• Introducing Image Options for ResetEra 2.0! Check the left side navigation bar to show or hide images, avatars, covers, and embedded media. More details at the link.

Global Automation - Will it occur in the next 15-20 years, or more realistic for the latter portion of this century?

Oct 27, 2017
1,730
#1
When do you think it will realistically occur?

If it happens sooner or later, do you think it will finally force countries to consider universal basic income? The majority of populations won't be able to sustain themselves if the workforce labour they do now is unavailable, and a country cannot run itself without some sort of economic stimulus from it's citizenry. For those more knowledgeable on this subject, what sort of strategies are being considered to allow people to continue living, at least decently?

Personally, I think this will occur within 20 years or so.
 
Oct 27, 2017
2,670
#2
Depends on the field, self driving cars and trucks for things like taxis and long distance hauling are gonna be here within the next five years. Many fields will take longer, but the transportation industry is gonna be a massive unemployment sector soon. As much as we will need UBI, i doubt we will see it outside of Canada, Japan and a few select Euro countries with the rest of the world dealing with massive unemployment and a new Feudal system from the power corporations and the 1% will yield in the automation era.
 
Oct 26, 2017
5,786
#3
It won't be *global* for a long time, perhaps not for a century.

But I think we'll just see corporations and governments not give a shit.
 
#4
It will occur slowly but with major impact. Remember a world without smartphones? Sounds like ages ago, but it's only been around 10 years.

Self checkout is here.
Delivery Robots/drones are here.
Self driving are he, though not perfectly fully autonomous.

It's only a question of time. Hell there is even an AI that makes news articles.

Are you talking about major changes in the working field and unemployment? 15-20 years.
 
Oct 25, 2017
2,308
#5
My truck driving in-laws say they'll be replaced in 10 years. You can already shop without a cashier at places like Walmart/Target. It's coming fast. UBI will have to happen but it won't unless the right people are in charge.
 
OP
OP
.Detective.
Oct 27, 2017
1,730
#6
Depends on the field, self driving cars and trucks for things like taxis and long distance hauling are gonna be here within the next five years. Many fields will take longer, but the transportation industry is gonna be a massive unemployment sector soon. As much as we will need UBI, i doubt we will see it outside of Canada, Japan and a few select Euro countries with the rest of the world dealing with massive unemployment and a new Feudal system from the power corporations and the 1% will yield in the automation era.
iirc, Bill Gates was proposing a heavy tax to be exacted upon corporations that will use robots to essentially replace their worker base. That additional tax revenue could then be used in part to fund UBI. Your part about this happening in only a small number of countries like Canada(where I live), Japan and EU nations is worrying though.

The US itself has corporate lobbyists essentially guiding governmental policy, and despite being the richest country in the world, it's actually one of the first in mind that I could see just doing absolutely the bare minimum during an automation era, because of the 1%'s continued greed. Which is fucking depressing and horrifying at the same time.

Are you talking about major changes in the working field and unemployment? 15-20 years.
Correct. I am talking about when the tipping point/major impact will occur. We already see minor changes, and escalating technology. But I am talking about the point in time when this will occur on a global, or at least continental scale. And what that would entail.
 
Oct 27, 2017
1,290
#7
It will occur slowly but with major impact. Remember a world without smartphones? Sounds like ages ago, but it's only been around 10 years.

Self checkout is here.
Delivery Robots/drones are here.
Self driving are he, though not perfectly fully autonomous.

It's only a question of time. Hell there is even an AI that makes news articles.
I would like to point out that self-checkout is not automation. That's self-service. The same work is getting done, only the customer is doing it rather than the employee. Same things with order kiosks at fast food places.

It always seems weird to me that people are so worried about automation, but no one mentions self-service, when that's the thing that is actually eliminating many jobs today.
 
Oct 25, 2017
1,992
#8
Depends on the field, self driving cars and trucks for things like taxis and long distance hauling are gonna be here within the next five years.
???

No way those things take full control of the industry in five years. It’ll be ten years at the earliest for that stuff.
 
#9
I would like to point out that self-checkout is not automation. That's self-service. The same work is getting done, only the customer is doing it rather than the employee. Same things with order kiosks at fast food places.

