Robots and their impact on the future

[Bakari]

Here, so people don't have to click on a link:






People keep posting about how we aren't that close yet.
The thing about exponential growth is that it doesn't look close even one year before you get there.

Remember the riddle from elementary school? 
"If a pond lily doubles everyday and it takes 30 days to completely cover a pond, on what day will the pond be 1/2 covered?[/size] [/color][/size]The answer is day 29.

We are here on day 26 saying "hey, the pond is barely 1/10th filled, we aren't close yet" even though we are only 4 days from total saturation.[/color]

[forummm]

Even in China, human labor is "too expensive". The program is called "replacing humans with robots".

http://www.nytimes.com/2015/04/25/technology/robotica-cheaper-robots-fewer-workers.html

[forummm]

Here, so people don't have to click on a link:






People keep posting about how we aren't that close yet.
The thing about exponential growth is that it doesn't look close even one year before you get there.

Remember the riddle from elementary school? 
"If a pond lily doubles everyday and it takes 30 days to completely cover a pond, on what day will the pond be 1/2 covered?[/size] [/color][/size]The answer is day 29.

We are here on day 26 saying "hey, the pond is barely 1/10th filled, we aren't close yet" even though we are only 4 days from total saturation.[/color]

This assume exponential trends will continue to infinity--something that doesn't happen in nature. Something always gets in the way. Maybe computational advances will continue at this rate for long enough to create the advances needed for AI exceeding human intelligence. Maybe not. There's a possibility, but don't take it as gospel truth.

[scottish]

I've read the discussion about how AI is growing exponentially.   But...  how do you 'measure' AI?    It sounds like we are using the number of neurons in a neural net as a proxy measurement for AI.   Is this the case?

[matchewed]

I've read the discussion about how AI is growing exponentially.   But...  how do you 'measure' AI?    It sounds like we are using the number of neurons in a neural net as a proxy measurement for AI.   Is this the case?

It's more that the capability of computers are growing exponentially. And if we consider that our computers are narrow AI then narrow AI capability is growing exponentially. There is still a barrier to the general side of it.

Consider it this way, in the 80's a computer would probably be able to be the "best tic-tac-toe" player in the world. Today a computer is the best chess player in the world. The complexity of the systems they are able to master is rising.

So it is not just number of neural net neurons, but our ability to program, computations a second, and precision with mathematical models which are measurements or benchmarks for AI growth. I wouldn't ever narrow it down to any one thing because then it's just way too simplified for discussion and is more demonstrable for showing.

[scottish]

Many computer engineers feel that Moore's law is finally coming to an end.   Citation:  http://www.theregister.co.uk/2015/04/23/moores_law_hits_50_intel/?page=3  I think this was mentioned before.     We get closer to the physical limits on transistor size every year.   And there's a big debate over parallelism, but this isn't very interesting to discuss.

It would be very interesting to find a way to quantitatively measure artificial intelligence.   Maybe it grows linearly with the capability of a computer.  Or maybe it's logarithmic (i.e. very sublinear).   Like an IQ measurement for AI.

[matchewed]

Many computer engineers feel that Moore's law is finally coming to an end.   Citation:  http://www.theregister.co.uk/2015/04/23/moores_law_hits_50_intel/?page=3  I think this was mentioned before.     We get closer to the physical limits on transistor size every year.   And there's a big debate over parallelism, but this isn't very interesting to discuss.

It would be very interesting to find a way to quantitatively measure artificial intelligence.   Maybe it grows linearly with the capability of a computer.  Or maybe it's logarithmic (i.e. very sublinear).   Like an IQ measurement for AI.

Let me pose it to you this way. Can you quantitatively measure human intelligence? You might want to jump in and answer IQ, but that only measures intelligence defined one particular way. A narrow oversimplification of what intelligence is isn't necessarily useful as a measurement. It may be interesting but it would be lacking. How do you measure synthesis of ideas given prior information? Is it the quantity of the ideas? The complexity? The "usefulness"?

It would get even more absurd if/when AI surpasses human intelligence. How do you measure something you can't even begin to understand?

Maybe we can stick with the human intelligence as an analogy. What measurement would you use to evaluate human intelligence from 60,000 years ago to today?

[scottish]

It's pretty hard to predict exponential growth in machine intelligence if you can't measure it and don't know its relationship to computational power, isn't it?   For people the IQ test measures how well you do on an IQ test and assigns you a number based on Gaussian statistics for intelligence.   I don't see any obvious way to apply this approach to machine  intelligence.

