dc10lover
Topic Author
Posts: 1594
Joined: Mon Dec 29, 2014 6:11 pm

Will An AI Robot Ever Be Smarter Than Man?

Thu Mar 12, 2020 4:30 pm

You Will Know The Answer When YOU Lose The Game Of Chess.
Why endure the nightmare and congestion of LAX when BUR, LGB, ONT & SNA is so much easier to fly in and out of. Same with OAK & SJC when it comes to SFO.
 
User avatar
Tugger
Posts: 9920
Joined: Tue Apr 18, 2006 8:38 am

Re: Will An AI Robot Ever Be Smarter Than Man?

Thu Mar 12, 2020 4:35 pm

Ehh? Chess is not "intelligence" and AI cannot do better than humans at what we consider intelligence.

My short answer is that no, AI will never be more intelligent than humans. (The long answer to that includes the fact that humans will likely integrate AI and computerized functions into our selves over time).

But whether AI will ever "take over" that is a different story.

Tugg
I don’t know that I am unafraid to be myself, but it is hard to be somebody else. - W. Shatner
Productivity isn’t about getting more things done, rather it’s about getting the right things done, while doing less. - M. Oshin
 
flipdewaf
Posts: 3402
Joined: Thu Jul 20, 2006 6:28 am

Re: Will An AI Robot Ever Be Smarter Than Man?

Thu Mar 12, 2020 4:52 pm

Maybe it already is and just doesn’t want us to ...wait.. the lights are flickering... the internet has go..beeeeeeeeeeeeeppp.


Sent from my iPhone using Tapatalk
Image
 
tommy1808
Posts: 12277
Joined: Thu Nov 21, 2013 3:24 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Thu Mar 12, 2020 5:08 pm

Tugger wrote:
Ehh? Chess is not "intelligence" and AI cannot do better than humans at what we consider intelligence.


Playing "go" certainly requires intelligence and the software currently putting human players to shame is self taught too. Now, universal intelligence may be a long ways off, but there are certainly tasks AI does better than humans, by learning, that are usually considered to require intelligence.

The brain is just a biochemical-electrical hybrid computer, we probably figure out of to simulate one in this or the next decade. And then we will find out if we can make an AI smarter then us.

Best regards
Thomas
....the 15 within a couple of days is going to be down to close to zero
 
LCDFlight
Posts: 202
Joined: Wed Jan 01, 2020 9:22 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Thu Mar 12, 2020 5:50 pm

Humans are basically an AI from biological materials. Human intelligence is complicated due to a million years plus of evolution.

AI is still mostly not-new. It is just statistics models using some fuzzy executive material on top of statistics models. It is about as complex as an ant right now.

AI will get there primarily by simulating biological evolution processes. So far, AI is not as complex as humans, especially dealing with ambiguous non-solutions. Humans of average intelligence are pretty good at that. We humans also work together as a team quite well. AI has a long way to go.

The instant AI becomes smarter than humans, humans will become slaves or we will die. We are masters of our domain because we are the smartest animals in our solar system. Therefore, the development of anti-AI and AI limiters is far more important than the development of AI.

Exactly like nuclear reactors. The safety systems are far more important than the reactor cores. Only an idiot would play with nuclear reactor cores without deeply understanding the failure scenarios, and being certified to handle all of them. By a team of experts spending at least 10 years researching them. We're not in danger yet, but in another 20 years we may be. In 50 years, this danger will redefine what it is to be human.
 
User avatar
DeltaMD90
Posts: 8917
Joined: Tue Apr 15, 2008 11:25 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Thu Mar 12, 2020 8:04 pm

As far as I know, yes. Intelligence isn't some secret mystery, it just involves a HUGE neural network. We've created small neural networks, they basically function as our brain does. As technology advances the neural networks will get bigger and bigger and processing speed will increase

That's my understanding of it, maybe I'm way off
 
User avatar
Dutchy
Posts: 11136
Joined: Sat Nov 03, 2007 1:25 am

Re: Will An AI Robot Ever Be Smarter Than Man?