It always seems weird to me that people are so worried about automation, but no one mentions self-service, when that's the thing that is actually eliminating many jobs today.
Your right. But instead of 10 people, you need just one to oversee the machines and that's it. I added self checkout because it will add on the job losses.
 
Oct 27, 2017
101
#14
Depends on the field, self driving cars and trucks for things like taxis and long distance hauling are gonna be here within the next five years. Many fields will take longer, but the transportation industry is gonna be a massive unemployment sector soon. As much as we will need UBI, i doubt we will see it outside of Canada, Japan and a few select Euro countries with the rest of the world dealing with massive unemployment and a new Feudal system from the power corporations and the 1% will yield in the automation era.
Transportation industry will be disrupted by automation very soon but will still need human operators 'just in case' until the tech is proven. I give it 8-10 years to past that point to totally eliminate human operators
 
Oct 27, 2017
148
#15
When do you think it will realistically occur?

If it happens sooner or later, do you think it will finally force countries to consider universal basic income? The majority of populations won't be able to sustain themselves if the workforce labour they do now is unavailable, and a country cannot run itself without some sort of economic stimulus from it's citizenry. For those more knowledgeable on this subject, what sort of strategies are being considered to allow people to continue living, at least decently?

Personally, I think this will occur within 20 years or so.
Some economist say that the society will cope and adapt (aka new kind of jobs)

As someone that work on industrial automation I think it won't be the hell that some people here think it will be.
 
OP
OP
.Detective.
Oct 27, 2017
1,730
#16
Some economist say that the society will cope and adapt (aka new kind of jobs)

As someone that work on industrial automation I think it won't be the hell that some people here think it will be.
What kind of jobs realistically? They will likely be associated with maintenance, programming and overseeing further automation related projects, correct? Which would require more education or time to train from scratch, if someone wasn't working in the sector to begin with.

What would occur to those who have lower qualified jobs, and how would they begin to transition? Especially the older generation.
 
Oct 27, 2017
3,624
#17
It’s gonna be an issue, but the problem is being overstated like...by a lot. You’d think with automation we wouldn’t be seeing an exponential increase in productivity with decreased overhead and the bottom line, but the exact opposite is true. The fear of automation has been present since the industrial revolution and they’ve always turned out wrong, and some more wrong than others.

https://www.vox.com/the-big-idea/20...s-artificial-intelligence-ai-mcjobs-low-skill

This article goes into a lot more detail.

That’s not to say we should pretend everything is fine, but we need to change the way we look at work rather than going full steam ahead with UBI until we see evidence that it’s going to be needed. So far, we are at least half a century off
 
#19
It’s gonna be an issue, but the problem is being overstated like...by a lot. You’d think with automation we wouldn’t be seeing an exponential increase in productivity with decreased overhead and the bottom line, but the exact opposite is true. The fear of automation has been present since the industrial revolution and they’ve always turned out wrong, and some more wrong than others.

https://www.vox.com/the-big-idea/20...s-artificial-intelligence-ai-mcjobs-low-skill

This article goes into a lot more detail.

That’s not to say we should pretend everything is fine, but we need to change the way we look at work rather than going full steam ahead with UBI until we see evidence that it’s going to be needed. So far, we are at least half a century off
Yes, but you can't compare this to the industrial revolution. This time is unique. I see it as an endgame. Of course not all jobs would be automated, but there definitely be a high unemployment.

This time, not just the low skilled jobs are being hit. There is already AI which can do the work of a lawyer.
 

GMM

Member
Oct 27, 2017
2,184
#23
It will happen to some industries sooner than others and people are oblivious if they don’t realize just how big of a threat it will be for the low income class over the next two decades, the impact will be huge.

Today we are already seeing it with self checkout machines at grocery stores and companies automating their support services with AI powered systems, these initiatives do remove the need for employing the same amount of people to service the same amount of customers.

In the very near future this will also apply for people working logistics with driverless long haul trucks, these people are generally unfit to take other positions at those companies and will be let go.

This is not like the industrial revolution where we made machines to amplify the output of a skilled workforce, this is about completely replacing the workforce in order to maximize profits.

I have nothing against these technology breakthroughs since they have no negative impact on me as a customer, I actually really love self checkout systems vs a cashier station, but I have major issues with how we as a society are failing to build towards a world where people might not be able to support themselves because they are being rendered obsolete by greedy companies who currently face no penalty for removing them in favor of technology.
 