I'm feeling argumentative today.   Taking the animated graphic of a pond filling at an exponential rate and applying it to machine intelligence bugs me.

[matchewed]

Well the animated graphic is talking about computations a second as the measuring stick. If you feel that is solely tied to the physical limits on transistor size then you've got your view on it. If you think that perhaps there may be some other way then it could in theory keep going.

[Bob W]

Only about 5 to 8 % of people in the US have "productive" jobs as it is.   Most of us are government workers or in businesses that owe their financial existence to government.     The balance is in professions like banking, insurance,  marketing and sales.     Very few people work at jobs that actually make things.     The number one thing sold by far in the US is debt.    So I'm guessing at best robots can replace 8% of us.     Of course once AI is fully in swing within 25 years most all jobs will be obsolete.     

[arebelspy]

Only about 5 to 8 % of people in the US have "productive" jobs as it is.

Citation?  And what does "productive" mean?

[forummm]

Only about 5 to 8 % of people in the US have "productive" jobs as it is.   Most of us are government workers or in businesses that owe their financial existence to government.     The balance is in professions like banking, insurance,  marketing and sales.     Very few people work at jobs that actually make things.     The number one thing sold by far in the US is debt.    So I'm guessing at best robots can replace 8% of us.     Of course once AI is fully in swing within 25 years most all jobs will be obsolete.   

Pretty much all of us are in jobs that owe their existence to government. Without government there would be so much inefficiency due to having the overhead of defending our own property, not having a stable financial system, courts, etc. Government, with all its problems and inefficiencies, is what has allowed an environment where the markets could function in a way for us to go from 100% of people in agriculture to only 2%.

A lot of the jobs you mention aren't making physical objects, but they are facilitating the ability of people to make physical objects. Whether it's marketing those objects (because people wouldn't buy as much of them otherwise and the manufacturing jobs would decrease), providing finance (so the business could afford to start up in the first place), or government ensuring safety and property rights and recourse for damages of defective products (without which fewer people would be willing to buy and sell). There is inefficiency in the system for sure.

And I believe the biggest employment role in government is teachers. At 3.3 million, that's far more than the total number of all federal government employees combined (about 2 million). Those teachers are facilitating the production of all things made in the economy.
http://nces.ed.gov/fastfacts/display.asp?id=28

[Albert]

Computing power increasing exponentially has been a useful rule of thumb, but it's not a fundamental law of nature and quite obviously one day will not be true anymore. Will it before or after we achieve general AI I cannot say, though. Nothing in nature expands exponentially forever...

And let me remind you all yet again that intelligence (or computing power if you wish) alone is not sufficient. If someone were to make a brain transplantation on every single cow on the planet making them as smart as the best among us there would still be no "cow civilisation"

[matchewed]

Computing power increasing exponentially has been a useful rule of thumb, but it's not a fundamental law of nature and quite obviously one day will not be true anymore. Will it before or after we achieve general AI I cannot say, though. Nothing in nature expands exponentially forever...

And let me remind you all yet again that intelligence (or computing power if you wish) alone is not sufficient. If someone were to make a brain transplantation on every single cow on the planet making them as smart as the best among us there would still be no "cow civilisation"

True but there is no entity behind the cows actively making them smarter either.

[scottish]

I like the cow analogy.   Cows would need the ability to use tools and machines and to communicate with each other beyond <moo>

[matchewed]

I like the cow analogy.   Cows would need the ability to use tools and machines and to communicate with each other beyond <moo>

But that's where the analogy fails, computers have those things.

[Bob W]

Only about 5 to 8 % of people in the US have "productive" jobs as it is.

Citation?  And what does "productive" mean?

. Produce means to make things or food.    http://www.nam.org/Newsroom/Facts-About-Manufacturing/     this from the national association of manufacturers.   Says 12 million people are employed directly in manufacturing.  That is what?   3.6 % of the population.  Take out those involved in munitions, manufacturing directly for the gov and the fudge factor of the Nam and we are at a realistic 2%.   Throw in farmers and food processor s and we might arrive at 5%.   Interesting enough we produce almost twice as much goods as we did 10 years ago with 1/2 the work force.     Point is in 10 years we may double production again and halve the manufacturing work force.      22% of the us economy is medical (government),  90% of teachers are either direct or indirect government.   All banks are quasi government.   Good or bad that is reality. Farm and ranch families comprise just 2 percent of the U.S. population. More than 21 million American workers (15 percent of the total U.S. workforce) produce, process and sell the nation's food and fiber.
Fast Facts About Agriculture - American Farm Bureau.           