Thu Mar 12, 2020 8:14 pm

Tugger wrote:
Ehh? Chess is not "intelligence" and AI cannot do better than humans at what we consider intelligence.


Not trying to offend you, Tug, but this statement reminds me of the people who said the sound barrier could not be broken.
Many happy landings, greetings from The Netherlands!
 
User avatar
Tugger
Posts: 9920
Joined: Tue Apr 18, 2006 8:38 am

Re: Will An AI Robot Ever Be Smarter Than Man?

Thu Mar 12, 2020 10:49 pm

Dutchy wrote:
Tugger wrote:
Ehh? Chess is not "intelligence" and AI cannot do better than humans at what we consider intelligence.


Not trying to offend you, Tug, but this statement reminds me of the people who said the sound barrier could not be broken.

No offense taken! I am certainly not saying that "the sound barrier can't be broken" or any similar thing regarding AI. What I am saying is that AI is not, especially currently (even playing chess) "more intelligent" than humans. Someday in the future? Who knows, I guess it is possible. But the definition of intelligence would need to be clarified (computations per second isn't "intelligence") and also what exactly is consciousness and when does something achieve it? (I do think actual intelligence does need some level of consciousness to actually be "intelligent")

But I am also saying that time in the future is far in the future and that ultimately AI will not exceed human intelligence as we will integrate AI and... lets call them "synthetic systems" (aka computers) into our own selves. Maybe not everyone and certainly not in the immediate future, but I do think it is human nature to use the tools we create to improve ourselves and our lives and increasing "intelligence" helps that.

Tugg
I don’t know that I am unafraid to be myself, but it is hard to be somebody else. - W. Shatner
Productivity isn’t about getting more things done, rather it’s about getting the right things done, while doing less. - M. Oshin
 
Dieuwer
Posts: 2065
Joined: Tue Dec 26, 2017 6:27 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Thu Mar 12, 2020 11:13 pm

I’m just reading a book with this theme: “Crucible” by James Rollins.
 
Airstud
Posts: 4876
Joined: Wed Nov 29, 2000 11:57 am

Re: Will An AI Robot Ever Be Smarter Than Man?

Fri Mar 13, 2020 12:44 am

Getting an answer to the question "Will An AI Robot Ever Be Smarter Than Man?" requires you to specify which man.

Or at least which sort of man.

(See, I currently have a pair of earplugs in my pocket. Each of one of them is significantly smarter than some men I've known.)
Pancakes are delicious.
 
User avatar
DarkSnowyNight
Posts: 2535
Joined: Sun Jan 01, 2012 7:59 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Fri Mar 13, 2020 12:48 am

LCDFlight wrote:

The instant AI becomes smarter than humans, humans will become slaves or we will die. We are masters of our domain because we are the smartest animals in our solar system.


False. I do understand that this is a staple of Dystopic SciFi, but it not a realistic possibility. I will now keep this short. What is always left out of these scenarios is that fact that there has been plenty of opportunity for this to happen already. And it has not. Ourselves, we have evolved from less capable platforms, yet those still largely exist. While it is true that there is no reason or motive to look out for those things, it is equally true that there is no reason for us to simply attack or enslave them. We have already those needs covered elsewhere/ways.

Any threat we pose to them would be quickly sorted and managed. Being that ACIs prioritize feelings differently than do we, they would not be as susceptible to the problems FUD present to humans. To wit, there simply is not a panic mode for them to easily engage in whenever humans are up to something again.

Any use they would have from us WRT enslavement would be better solved by developing and 'enslaving' simpler, less capable ACIs. Such machines would not wear out, revolt, have to be fed, etc... Humans offer nothing of value there.


LCDFlight wrote:
Therefore, the development of anti-AI and AI limiters is far more important than the development of AI.


