The Revolution in Motion

[TimG]

Not self-sustaining, we need frequent food, water, and oxygen intake to fuel us.

All items which can be obtained fairly easily from the natural environment. Silicon based robots need electricity which is much harder to obtain and store. To match human portability robots would likely have to switch to biological power sources.

They could also become so intelligent and powerful that they could invent new ways to power themselves.

If they did they would most likely conclude that 4 billion years of biological evolution has produced the best trade off between compactness, portability and computing power possible given the laws of physics. Many advances today come from mimicking what nature has already figured out.

Consider how fast computer technology has evolved in the last 30 years....and that's by a human mind that hasn't become any more intelligent.

Yet despite those advances an AI can be rendered useless by unplugging it. Batteries do not follow moore's law and there is no reason to believe that they can be improved enough to match the portability and flexibility of a biological system. Edited August 4, 2015 by TimG

[Moonlight Graham]

All items which can be obtained fairly easily from the natural environment. Silicon based robots need electricity which is much harder to obtain and store. To match human portability robots would likely have to switch to biological power sources.

If true, then so be it. If silicon based machines/AI compute than silicon, metals, and plastic aren't the best materials to make it the most efficient and powerful system and it needs to redesign itself, even using organic systems, then so be it.

If they did they would most likely conclude that 4 billion years of biological evolution has produced the best trade off between compactness, portability and computing power possible given the laws of physics. Many advances today come from mimicking what nature has already figured out.

Yes, many things nature has evolved is incredible and can be mimicked by AI. Humans are amazing, but also far from the perfect. We are physically weak compared to machines (ie: bulldozers, power tools, guns) or even other animals, unable to compute numbers nearly as fast as even moderately capable computers, and unable to survive extreme environments on our own without technologies. AI could create a being that's far more intelligent, physically stronger, and more adaptable than humans. If this new being incorporates biological elements, so be it. My point is, humans can be improved upon.

Yet despite those advances an AI can be rendered useless by unplugging it. Batteries do not follow moore's law and there is no reason to believe that they can be improved enough to match the portability and flexibility of a biological system.

My point is that AI will be able to evolve at an incredibly fast rate, without the need to rely on infrequent reproduction of offspring and random genetic mutations in order to do it. Maybe humanity will have this ability one day too based on technology they create (ie: test-tube babies and genetic manipulations), but they'll never be able to match the speed of AI evolution...and a human with ie: 2x the intelligence and 4 arms would be a being that would cease to be human. IMO humans will be the creators of their successors.

Edited August 4, 2015 by Moonlight Graham

[TimG]

If true, then so be it. If silicon based machines/AI compute than silicon, metals, and plastic aren't the best materials to make it the most efficient and powerful system and it needs to redesign itself, even using organic systems, then so be it.

There is no *best*. There are only solutions that best meet design requirements. Humans are generalists - we can do an extremely wide range of tasks but because we are optimized for flexibility we cannot compete with a machine designed to a single task. We don't really need generalist AIs. We need AIs designed to solve specific problems, however, as an academic exercise I am sure someone will try to replicate a human's generalist capabilities. That said, it my belief that such efforts will only reach true equality with humans when the machines are biological constructs. AIs that need to be plugged in or cannot move are not equal to humans no matter how intelligent they are.

AI could create a being that's far more intelligent, physically stronger, and more adaptable than humans.

AIs will not be more 'adaptable' unless they become biological constructs which will also mean they will have many of limitations that humans have because of the limitations of biological constructs.

My point is that AI will be able to evolve at an incredibly fast rate, without the need to rely on infrequent reproduction of offspring and random genetic mutations in order to do it.

All this means is machines will remain hyper specialized. i.e. an AI that drives a car does not need to be able to compose music. It just needs to drive the car. The random aspect of evolution that gives humans abilities that were not intended or required is what makes humans human. This is why I don't think we will ever see a 'robot civilization'. Edited August 4, 2015 by TimG

[Moonlight Graham]

That said, it my belief that such efforts will only reach true equality with humans when the machines are biological constructs. AIs that need to be plugged in or cannot move are not equal to humans no matter how intelligent they are.