[Bob W]

Only about 5 to 8 % of people in the US have "productive" jobs as it is.   Most of us are government workers or in businesses that owe their financial existence to government.     The balance is in professions like banking, insurance,  marketing and sales.     Very few people work at jobs that actually make things.     The number one thing sold by far in the US is debt.    So I'm guessing at best robots can replace 8% of us.     Of course once AI is fully in swing within 25 years most all jobs will be obsolete.   

Pretty much all of us are in jobs that owe their existence to government. Without government there would be so much inefficiency due to having the overhead of defending our own property, not having a stable financial system, courts, etc. Government, with all its problems and inefficiencies, is what has allowed an environment where the markets could function in a way for us to go from 100% of people in agriculture to only 2%.

A lot of the jobs you mention aren't making physical objects, but they are facilitating the ability of people to make physical objects. Whether it's marketing those objects (because people wouldn't buy as much of them otherwise and the manufacturing jobs would decrease), providing finance (so the business could afford to start up in the first place), or government ensuring safety and property rights and recourse for damages of defective products (without which fewer people would be willing to buy and sell). There is inefficiency in the system for sure.

And I believe the biggest employment role in government is teachers. At 3.3 million, that's far more than the total number of all federal government employees combined (about 2 million). Those teachers are facilitating the production of all things made in the economy.
http://nces.ed.gov/fastfacts/display.asp?id=28

interesting thoughts,

[arebelspy]

Only about 5 to 8 % of people in the US have "productive" jobs as it is.

Citation?  And what does "productive" mean?

. Produce means to make things or food.

Ah.  Well I think that's way too narrow of a definition.

To say that a teacher's, or a firefighter's, or a librarian's job is non-productive is--while technically accurate under that definition--ridiculous, IMO.

Or, in other words, if you want to use that definition of productive, I'd say it's a good thing that most jobs aren't "productive," and I don't think most jobs should be productive.  If all jobs were just making food or manufacturing things.. eh.  Doesn't say much for that society, IMO.

[matchewed]

Only about 5 to 8 % of people in the US have "productive" jobs as it is.

Citation?  And what does "productive" mean?

. Produce means to make things or food.

Ah.  Well I think that's way too narrow of a definition.

To say that a teacher's, or a firefighter's, or a librarian's job is non-productive is--while technically accurate under that definition--ridiculous, IMO.

Or, in other words, if you want to use that definition of productive, I'd say it's a good thing that most jobs aren't "productive," and I don't think most jobs should be productive.  If all jobs were just making food or manufacturing things.. eh.  Doesn't say much for that society, IMO.

Best Korea takes umbrage at your comments.

Engage saber rattling in 3.. 2.. 1..

[arebelspy]

Only about 5 to 8 % of people in the US have "productive" jobs as it is.

Citation?  And what does "productive" mean?

. Produce means to make things or food.

Ah.  Well I think that's way too narrow of a definition.

To say that a teacher's, or a firefighter's, or a librarian's job is non-productive is--while technically accurate under that definition--ridiculous, IMO.

Or, in other words, if you want to use that definition of productive, I'd say it's a good thing that most jobs aren't "productive," and I don't think most jobs should be productive.  If all jobs were just making food or manufacturing things.. eh.  Doesn't say much for that society, IMO.

Best Korea takes umbrage at your comments.

Engage saber rattling in 3.. 2.. 1..

Ah, but see, if they did that, Dear Leader would have to admit his job is not productive...  ;)

[forummm]

Only about 5 to 8 % of people in the US have "productive" jobs as it is.

Citation?  And what does "productive" mean?

. Produce means to make things or food.    http://www.nam.org/Newsroom/Facts-About-Manufacturing/     this from the national association of manufacturers.   Says 12 million people are employed directly in manufacturing.  That is what?   3.6 % of the population.  Take out those involved in munitions, manufacturing directly for the gov and the fudge factor of the Nam and we are at a realistic 2%.   Throw in farmers and food processor s and we might arrive at 5%.   Interesting enough we produce almost twice as much goods as we did 10 years ago with 1/2 the work force.     Point is in 10 years we may double production again and halve the manufacturing work force.      22% of the us economy is medical (government),  90% of teachers are either direct or indirect government.   All banks are quasi government.   Good or bad that is reality. Farm and ranch families comprise just 2 percent of the U.S. population. More than 21 million American workers (15 percent of the total U.S. workforce) produce, process and sell the nation's food and fiber.
Fast Facts About Agriculture - American Farm Bureau.         