Oct 25, 2017
6,618
#26
Like everything, automation is implemented incrementally and at different rates in different industries and societies around the world. There isn't a point where things will be automated or not automated. Today, there is more automation than there was yesterday. Yesterday there was more automation than there was 100 years ago. 100 years ago, there was more automation than there was 1,000 years ago. A lot of people mistakenly think of automation as just new automation, like how automation is going to make service industry jobs obsolete... but forget that most service industry jobs are a form of automating something that was done differently 10, 20, 50, or 100 years ago.
 
Oct 25, 2017
3,921
#27
The famous Frey/Osbourne study is too pessimistic in my opinion. That said, automation will come or rather continue and it will have a major impact.
 
Oct 25, 2017
6,497
#28
Nope - probably not even in my lifetime.

Robots are so expensive to build and maintain, I have a hard time imagining them replacing humans. We can only barely see robots capable of doing menial chore-work or highly rote automated assembly-line type stuff.

The capabilities of "AI" also tend to be massively overstated by doomsayers. See Above Article -- somebody makes a program that is good about scanning legal documents, and the headline is "Look out lawyers, you're gonna be obsolete!"
 
#29
Nope - probably not even in my lifetime.

Robots are so expensive to build and maintain, I have a hard time imagining them replacing humans. We can only barely see robots capable of doing menial chore-work or highly rote automated assembly-line type stuff.

The capabilities of "AI" also tend to be massively overstated by doomsayers. See Above Article -- somebody makes a program that is good about scanning legal documents, and the headline is "Look out lawyers, you're gonna be obsolete!"
It starts with scanning legal documents and then understanding them. AI moves very fast. Look at GO player.
 
Nov 13, 2017
522
#31
I watched a good talk a while ago from head of AI at Baidu or Tencent. He said anything that takes 3 seconds is in the process of being automated. So think of a security guard glancing at a monitor. Then alerting someone if there’s an intruder. That can be done by a computer. Many activities are comprised of quick, discrete actions. For example driving can be broken down to processing/sampling visual field then taking appropriate action. Will take a long time for high touch, non linear actions/activities.
 
Oct 27, 2017
3,072
Spain
#34
Machine Learning and AI are very overrated and misunderstood techniques. It's a bunch of computer scientists who have no knowledge of statistics creating models that perform very well in-sample, but which do not infer any causal relation whatsoever, using instead a bunch of un-controlled spurious relationships that exist in-sample to infer results out of sample.
As a result, it's unscientific and arguably morally wrong to use machine learning or AI models to automate jobs that require critical decision making. As a notorious case in point, there's the Amazon CV evaluation tool that discriminated against women, and many others. There are valuable uses to AI and machine learning, and the techniques they are creating will at some point go into proper scientific modelling, but the whole "In ten years there will be no doctors and AI will do their job instead" is complete nonsense.
 
Oct 27, 2017
148
#35
What kind of jobs realistically? They will likely be associated with maintenance, programming and overseeing further automation related projects, correct? Which would require more education or time to train from scratch, if someone wasn't working in the sector to begin with.

What would occur to those who have lower qualified jobs, and how would they begin to transition? Especially the older generation.
Some will be maintenance, also of course automation development.
But my guess most of jobs will be created with the improved efficiency.

If prices of good decrease people will have more money to spend on other stuff.
 
Mar 29, 2018
1,196
#36
IIRC the stats indicate the number of jobs keep increasing regardless of automation (the world always getting more complex)

But yeah something gotta give
 
OP
OP
.Detective.
Oct 27, 2017
1,730
#37
If prices of good decrease people will have more money to spend on other stuff.
What incentive or punishment does a corporation have to decrease the price of their goods, if they can maximize profit margins while reducing cost?

This is something AOC was stating in an article published recently. She referenced Bill Gates' robot tax, but mentioned it should be more about taxing companies much more than we currently do, because when(not if) they decide to rollout more automation, there isn't any checks in place to ensure they don't just dump their workforce. She seems automation as a good thing, because it should lead to more time for people to do the things like enjoy, but also points out that current salaries are basically just a concept where we are desperate enough to accept whatever is offered, whereas the majority of profits go to the 1%.