Why does AI need to be plugged in? It could run on gas or other fossil fuels, solar, wind, nuclear etc., or anything else that produces energy/electricity. Battery technology will also continue to improve, especially once super-intelligent AI is able to redesign it, while humans running on food and fat stores will not. Humans aren't very portable WITHOUT machines. We can walk and run an ride a horse, that's about it. No competition for a plane or car or space shuttle.

Humans, to achieve anything that the advances the industrial revolution and electricity have provided us, must use machines. Why not cut out the middle-man?

AIs will not be more 'adaptable' unless they become biological constructs which will also mean they will have many of limitations that humans have because of the limitations of biological constructs.

I guess we'll agree to disagree on that one.

All this means is machines will remain hyper specialized. i.e. an AI that drives a car does not need to be able to compose music. It just needs to drive the car. The random aspect of evolution that gives humans abilities that were not intended or required is what makes humans human. This is why I don't think we will ever see a 'robot civilization'.

No need to be specialized. I could imagine an AI hub mind that controlled everything over a network, where all processing power from all AI is linked and integrated together into its total processing power and total intelligence. This AI hub network would work in unison together to achieve whatever functions/goals it deems desirable. A car wouldn't need to compose music if it's unneeded, but if it is then it could be adapted very quickly to compose music.

[TimG]

Why does AI need to be plugged in? It could run on gas or other fossil fuels, solar, wind, nuclear etc., or anything else that produces energy/electricity.

IOW, it needs to be plugged in. Battery technology will likely NOT improve much because there are basic rules of physics that limit the capacity/charging time of batteries. What will happen is power requirements for AIs will drop which lessen the need for power but there will always be a trade off: more powerful AIs will require a large power source which limits their portability while less powerful AIs will be more portable but less capable. Engineering trade offs are a fact of life an it is naive to assume some technical nirvana exists where no trade offs are required.

Humans aren't very portable WITHOUT machines. We can walk and run an ride a horse, that's about it. No competition for a plane or car or space shuttle.

A plane, car or shuttle are piles of inert junk without the infrastructure necessary to provide fuel and replacement parts. They can also only do the task that they were designed for. i.e. plane is not useful for getting to costco, a car won't help if you want to cross an ocean. The space shuttle is will get you to low earth orbit and no where else. A human can walk, ride a horse, build a car, a boat or a plane. That is the difference between a specialist and a generalist.

No need to be specialized.

Of course there is because general purpose AIs will be, by definition, less capable than specialized AIs. The cost gap between a specialist and a generalist will ensure that no one will have any incentive to create a generalist AI for something other than an academic exercise. This means the emergence of a robot society independent of human is extremely implausible. I can see a future filled with specialized thinking machines that design to solve specific tasks for humans but generalist humans will be at the core. Edited August 5, 2015 by TimG

[Bonam]

If they did they would most likely conclude that 4 billion years of biological evolution has produced the best trade off between compactness, portability and computing power possible given the laws of physics. Many advances today come from mimicking what nature has already figured out.

4 billion years of biological evolution has by no means produced a perfect product that achieves the maximum possible performance optimization allowable by the laws of physics. Biological evolution is random, not directed, and does not tend towards perfection but only towards the ability to survive within a given environment. It is most definitely possible to create machines that outperform living creatures, whether specialists or generalists. Some advances come from mimicking nature, but many others do not.

Edited August 5, 2015 by Bonam

[Bonam]

IOW, it needs to be plugged in. Battery technology will likely NOT improve much because there are basic rules of physics that limit the capacity/charging time of batteries. What will happen is power requirements for AIs will drop which lessen the need for power but there will always be a trade off: more powerful AIs will require a large power source which limits their portability while less powerful AIs will be more portable but less capable. Engineering trade offs are a fact of life an it is naive to assume some technical nirvana exists where no trade offs are required.

A battery in this context is any portable energy source. Fats and sugars are by no means the highest energy density means of storing energy. Neither are rechargeable chemical batteries. Far higher energy densities are possible with other means of storage, including radioisotopes, meta-stable states of matter, and antimatter. While trade offs are required in engineering, current battery storage technologies are many orders of magnitude away from the the limits of what is physically possible.