That's just manufacturing in the US. We've outsourced most of ours. If you look globally, the percent involved in agriculture and manufacturing and other things in your narrow definition of "productive" is going to be very high. Mostly because many things in your definiteion of productive are not lucrative, and are easy to outsource.

[Bob W]

Only about 5 to 8 % of people in the US have "productive" jobs as it is.

Citation?  And what does "productive" mean?

. Produce means to make things or food.    http://www.nam.org/Newsroom/Facts-About-Manufacturing/     this from the national association of manufacturers.   Says 12 million people are employed directly in manufacturing.  That is what?   3.6 % of the population.  Take out those involved in munitions, manufacturing directly for the gov and the fudge factor of the Nam and we are at a realistic 2%.   Throw in farmers and food processor s and we might arrive at 5%.   Interesting enough we produce almost twice as much goods as we did 10 years ago with 1/2 the work force.     Point is in 10 years we may double production again and halve the manufacturing work force.      22% of the us economy is medical (government),  90% of teachers are either direct or indirect government.   All banks are quasi government.   Good or bad that is reality. Farm and ranch families comprise just 2 percent of the U.S. population. More than 21 million American workers (15 percent of the total U.S. workforce) produce, process and sell the nation's food and fiber.
Fast Facts About Agriculture - American Farm Bureau.         

That's just manufacturing in the US. We've outsourced most of ours. If you look globally, the percent involved in agriculture and manufacturing and other things in your narrow definition of "productive" is going to be very high. Mostly because many things in your definiteion of productive are not lucrative, and are easy to outsource.

I stand by my definitions of "to produce" things.   The reason I do in this context is we are talking about robotics and not software or other soft products.   Most people would say "we don't make anything anymore in this country"  when that is just a too simple a way to look at it.  Fact is we make more than every in the US but the robots, software and streamlined manufacturing has reduced the need for human labor year after year after year.   So the correct saying in my mind should be "Robots make a heluva lot of stuff in the US."     

[matchewed]

Only about 5 to 8 % of people in the US have "productive" jobs as it is.

Citation?  And what does "productive" mean?

. Produce means to make things or food.    http://www.nam.org/Newsroom/Facts-About-Manufacturing/     this from the national association of manufacturers.   Says 12 million people are employed directly in manufacturing.  That is what?   3.6 % of the population.  Take out those involved in munitions, manufacturing directly for the gov and the fudge factor of the Nam and we are at a realistic 2%.   Throw in farmers and food processor s and we might arrive at 5%.   Interesting enough we produce almost twice as much goods as we did 10 years ago with 1/2 the work force.     Point is in 10 years we may double production again and halve the manufacturing work force.      22% of the us economy is medical (government),  90% of teachers are either direct or indirect government.   All banks are quasi government.   Good or bad that is reality. Farm and ranch families comprise just 2 percent of the U.S. population. More than 21 million American workers (15 percent of the total U.S. workforce) produce, process and sell the nation's food and fiber.
Fast Facts About Agriculture - American Farm Bureau.         

That's just manufacturing in the US. We've outsourced most of ours. If you look globally, the percent involved in agriculture and manufacturing and other things in your narrow definition of "productive" is going to be very high. Mostly because many things in your definiteion of productive are not lucrative, and are easy to outsource.

I stand by my definitions of "to produce" things.   The reason I do in this context is we are talking about robotics and not software or other soft products.   Most people would say "we don't make anything anymore in this country"  when that is just a too simple a way to look at it.  Fact is we make more than every in the US but the robots, software and streamlined manufacturing has reduced the need for human labor year after year after year.   So the correct saying in my mind should be "Robots make a heluva lot of stuff in the US."   

Not sure if anyone would disagree with that last part, just in particular with attempting to narrowly define productive jobs.

[scottish]

But that's where the analogy fails, computers have those things.

That's true, but cows have many things computer don't.    They have senses, and they use them to observe the world and what happens when they interact with it.   They can reproduce, and they have a biological drive to do so.

To be fair, computers do have limited senses, such as vision.  Some of them even have effectors and actuators and so on.   So cows aren't everything.

[matchewed]

But that's where the analogy fails, computers have those things.

That's true, but cows have many things computer don't.    They have senses, and they use them to observe the world and what happens when they interact with it.   They can reproduce, and they have a biological drive to do so.

To be fair, computers do have limited senses, such as vision.  Some of them even have effectors and actuators and so on.   So cows aren't everything.