“We should be excited about automation, because what it could potentially mean is more time educating ourselves, more time creating art, more time investing in and investigating the sciences, more time focused on invention, more time going to space, more time enjoying the world that we live in,” she said. “Because not all creativity needs to be bonded by wage.”

But her answer placed it in the familiar context of a larger fight against economic inequality and corporate greed.

“We should be working the least amount we’ve ever worked, if we were actually paid based on how much wealth we were producing,” she said. “But we’re not. We’re paid on how little we’re desperate enough to accept. And then the rest is skimmed off and given to a billionaire.”
https://www.theverge.com/2019/3/10/18258134/alexandria-ocasio-cortez-automation-sxsw-2019
 
Oct 25, 2017
3,352
#38
Alot of people are gonna suffer from this and climate change but eventually sanity will prevail and basic income will be introduced. Doesn't fix the climate change issue, so yeah. I'm optimistic.

A person's worth only being determined by what they do is so ingrained in our society, any efforts to change that are always met with heavy resistance. Basically you need it to start to hit middle class income people before anyone gives a shit.
 
Oct 25, 2017
10,585
#39
All I know is that when it happens, the US will quickly lose its superpower status. The economy is gong to completely tank, because things like UBI are completely unattainable in this fucked up country. I mean, we're completely incapable of giving our citizens basic health care rights. Automation is going to completely break us, though it will be an incredible advancement for most other civilized nations.
 
Dec 17, 2018
522
#40
Of course not. I keep reading people hype up the automation utopia for no reason. You can only automate task and jobs that are similar, such as driving on a standardize road. You can't automate a robot to install a garbage disposal for you, because each installation job is so different. So all you are doing is moving the repetitive jobs to the machine slowly and industry by industry but you keep the complicated jobs (not complicated as in difficult but in too costly to program robot to do)

Think about the percentage of population who work on agriculture 100 years ago and the percentage now. Does anyone start yelling the machines are taking over the farming jobs?

My theory about the liberal types who play up the effect of automation to sidestep the problem of unfair distribution of wealth in the current society.
 
Oct 25, 2017
6,497
#41
It starts with scanning legal documents and then understanding them. AI moves very fast. Look at GO player.
Your landlord hands you a new lease. Are you comfortable signing it sight-unseen because an AI said it looked good?

It's an AI that can rate contracts for "favorability". Lawyers do more than read contracts all day, and most clients will insist on having some form of human analysis before entering into a binding contract. Not to mention that people will learn to write contracts in a way that obfuscates information by confusing the AI.

Like most everything else, they've found a way to automate some menial part of a profession, but not entirely eliminate a profession.
 
Oct 25, 2017
610
#42
Machine Learning and AI are very overrated and misunderstood techniques. It's a bunch of computer scientists who have no knowledge of statistics creating models that perform very well in-sample, but which do not infer any causal relation whatsoever, using instead a bunch of un-controlled spurious relationships that exist in-sample to infer results out of sample.
As a result, it's unscientific and arguably morally wrong to use machine learning or AI models to automate jobs that require critical decision making. As a notorious case in point, there's the Amazon CV evaluation tool that discriminated against women, and many others. There are valuable uses to AI and machine learning, and the techniques they are creating will at some point go into proper scientific modelling, but the whole "In ten years there will be no doctors and AI will do their job instead" is complete nonsense.
While I agree that a lot of people tend to overrate AI, saying that scientists working on it have no knowledge of statistics is incorrect. Plenty of them understand statistics perfectly well, and pretending the whole field is terrible just because some of them don't is a weird take.
 
Dec 12, 2017
1,015
#43
While I agree that a lot of people tend to overrate AI, saying that scientists working on it have no knowledge of statistics is incorrect. Plenty of them understand statistics perfectly well, and pretending the whole field is terrible just because some of them don't is a weird take.
Indeed. It's a bit like mathematicians looking down on statisticians. Computer science has its roots in mathematics just like statistics.
 
Oct 25, 2017
4,660
#44
I think we have to consider scale and depth. Most people think this will be an issue when we see full displacement, but all you need is to delegate enough of the laborer's skillset to technology to justify weaker pay and hours. For example, platooning will eradicate working truck drivers, yet we're all worried about fully driverless trucks. This becomes an issue before we even enter that endgame scenario.