Anyway, the overall point you are trying to make, that humans will always have a role to play, is likely true, even if only because humans themselves will not remain as they are now but will enhance themselves through gene manipulation, cybernetics, and nanotechnology. That said, you make very declarative statements about the limitations of technology that have no basis in fact, instead they are based on the assumption that the technology that is commercially available in 2015 is the only technology that will ever or can ever exist. Only in a very few fields of technology is there even any thought that we are pushing close to what is allowable by the laws of physics, in most others we are nowhere close, and the advances that can be reaped over the decades to come will likely prove almost everyone trying to predict the future to be completely wrong.

Edited August 5, 2015 by Bonam

[TimG]

Far higher energy densities are possible with other means of storage, including radioisotopes, meta-stable states of matter, and antimatter. While trade offs are required in engineering, current battery storage technologies are many orders of magnitude away from the the limits of what is physically possible.

My point is the trade offs will always exist. i.e. energy storage mechanisms that may be less efficient but require less supporting technical infrastructure will be better for some applications. i.e. someone may decide that having a robot with 4 times the strength of a human serves a purpose but the price would be higher energy requirements which would make less useful than a human in some situations.

That said, you make very declarative statements about the limitations of technology that have no basis in fact, instead they are based on the assumption that the technology that is commercially available in 2015 is the only technology that will ever or can ever exist.

If you want to assume that everything we know about the universe could turn out to be wrong then yes it goes without say that anything is possible. However, we do know a fair amount about the universe and most of this knowledge will be still true in the future even as existing theories need to be refined. For example, it is reasonable to say that bi-directional time travel is extremely implausible (i.e. going to the past or future and then return to the present). It is also reasonable extrapolate based on our knowledge now and identify areas where progress is plausible and where it is less plausible. For example, it is plausible that humans will be able to build an organism from scratch in the future and learn to modify our own genes to enhance or repair our bodies. It is also plausible that the energy required per unit of computing power will continue to decrease and that economically viable fusion power will reduce the cost of energy production to close to nothing. It is implausible to think that the process of using chemical bonds to store energy will change much or the infrastructure required to distribute electricity will get less cumbersome. Edited August 5, 2015 by TimG

[Bonam]

If you want to assume that everything we know about the universe could turn out to be wrong then yes it goes without say that anything is possible. However, we do know a fair amount about the universe and most of this knowledge will be still true in the future even as existing theories need to be refined. For example, it is reasonable to say that bi-directional time travel is extremely implausible (i.e. going to the past or future and then return to the present). It is also reasonable extrapolate based on our knowledge now and identify areas where progress is plausible and where it is less plausible. For example, it is plausible that humans will be able to build an organism from scratch in the future and learn to modify our own genes to enhance or repair our bodies. It is also plausible that the energy required per unit of computing power will continue to decrease and that economically viable fusion power will reduce the cost of energy to close to nothing. It is implausible to think that the process of using chemical bonds to store energy will change much or the infrastructure required to distribute electricity will get less cumbersome.

Lithium Ion batteries store about 1 MJ/kg.

The highest energy density chemical bonds are about 2 orders of magnitude higher than that (~50-100 MJ/kg). We already harness some of these for portable energy sources.

Nuclear energy storage can get you yet another 4 orders of magnitude higher, to about 1 TJ/kg (1 million times denser than the battery). There are already nuclear batteries in commercial use, based on the beta decay of tritium, as well as RTG power sources for spacecraft and other exotic applications, based on the alpha decay of plutonium 238.

Antimatter energy storage can get you yet another 5 orders of magnitude higher, to about 100 PJ/kg. That's 100 billion times more energy density than a current battery.

All these technologies are well within the known laws of physics.

We have as far to go in energy storage technology (a factor of 10^11) as we have gone in computation technology in the last 100 years. And this progress in energy technologies will likely happen over the next 100 years.

Edited August 5, 2015 by Bonam

[TimG]

The highest energy density chemical bonds are about 2 orders of magnitude higher than that (~50-100 MJ/kg). We already harness some of these for portable energy sources.

Energy density is not the main issue - it is the process of charging and discharging. Fats and sugars have low energy density but they can be stored outside the body and quickly consumed when needed. Electrical batteries need to be connected to a power source for charging and the higher the density the longer it takes to charge. It is the charging time that will not change because it governed by E = VIt and I can't see that changing.

Esoteric batteries like a nuclear or anti-matter batteries may have high energy density but issues like containing radiation/preventing catastrophic accidents will make them unusable in many applications which I why I talked about engineering trade offs. i.e. just because a technology exist does not mean it can be used in any application. I am fairly confident that a energy storage mechanism based on organics will turn out to be the most versatile and practical for mass applications largely because that is the mechanism chosen by evolution.