Many robots have senses, we have software that can reproduce, robots build other robots, all that is lacking is that biological drive. The point of the analogy was that cows as smart as humans couldn't take over the world. But a world with every computer as intelligent as a human? Yeah they'd run the world. I'd bet by that point consciousness wouldn't be far behind.

[Bob W]

I just reread Tim Urban's posts on Waitbutwhy.com regarding the Fermi Paradox and ASI (Artificial Super Intelligence).   

My conclusion was that most readers here will see ASI in their lifetimes much sooner than they think.

My second conclusion was that ASI has been around what we consider the Universe for way longer than we imagine (perhaps trillions of what we call years).  and that the reason we perceive no other intelligent life in the Universe is that the entire universe as we perceive it is merely a holographic projection.  (this is now a widely accepted thought in the physics community).  It appears that Super Intelligence can do what ever it likes with just about anything.  Especially if it has a few billion years to become smarter at a pace of doubling every hour.  (this fits in well with the intelligent design folks)

It is very likely that Super Intelligence can do all sorts of cool stuff we can't imagine like go back in time and manipulate the 10 know dimensions. 

So the likely outcome is that ASI will be that A. We are likely to become nonexistent.   B.  Be treated like a son of the already existing SI entities.  C.  Be sentient whatever that means.

Meanwhile back at the Robot Ranch here.   

Prior to the ASI explosion we most definitely will see computers and robots in virtually every aspect of our daily lives. 

Tim made the point that change is exponential at this point and I can attest to that.   

My dad was born in 1919 --  Cars were already a thing by the time he was a teenager.  So from 1920 to 1960 there wasn't a big leap or a whole lot of change.  (Imagine my dad showing up in 1919 with a cell phone, tablet, flat screen TV, computer, digital video,  digital everything and  the internet.  He would have been God like!)

I was born in 1959  -  We had black and white broadcast TV with poor reception.  We had a party line with a phone with a 5 foot cord.  We did not have microwave ovens.    In 1978 we had 1 computer in our high school with a little green screen that you had to program yourself. 

So from 59 to 78 not much really changed.   Cell phones were still a fantasy.

Thing started to move along a little quicker an as the 80s progressed we got huge brick cell phones  -- 90s internet and real cell phones wow.  2000s smart phones, the cloud,  damn near everything.

The period from 2007 to now has been a little slow and Tim explains this as the S curve way that the exponential shit happens.  So I think in the next 3-5 years we will see a very huge "Next Thing" happen.   

Smart cars and trucks seems like the logical deal.  (Some people say that it may even be made illegal for humans to drive cars)   The implications of 3D printing have yet to be fully revealed but this is some Star Trek shit for real.   You may not print it at home but I can imagine going to your local 3D store and picking up the meat they just printed for you. 

We are about 5 foot steps from all the really cool shit to come but just can't see it or wrap our head around it because we base our estimate of pace of change on what has happened recently.   

Unfortunately,  in the end it appears that ASI will rule the Universe as we know it and we as humans will become irrelevant or extinct as we now exist. 

One thing for sure is that this stuff is damn fascinating for an old guy like me.   I only hope I will make it to the promised 1945 date of ASI. 

[arebelspy]

and that the reason we perceive no other intelligent life in the Universe is that the entire universe as we perceive it is merely a holographic projection.  (this is now a widely accepted thought in the physics community).

[Citation needed.]  (Not for the theory, but for its "wide" acceptance.)

My second conclusion was that ASI has been around what we consider the Universe for way longer than we imagine (perhaps trillions of what we call years).
...
It appears that Super Intelligence can do what ever it likes with just about anything.  Especially if it has a few billion years to become smarter at a pace of doubling every hour.  (this fits in well with the intelligent design folks)

It is very likely that Super Intelligence can do all sorts of cool stuff we can't imagine like go back in time and manipulate the 10 know dimensions. 

So the likely outcome is that ASI will be that A. We are likely to become nonexistent.

This seems contradictory.  If it's already existed for billions of years, why would it decide now to wipe us out?

Either ASI already exists, as you claim, and it's okay with us, or it doesn't yet exist, but when we create it it will not be okay with us and wipe us out, as you also claim.  Which is it?

I only hope I will make it to the promised 1945 date of ASI.

Who's promising that?

Cause AFAIK, ASI wasn't invented around the end of WWII.  Even if I assume that's a typo and you meant 2045, I still don't know who's promising anything around that timeframe...

[Bob W]

and that the reason we perceive no other intelligent life in the Universe is that the entire universe as we perceive it is merely a holographic projection.  (this is now a widely accepted thought in the physics community).