I'm of the opinion of the Obama administration: it currently is the leading middle class issue regarding erosion of said class. It's already here, compounding the already existing issues of a precariat crisis. Technology and automation today has overlapped with the reality that full time employment isn't even a quarter of job growth in the United States, so it looks to solidify the issues of a gig economy and the precarious tightrope that puts on people.
 
Oct 27, 2017
3,072
Spain
#45
While I agree that a lot of people tend to overrate AI, saying that scientists working on it have no knowledge of statistics is incorrect. Plenty of them understand statistics perfectly well, and pretending the whole field is terrible just because some of them don't is a weird take.
AI research has advanced disregarding most or all of the concerns, limitations and implications that econometric analysis utilizes to infer causal relations. It's not terrible nor good, it's what it is. The implication is that, while machine learning models may perform very well at forecasting or inferring, this doesn't mean they are picking up any causal relations, in fact most likely they are not, and some unexpected change out of sample might make them produce unusable results.

Even if a machine learning model ends up producing the results the creators intended, the very nature of their estimation means it's impossible to prevent them from picking up all sorts of biases that are present in the training samples, and there is no way to control for those biases.

Machine learning research is very valuable, and it's creating useful techniques, but it's not some sort of panacea. More importantly, it has to and will eventually merge with the broader field of econometrics, which will benefit both, econometrics will get all the awesome data management and non-linear estimation techniques machine learning uses while machine learning will get all the theory that lets econometric techniques infer causality.

That's why machine learning is not a panacea that will substitute all jobs, in order to substitute humans you need to get a model that really understands how things work, but machine learning models may and will fail under certain conditions, and on their own they will never amount to containing any explatory power, by virtue of their very nature. They are a powerful tool, that's all. I'm wary about all the optimism on machine learning because of all this is the ways it is being mis-used by big actors in the private sector, as we speak, in many cases automating things that should never be left to a model that doesn't infer any causality at all.
 
Oct 25, 2017
610
#46
AI research has advanced disregarding most or all of the concerns, limitations and implications that econometric analysis utilizes to infer causal relations. It's not terrible nor good, it's what it is. The implication is that, while machine learning models may perform very well at forecasting or inferring, this doesn't mean they are picking up any causal relations, in fact most likely they are not, and some unexpected change out of sample might make them produce unusable results.

Even if a machine learning model ends up producing the results the creators intended, the very nature of their estimation means it's impossible to prevent them from picking up all sorts of biases that are present in the training samples, and there is no way to control for those biases.

Machine learning research is very valuable, and it's creating useful techniques, but it's not some sort of panacea. More importantly, it has to and will eventually merge with the broader field of econometrics, which will benefit both, econometrics will get all the awesome data management and non-linear estimation techniques machine learning uses while machine learning will get all the theory that lets econometric techniques infer causality.

That's why machine learning is not a panacea that will substitute all jobs, in order to substitute humans you need to get a model that really understands how things work, but machine learning models may and will fail under certain conditions, and on their own they will never amount to containing any explatory power, by virtue of their very nature. They are a powerful tool, that's all. I'm wary about all the optimism on machine learning because of all this is the ways it is being mis-used by big actors in the private sector, as we speak, in many cases automating things that should never be left to a model that doesn't infer any causality at all.
Thanks for elaborating. I agree with a lot of this actually, though saying that machine-learning researchers have no knowledge of statistics still seems a bit harsh. I think a lot of people that are active in the field are well aware of the limitations of machine learning (even some big names have been very vocal about that), though the way some people try to push it really hard in the private sector is definitely scummy in a lot of ways.

I do think some people go a bit too hard trying to trash machine learning though. It's definitely flawed, but just because it's a limited tool doesn't mean its not extremely useful when applied to the correct problem, as many recent practical examples have shown. One of the answers from this interview (which talks about the problem of machine learning lacking casual inference) sums it up pretty well, I think: it's pretty much just curve fitting (well, plus a lot of things related to that), but it's really impressive how many problems can be solved extremely well with just curve fitting. This means it will be used to automate a lot of things, no matter how morally wrong it might be.

This is slightly incorrect though:
Even if a machine learning model ends up producing the results the creators intended, the very nature of their estimation means it's impossible to prevent them from picking up all sorts of biases that are present in the training samples, and there is no way to control for those biases.
There's been quite a few work done in trying to create unbiased/fair classifiers in recent years (well, less biased at any rate). This actually does require careful knowledge of statistics as well. Not that I expect it will see use in commercial applications any time soon, but it's something at least.