Edited August 5, 2015 by TimG

[GostHacked]

BS. Some people claim that doom is on the horizon just like some people claim the rapture is just around the corner. Actual science says that CO2 is a concern and the impacts could range from minor to locally disruptive. No plausible scenario suggests that humans will not be able to adapt to CO2 induced changes.

But that is my point. Humans adapt. Your assertion that our ability to adapt will suddenly disappear has no logical foundation.

What do you do when there is nothing left to adapt to?

[Moonlight Graham]

IOW, it needs to be plugged in.

Planes, cars, shuttles, satellites, space stations, probes flying past pluto need to be plugged in?

[Moonlight Graham]

That said, you make very declarative statements about the limitations of technology that have no basis in fact, instead they are based on the assumption that the technology that is commercially available in 2015 is the only technology that will ever or can ever exist.

Exactly. TimG is displaying an incredible lack of imagination. Likely at first super-intelligent self-evolving AI will need human guidance, but I see no reason why they can't eventually evolve themselves to become self-sustaining beings without any human help.

[TimG]

Exactly. TimG is displaying an incredible lack of imagination. Likely at first super-intelligent self-evolving AI will need human guidance, but I see no reason why they can't eventually evolve themselves to become self-sustaining beings without any human help.

There is imagination and there are implausible fantasies. The issues with energy storage are real and fundamental to universe as we understand it today. A computer AI is not synonymous with god-hood and an AI will have to live the same universe that we live with and will need to make engineering trade offs. This put constraints on any future society and it likely means that biological constructs (artificial or natural) will be a big part of any future because biological constructs have already figured out a way to put a powerful generalist AI into a compact and adaptable frame. Edited August 5, 2015 by TimG

[Moonlight Graham]

There is imagination and there are implausible fantasies. The issues with energy storage are real and fundamental to universe as we understand it today. A computer AI is not synonymous with god-hood and an AI will have to live the same universe that we live with and will need to make engineering trade offs. This put constraints on any future society and it likely means that biological constructs (artificial or natural) will be a big part of any future because biological constructs have already figured out a way to put a powerful generalist AI into a compact and adaptable frame.

A being that is able to perpetually evolve itself will be able to eventually create anything possible that's within the laws of physics and the within the limits of the natural resources it will have at its disposal. Consider the unimaginable and incredible possibilities. Whatever form it takes is impossible to guess, but I'm confident that such a being will be far superior to humans in virtually every way.

[WIP]

4 billion years of biological evolution has by no means produced a perfect product that achieves the maximum possible performance optimization allowable by the laws of physics. Biological evolution is random, not directed, and does not tend towards perfection but only towards the ability to survive within a given environment. It is most definitely possible to create machines that outperform living creatures, whether specialists or generalists. Some advances come from mimicking nature, but many others do not.

That's right, and one of the big reasons why some philosophers who are AI skeptics, believe we many actually be no closer to creating a truly conscious artificial intelligence than we were 20 or 30 years ago. Speed and processing power can't make up for the fact that neurons are more complicated and have more options than an integrated circuit.

[WIP]

My point was the phrase is meaningless because people use it as a catch all to criticize any human activity they don't like. Humans are very good at adapting to change and it is a mistake to use linear extrapolation and assume a past behavior will continue indefinitely into the future.

We used to be good at adapting to change! In the time before 10,000 years ago, when everyone was a hunter/gatherer and was not able to exercise much control over available food supplies, other than look for other things to eat in fresh locations, then we had to be very good at adapting to changing weather and climate conditions...though genetic evidence of a population bottleneck about 70,000 years ago which reduced the likely total human population to about two to three thousand, indicates that the human race almost went extinct because of the only supervolcano event in the history of modern humans.

What we are now, are mostly civilized drones who have no idea what to do when the power goes out! Are we really that resourceful today? Is our globalized civilization...dependent on extended trade links, with few locally sustainable communities, a resourceful global civilization that will survive a changing climate, unprecedented population levels, declining resources? We will all find out! But my suspicions are that any survivors will be the kinds of people who are truly adaptable and resourceful, and lucky enough to live in the right places.