[Citation needed.]  (Not for the theory, but for its "wide" acceptance.)

My second conclusion was that ASI has been around what we consider the Universe for way longer than we imagine (perhaps trillions of what we call years).
...
It appears that Super Intelligence can do what ever it likes with just about anything.  Especially if it has a few billion years to become smarter at a pace of doubling every hour.  (this fits in well with the intelligent design folks)

It is very likely that Super Intelligence can do all sorts of cool stuff we can't imagine like go back in time and manipulate the 10 know dimensions. 

So the likely outcome is that ASI will be that A. We are likely to become nonexistent.

This seems contradictory.  If it's already existed for billions of years, why would it decide now to wipe us out?

Either ASI already exists, as you claim, and it's okay with us, or it doesn't yet exist, but when we create it it will not be okay with us and wipe us out, as you also claim.  Which is it?

I only hope I will make it to the promised 1945 date of ASI.

Who's promising that?

Cause AFAIK, ASI wasn't invented around the end of WWII.  Even if I assume that's a typo and you meant 2045, I still don't know who's promising anything around that timeframe...

Obvious typo on the 45 deal.  2045 is the date of median acceptance for AGI with ASI a few moments after that in some scenarios.   Personally I feel that the date is much sooner than that.   If we are 1 percent to AGI now, then the exponential S curve theory leads me to believe 15 years is the longest out.   Yeah, this will probably fuck up a lot of people's retirement plans.

One assumption is that the US and China are behind the curtains pumping 10s of billions into this.   If they aren't currently, one would assume that the Pentagon will be getting on this soon.   At very least the NSA has all the information ever produced on this and is keeping a very close eye on it.   It may be the NSA's primary focus now as it is the most likely big threat on the horizon.   

 You'll have to do your own research on the holographic universe theory.   (I don't make this shit up as you know)

To simplify why ASI has probably been around for billions or trillions or more years.

The Fermi paradox clearly states we can't explain why we can't find any intelligent technologically advanced life in the universe?   The probability of us being the only ones ever is as close to zero as one could get,  given the size and age of the known Universe.   

So therefore the opposite must be true --- that technologically advanced life must have existed before us.    (I know we like to think we're special, but that probably isn't the case)

One would also surmise that exponentially advances in technology apply to other places besides the earth. 

 Therefore, even if an entity arrived at ASI just 20 years ahead of us anywhere on the quadrillion solar systems,  it would now be at the point where it was trillions and trillions of times more intelligent. 

Once you do your research on the holographic universe explanation,  you will see that what we see as a reality is merely a holographic projection.  So it must be projected from somewhere and that somewhere leads us to the SI entity whose existence is mathematically as close to certain as you can get.   In the parallel universe theory it is a certainty by definition. 

In our local time horizon when ASI pops it will mean either the end of humans or the end of our relevance.   The idea of us melding our brains into SI is certainly a possibility and may in fact happen.   That would allow SI to develop with a human conscious, soul, or be sentient.  One would assume that most  scientist would want to meld the SI with people who are generally "good" and have empathy.   Of course they could botch it as well. 

Will that entity ever reach the level where it can create, expand,  constrict and travel back and fourth on the time dimension?  Probably so.   Will that entity reach a level where it can function on the existing known 10 dimensions.   Probably so.   Will it be able to create new dimensions as a  fun game for a Saturday afternoon.  Probably so.  Will it be able to create what we perceive as an entire universe in its spare time?  Yep,  the exponential theory would lead us to believe that and that fits right in with the holographic universe thinking.

Probably our ASI child will meld into the existing SI at some point within a very short time.

You'll have to excuse me for a moment --- a bit of my brain just melted and dripped out of my ear. 

I hope Tim writes about a few things in the future 1.  Holographic Universe (which is even more interesting than AI)  2. Dark Matter  3. Dark Energy  4. The 10 living people whose father's "fought" in the civil war.   5. Do we actually "exist"

What a crazy interesting universe!

(I also wish there was a forum on Tim's site rather that that crazy 3000 comments per post thing)

[matchewed]

A) He's not saying it's not a theory, he's saying he doubts the wide acceptance and is asking for some citation that it is widely accepted. I would also like to see that as there are several physicists I've seen interviewed who do not buy into it, nor the 10 dimensions you mentioned, there are probably some theoretical physicists who believe so but so far we have no experimental proof to back this up. This lack of proof would make me believe that wide acceptance would be a stretch.

