When will the robots take over?

[mad_scientist]

When will most jobs be lost to robots due to automation?

 

Are robots working and doing all the work for us a good thing?

 

Would you be happy doing what you wanted with a guaranteed basic income instead of having to earn a living and robots making the money for us?

 

What would you do with all the free time?

 

[EdEarl]

Robots and AI are already taking jobs, and the number of jobs they take will increase until people have no jobs. For example, car autopilots are expected to take about 12M jobs by 2025. In general, I believe it will be a good thing. People didn't have paid jobs until the industrial revolution. I am retired, and happy not going to work. A guaranteed basic income would be nice. I take care of myself, interact with people, and grow a garden.

[Sriman Dutta]

The pace with which our technology is growing, I believe the age of robots isn't too far.

[dimreepr]

Are robots working and doing all the work for us a good thing?

 

If they work for us yes; if they work for their investors no.

 

Would you be happy doing what you wanted with a guaranteed basic income instead of having to earn a living and robots making the money for us?

 

 

That's a loaded question, it revolves around what you mean by 'earn'?

 

What would you do with all the free time?

 

 

Spend it.

[Endy0816]

Yeah, I would have a blast.

 

In theory the free market should still assign you a value though. We have artificial diamonds these days yet people still value the "real" thing, despite artificial diamonds being superior quality.

 

Some positions are not going to be easy to automate, so those will keep going for some time. For others, I think specifically human performed/produced artwork and services will be a big thing in the future.

Edited April 14, 2017 by Endy0816

[Klaynos]

Technology removing jobs is not a new thing. What proportion of people work on food growing now compared to the 1500s or clothing production compared to the 1800s?

 

Whilst a lot of the jobs we do now may be automated there will be other, probably safer more relaxed jobs for us.

[Sensei]

Tasks that can be easily automated, are typically hated by people having to do them. They are dreaming about better job..

Imagine work in warehouse:

Would you like to work in warehouse?

8-12 hours per day scan box, move to the right place on the feet and put on the right shelf.. ? Day by day, the same stupefying job.

If I would have such job in warehouse, I would invent robot that would do it for me instead, and just remotely monitor what he is doing.

I think so human should do jobs that they at least like, if not, love.
Take for example scientists. If they would be fired from university, in the majority of cases, they would work for free at home.
Especially theoretical physicists and mathematicians, working on their own theories, trying to discover something.

Edited April 14, 2017 by Sensei

[EdEarl]

Technology removing jobs is not a new thing. What proportion of people work on food growing now compared to the 1500s or clothing production compared to the 1800s?

 

Whilst a lot of the jobs we do now may be automated there will be other, probably safer more relaxed jobs for us.

True, jobs have been automated since the industrial revolution. However, AI and robots will be able to do anything we can do, and they don't need to be paid for working 24/7. We will be without jobs as we know it, just as most people before the industrial revolution didn't have jobs. The difference is that before industrial jobs, people worked hard to grow food to eat. Robots will do it for us in the future.

[Sensei]

True, jobs have been automated since the industrial revolution.

 

Wood watermill and windmill, earlier than industrial revolution, are examples of automatization of human work as well.

Edited April 14, 2017 by Sensei

[EdEarl]

Youtube video A.I. is progressing faster than you think gives some interesting examples of how quickly AI is progressing. In addition Google has developed an AI coprocessor that is an order of magnitude faster than existing graphics processing units recently used for AI. The rate of improvement in AI is surprising; thus, IMO it will occur faster than we imagine.

[Delta1212]

Youtube video A.I. is progressing faster than you think gives some interesting examples of how quickly AI is progressing. In addition Google has developed an AI coprocessor that is an order of magnitude faster than existing graphics processing units recently used for AI. The rate of improvement in AI is surprising; thus, IMO it will occur faster than we imagine.

I've been following AI development pretty closely for roughly 5-10 years now, and it's well past the point where I've begun seriously wondering about about whether some of them have any kind of subject experience or not.

 

(Keeping in mind that I'm talking about a level on par with my wondering about the internal lives of bugs rather than anything approaching human-level intellect).

[dimreepr]

I've been following AI development pretty closely for roughly 5-10 years now, and it's well past the point where I've begun seriously wondering about about whether some of them have any kind of subject experience or not.

 

(Keeping in mind that I'm talking about a level on par with my wondering about the internal lives of bugs rather than anything approaching human-level intellect).

 

I think it's more like, a human-level intelligence with an inhuman intellect.

Edited April 14, 2017 by dimreepr

[Delta1212]

 

I think it's more like, a human-level intelligence with an inhuman intellect.

Even the best ones currently don't really have human-level intelligence, though. There are some that can do specific tasks better than unaided human can, but that isn't really the same thing.

[dimreepr]

Even the best ones currently don't really have human-level intelligence, though. There are some that can do specific tasks better than unaided human can, but that isn't really the same thing.

 

But there are areas/tasks in which the the level, of intelligence, is equivalent; the question then is, how many specialised AI computers does it take to equal a polymath human?

 

The number of computers is irrelevant, because they talk; so the only relevant question is, 'what's the lead time?'.

https://www.youtube.com/watch?v=sQNtu4kpd64#t=767.217329

[Delta1212]

 

But there are areas/tasks in which the the level, of intelligence, is equivalent; the question then is, how many specialised AI computers does it take to equal a polymath human?

 

The number of computers is irrelevant, because they talk; so the only relevant question is, 'what's the lead time?'.

They have to be designed to actually talk, though. Nobody has built a supernetwork linking up all of the various skills currently available, nor built a network capable of assembling all of those skills into any kind of coherent whole.

 

Also, be careful conflating skill with intelligence. There are many animals with individual behaviors that approach or surpass what a human is capable of accomplishing, but that is not the same as having human-level general intelligence. The fact that we have designed networks whose single specialized skill happens to fall into an area that we have traditionally assigned to intelligence has more to do with the lack of evolutionary pressure for Go-playing ants than because the skill in question requires human-like intelligence to do well with.

 

I am a big proponent of AI and its potential, and I do wonder a bit about what the threshold for subject experience is and whether we have crossed it yet if we ever possibly can. Especially with some of the more advanced object recognition systems where are essentially building internal models of meaning out of the raw visual data that they are presented with. Is that enough to get quails out of, or is there something more that is required?

 

Regardless, it is important to understand what our more advanced AIs are actually doing and how that relates to intelligent behavior and learning ability. This is a subject where there tends to be a major disconnect between the perception of how smart an AI needs to be to perform a certain task and how smart it actually needs to be.

 

AlphaGo and its ilk are extremely impressive, but nothing thus far is anywhere close to a true general intelligence AI.

 

I think the path we're currently on certainly leads there, and that the challenges are mostly ones of engineering, funding and technique rather than high level concept, but just because you can see the broad strokes of how to get from here to there doesn't mean that we are already there.

[dimreepr]

I think the path we're currently on certainly leads there, and that the challenges are mostly ones of engineering, funding and technique rather than high level concept, but just because you can see the broad strokes of how to get from here to there doesn't mean that we are already there.

 

Agreed, "There's many a slip twix cup and lip".