And for what it's worth, there does seem to be a recent rise of interest in combining machine learning with casual inference.
 
Jan 5, 2019
3
#47
Foffy's comment about platooning is very important but I worry that people will ignore it because they've never heard of that term.

The problem is that too many people assume a robot has to directly and perfectly replace a human to take your job. This isn't how it works. You don't walk into a hardware store and interact with a robot cashier manning a register. In reality, we have stores collapsing due to being undercut by companies like Amazon who run their warehouses and fulfillment centers with robots.

For example, robot drivers don't have to be perfect to displace millions of truck drivers. Instead, you'd have a fleet of 20 robot trucks following one human point driver most of the way. Then outside a city, they switch to remote control for the hard 5% of the trip that involves traffic, parking, etc.

Ai doesn't have to be as good as humans to take human jobs. It just needs to be able to perform a useful task, then businesses will find a way to shift their business model to leverage that labor instead of human labor.
 
Oct 27, 2017
3,072
Spain
#48
Thanks for elaborating. I agree with a lot of this actually, though saying that machine-learning researchers have no knowledge of statistics still seems a bit harsh. I think a lot of people that are active in the field are well aware of the limitations of machine learning (even some big names have been very vocal about that), though the way some people try to push it really hard in the private sector is definitely scummy in a lot of ways.

I do think some people go a bit too hard trying to trash machine learning though. It's definitely flawed, but just because it's a limited tool doesn't mean its not extremely useful when applied to the correct problem, as many recent practical examples have shown. One of the answers from this interview (which talks about the problem of machine learning lacking casual inference) sums it up pretty well, I think: it's pretty much just curve fitting (well, plus a lot of things related to that), but it's really impressive how many problems can be solved extremely well with just curve fitting. This means it will be used to automate a lot of things, no matter how morally wrong it might be.

This is slightly incorrect though:

There's been quite a few work done in trying to create unbiased/fair classifiers in recent years (well, less biased at any rate). This actually does require careful knowledge of statistics as well. Not that I expect it will see use in commercial applications any time soon, but it's something at least.

And for what it's worth, there does seem to be a recent rise of interest in combining machine learning with casual inference.
Well, of course the fields are merging, it's the natural course, imo. Machine learning is a new and shiny toolbox of non-linear estimation. So naturally there is interest to merge the field with techniques from econometrics, I'm aware that exists. The problem is that, as you say, commercial implementations of these models can often be "problematic". The problematic part is not just about being morally wrong or not, it's that without causal inference there's many uses for which a model cannot be accepted for true automation. Like yeah, a ML model can recommend your YouTube feed and the most terrible thing that can happen is that it will only recommend far right propaganda (That's bad enough) , but it cannot reliably diagnose an X-Ray, or a CV, and god knows what it will say the moment it's extrapolating out of sample whenever anything changes that the model (spuriously) relies on for its inference.

I think at some point its going to fluke in the private sector really bad, we have idiot executives thinking in ten years they will be the only employees in their companies and that's just complete rubbish, and that will be a good thing for the field because it will weed out the snake oil salesmen.
 
Nov 11, 2017
5,225
#49
You arent gunna wake up to automation. Its already happening in retail construction banking.. Etc. So the next 20 years will increase pace with development in ai and processor speed/gpu speed.
 
Oct 27, 2017
101
#50
It’s gonna be an issue, but the problem is being overstated like...by a lot. You’d think with automation we wouldn’t be seeing an exponential increase in productivity with decreased overhead and the bottom line, but the exact opposite is true. The fear of automation has been present since the industrial revolution and they’ve always turned out wrong, and some more wrong than others.

https://www.vox.com/the-big-idea/20...s-artificial-intelligence-ai-mcjobs-low-skill

This article goes into a lot more detail.

That’s not to say we should pretend everything is fine, but we need to change the way we look at work rather than going full steam ahead with UBI until we see evidence that it’s going to be needed. So far, we are at least half a century off
That article, and everything else put out by Vox on the subject of Automation are incredibly embarrassing. How anyone can put forward as a serious argument, that because people said they were worried about something called automation in the past and were wrong about it, then we are probably wrong about it now, is astonishing.