B) The Fermi Paradox does not explain why we don't see evidence of life beyond our earth. It only asks why, given the size and age of the universe, do we not see any evidence of life beyond earth. There are several hypothesis of the explanation, one being that they (other civilizations) tend to have a technological singularity. This tends to fragment into many other smaller interpretations as to what that technological singularity is, and one small narrow possibility is what you posted.

None of that is what I would call widely accepted. And there are other plausible explanations for the Fermi Paradox that do not invoke AI at all. 

[sol]

I have no problem believing in a universe in which non-organic intelligences inhabit the universe.  Organic life is just too fragile to exist long term in a universe like ours. 

The easiest explanation for dark matter is that most of the mass in the universe exists in luminescent stars that currently have all of their energy output that earth can see harvested.  In that view, the tiny fraction of mass that constitutes the billions of stars that we can see is just the leftover table scraps that synthetic life has yet to fully colonize.

I expect that such a universe would eventually have each galaxy owned by a single synthetic intelligence or society, but that the distances between galaxies probably inhibit intergalactic conflict.

Human society is just a speed bump.  If we ever do give rise to a superintelligence of our own, it's likely to be rapidly outnumbered or extinguished by the nearest existing superintelligence.  I would expect such an entity would view human life as an interesting oddity, like we view deep sea hydrothermal vent communities, alien aberrations unlike everything else known. 

The universe is approximately sixty five thousand times older than all human life. Whole generations of stars like our sun have come and gone before our solar system was even formed, and our little planet has only had intelligent life for a tiny fraction of its brief history. 

The universe looks capable of continuing on in much its present form for much longer than the 13 billion years it has already existed.  It took less than a billion years to go from single celled organisms to spacefaring intelligence.  What will it look like in another thirteen billion years?

[Albert]

It all sounds too much like a religion to me believing in this or believing in that... One of the key properties of a truly scientific theory that there should be a way to prove it wrong. That is if you propose a theory A there should be an experiment (preferably real, but could also be something we can't do technologically yet) which by giving result B would prove that theory A is invalid.

[matchewed]

It all sounds too much like a religion to me believing in this or believing in that... One of the key properties of a truly scientific theory that there should be a way to prove it wrong. That is if you propose a theory A there should be an experiment (preferably real, but could also be something we can't do technologically yet) which by giving result B would prove that theory A is invalid.

Well yeah, theory is being used two different ways here. The theory of evolution is used in a way which means we have a preponderance of evidence showing it happens. The idea that physicists have a theory that the world is just a hologram, is really a hypothesis one which we have no real method of testing right now.

I hate the phrase but this truly is a semantic issue rather than an actual one.

[Albert]

If you can't test your hypothesis even theoretically then it's not a science anymore. At best it's a philosophy of the sort ancient Greeks engaged in with some chance of making a lucky guess (Democritus and his atoms, for example).

[matchewed]

If you can't test your hypothesis even theoretically then it's not a science anymore. At best it's a philosophy of the sort ancient Greeks engaged in with some chance of making a lucky guess (Democritus and his atoms, for example).

I'm not sure if I can agree 100% with that. You need the hypothesis. And just because we can't test today doesn't mean we can't test tomorrow. Theoretical physics is what led us to many major discoveries. Einstein was a theoretical physicist. I doubt anyone could say he wasn't a scientist.

[Albert]

I'm not sure if I can agree 100% with that. You need the hypothesis. And just because we can't test today doesn't mean we can't test tomorrow. Theoretical physics is what led us to many major discoveries. Einstein was a theoretical physicist. I doubt anyone could say he wasn't a scientist.

I think you didn't understand me correctly. Theoretical physics is a proper science because its theories are in principle testable. Pretty much everything Einstein proposed has been later verified, not by him and in some cases long after he died but that doesn't matter.  When you propose something new whether it be in particle physics or in psychology of monkeys you also need to come up with some experiment which could in principle either verify or invalidate it. As I wrote before the best is if you could do it yourself or base it on earlier work, but if that is not possible and it wasn't for some of Einstein's stuff a thought experiment would suffice. Of course this also leaves your theory more open to later invalidation. The problem arises if your theory is so esoteric (as discussed upthread) that you can't come up even with a plausible thought experiment. Statistics and probability in this case is not good enough.

[gimp]

All I know is that I want to be one of the people producing tech. Until robots can design better robots, I'll make a living :)

[EscapeVelocity2020]

We should also be more clear about what we are talking about.  Just because computers will be 'exponentially more powerful' in the future (to whatever extent you wish to believe), it does not make them any more 'intelligent'.  More capable, but not even incrementally more capable of what people are after when they imagine machine intelligence (e.g. solving problems and self-learning).

