, but the second is much father away then many (esp. in this thread) believe.


The article about relatively safe jobs is looking for those which require human levels of ability (at least for some parts of the job description), which, although it may have not picked entirely accurately, is a legitimate differentiation. 
Yes, even within those jobs there will be some job loss as some parts are automated, but the job will still probably exist in 20 years.

I think the point Sol is trying to make though is that in real life shortcuts and "cheating" are accepted and even encouraged if it allows a job to be completed more efficiently.

jfolsen

  • 5 O'Clock Shadow
  • *
  • Posts: 62
Re: Robots and their impact on the future
« Reply #1497 on: August 02, 2017, 11:39:47 AM »
I think automation is and will continue to have a large and growing impact on the labor market. But eliminate? Not so likely. You will now need robot repair techs, robot software upgrades and programmers. And especially they still require human supervision, because there are limits to what robots can figure out, as others have stated.

There was a Japanese car transmission plant that was automated (this was a while ago). One robot assembly put on the main nut and torqued it down to a specific foot/pound of pressure. Problem: due to wear, the nut was slightly mis-aligned and started cross threaded. The robots dutifully applied the correct amount of torque. Several thousand transmissions were put into cars and shipped out before a human figured out what was going on.

Bakari

  • Handlebar Stache
  • *****
  • Posts: 1734
  • Age: 37
  • Location: Oakland, CA
  • Veggie Powered Handyman
    • The Flamboyant Introvert
Re: Robots and their impact on the future
« Reply #1498 on: August 02, 2017, 11:40:36 AM »
Here's why I think there is a big difference between being to automate an entire job (like, for example, toll collector), vs most of a job (like police officer or general repair person).

If a robot can do 100% of a job, the job is gone.  Period.

If a robot can do 90% of a job, then EITHER you have 10% of jobs left, OR you have 100% of jobs with 4 hours of work a week for each. 
In the latter, you can maintain full employment, and if wages go up enough (but not too much), it can be better overall for everyone, both employees and investors. 
This is basically what happened with the 40 hour work week following the industrial revolution, more pay per hour, but less hours, with a net gain for everyone.



But if robots can do the entire job, including every individual subtask with no oversight at all, distribution of remaining (human) work hours isn't even a consideration, and a far more drastic change in the way the economy is set up has to happen.
Anything I've said here useful or interesting?  Find a lot more of my thoughts here: http://randomthoughts.fyi

Bakari

  • Handlebar Stache
  • *****
  • Posts: 1734
  • Age: 37
  • Location: Oakland, CA
  • Veggie Powered Handyman
    • The Flamboyant Introvert
Re: Robots and their impact on the future
« Reply #1499 on: August 02, 2017, 11:42:24 AM »
I think the point Sol is trying to make though is that in real life shortcuts and "cheating" are accepted and even encouraged if it allows a job to be completed more efficiently.


And my point is that in real life there really are limits to the acceptable shortcuts.
I already addressed that with " like making anautonomous tank that can roll over other cars and claiming you perfected urban self-driving because it reached the destination."

A robot car does not have to do things the same way a person would, as in with hands on a steering wheel and eyes in a head, but they do have to follow the same traffic rules and predict what the bicyclist going the wrong way down the street is going to do.  There is more than one way to do it, but not any real "shortcuts". 
There is no shortcut to figuring out a wiring problem, or negotiating in a domestic violence situation.
« Last Edit: August 02, 2017, 11:50:07 AM by Bakari »
Anything I've said here useful or interesting?  Find a lot more of my thoughts here: http://randomthoughts.fyi