On the other hand, from the Nature article, even with the amazing networked parallel processing of Deep Q, the real breakthrough was implementation of more simplistic algorithms to simulate 'motivation'.  Basically, the computers were seeking a goal of the highest end state (score) with just the right amount of learning carried over from past trials.  But it is this 'motivation' component that is still elusive. 

I have been thinking about it, and it really is a hard thing to define or quantify (let alone express in mathematical terms).  For instance, many of the people on this forum are 'motivated' to work and save and forego the immediate fruits of their labor in exchange for the promise of earlier retirement.  Kinda makes sense, but if you tried to express that in simple, inflexible language to a computer, it would sound like you are doing what you don't want to be doing now so that you don't have to do it any more of it than you have to do....  So how much do you do?  Solving a problem like this is (seemingly) possible for all of us here, but I can't even imagine how an AI would go about it...  It can do the simulations (ANI) faster and faster each year (shave off another fraction of a second of all those complex calculations, woo-hoo!), but without understanding how to program 'motivation', it will never reach AGI.

[EscapeVelocity2020]

Thought this was interesting: http://time.com/3860218/robot-salad-youtube/

Using pattern recognition software designed by the interdisciplinary robotics team at the College Park campus, Julia the robot watched YouTube videos of people making salads to learn the steps, from cutting vegetables to tossing the ingredients and even pouring the salad dressing at the end.

[Insanity]

amazon released a Machine Learning service as part of its AWS suite.   

More systems will be smart and will learn.  All learning is simply gaining rules and evaluating prioritization a based on another set of rules. 

Motivation, emotion, they are all based on rules.  That is it.  That doesn't mean it is easy. But it is pretty simple.

[scottish]

Making salad would be an impressive demo.    If they actually showed it.    There's a more complete video here:  https://www.youtube.com/watch?v=IFKtL1BWQ9Q   It's not clear how much autonomous learning the robot is doing, the video suggests that the researchers are segmenting the video into the different salad making tasks and using them to train the robot.  Prof Alimonos' work certainly appears impressive though.   Robotics are great!

[Bob W]

This just in ---

http://www.salon.com/2015/05/10/robots_are_coming_for_your_job_amazon_mcdonalds_and_the_next_wave_of_dangerous_capitalist_disruption/

It appears that McDonalds will be testing an almost fully robotic restaurant with a small crew to assist the robots.   Makes sense.   What next drones delivering your pizzas?!

[tomsang]

This just in ---

http://www.salon.com/2015/05/10/robots_are_coming_for_your_job_amazon_mcdonalds_and_the_next_wave_of_dangerous_capitalist_disruption/

It appears that McDonalds will be testing an almost fully robotic restaurant with a small crew to assist the robots.   Makes sense.   What next drones delivering your pizzas?!

Nice article.  Comments are interesting as well.

[forummm]

Next they're going to create bots that post on Internet forums. Then what would I do for money?

[arebelspy]

Next they're going to create bots that post on Internet forums.

[brooklynguy]

Next they're going to create bots that post on Internet forums.

As I've said before, I think they already have.

[arebelspy]

Next they're going to create bots that post on Internet forums.

As I've said before, I think they already have.

Tell me more about As I've said before.

[2lazy2retire]

Here, so people don't have to click on a link:






People keep posting about how we aren't that close yet.
The thing about exponential growth is that it doesn't look close even one year before you get there.

Remember the riddle from elementary school? 
"If a pond lily doubles everyday and it takes 30 days to completely cover a pond, on what day will the pond be 1/2 covered?[/size] [/color][/size]The answer is day 29.

We are here on day 26 saying "hey, the pond is barely 1/10th filled, we aren't close yet" even though we are only 4 days from total saturation.[/color]

This assume exponential trends will continue to infinity--something that doesn't happen in nature. Something always gets in the way. Maybe computational advances will continue at this rate for long enough to create the advances needed for AI exceeding human intelligence. Maybe not. There's a possibility, but don't take it as gospel truth.

There's been  a bit of that exponential stuff going on - 5Meg Hard Drive  1956

[brooklynguy]

Tell me more about As I've said before.

Not sure I follow...

If your point was I should stop recycling jokes I've already made, point taken.

If your point was I should stop speculating about that which I speculated, given that you are the most likely candidate to whom my statement applies, point taken :)

[forummm]

Oh no! They're here!