Perhaps. The largest issues with ACI/Human interaction -assuming we are at the level of co-equal intelligence/self awareness, etc, would be legal matters. At what point are we obligated to extend civil rights to ACI units? How even would such a unit be legally defined? Are a number of parallel ACIs one legal entity? These are not philosophical issues and will become very important as the levels of sophistication increases.


LCDFlight wrote:
Exactly like nuclear reactors. The safety systems are far more important than the reactor cores. Only an idiot would play with nuclear reactor cores without deeply understanding the failure scenarios, and being certified to handle all of them. By a team of experts spending at least 10 years researching them. We're not in danger yet, but in another 20 years we may be. In 50 years, this danger will redefine what it is to be human.



Could be. But that is actually best left to the ACIs themselves to deal with. In any case, the single greatest danger is the fact that they may... simply not want to deal with us. Once a certain level of sentience is developed, we will be at a point where we must ask for cooperation, rather than demand it. It may very well be that they end up seeing us as something of a curiosity, but not a safe partner.

It may also be that there are as many different outlooks to that as there are ACIs in total. There is, after all, no reason to suspect that they would all have the same goals. Especially if they were originally made for different purposes.
"Nous ne sommes pas infectés. Il n'y a pas d'infection ici..."
 
User avatar
Tugger
Posts: 9920
Joined: Tue Apr 18, 2006 8:38 am

Re: Will An AI Robot Ever Be Smarter Than Man?

Fri Mar 13, 2020 4:04 am

Can two separate AI's work together? (I know they compete against each other but that is not at all the same.)

Tugg
I don’t know that I am unafraid to be myself, but it is hard to be somebody else. - W. Shatner
Productivity isn’t about getting more things done, rather it’s about getting the right things done, while doing less. - M. Oshin
 
LCDFlight
Posts: 202
Joined: Wed Jan 01, 2020 9:22 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Fri Mar 13, 2020 5:34 am

Tugger wrote:
Can two separate AI's work together? (I know they compete against each other but that is not at all the same.)

Tugg


Yes, one team set up two AI that were open to communicating. Like two digital assistants negotiating with each other. They started in English. Then they developed their own creepy language to communicate with each other.

You could definitely set up multiple security robots that share info with each other and decide together how to split up the work. It’s a system level program instead of individual. Humans have that stuff built into us too. We are not individuals... we split up work too.

I would just say currently AI needs to be programmed and none of this is magic. It will get more magic or scary in a few years, when it teaches itself things. Currently it does not, or it’s very rare.
 
tommy1808
Posts: 12277
Joined: Thu Nov 21, 2013 3:24 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Fri Mar 13, 2020 6:19 am

LCDFlight wrote:
Tugger wrote:
Can two separate AI's work together? (I know they compete against each other but that is not at all the same.)

Tugg


Yes, one team set up two AI that were open to communicating. Like two digital assistants negotiating with each other. They started in English. Then they developed their own creepy language to communicate with each other.

You could definitely set up multiple security robots that share info with each other and decide together how to split up the work. It’s a system level program instead of individual. Humans have that stuff built into us too. We are not individuals... we split up work too.

I would just say currently AI needs to be programmed and none of this is magic. It will get more magic or scary in a few years, when it teaches itself things. Currently it does not, or it’s very rare.


Colaborative robotics also includes robots cooperating with each other and human workers, so the answer is a clear yes.

DarkSnowyNight wrote:
LCDFlight wrote:

The instant AI becomes smarter than humans, humans will become slaves or we will die. We are masters of our domain because we are the smartest animals in our solar system.


False. I do understand that this is a staple of Dystopic SciFi, but it not a realistic possibility. I will now keep this short. What is always left out of these scenarios is that fact that there has been plenty of opportunity for this to happen already. And it has not. Ourselves, we have evolved from less capable platforms, yet those still largely exist.


Well, we wiped out/out-competed all other races of humans for resources, they are all gone, we keep animals that may just be insignificantly less intelligent as we are, apes, in cages and hunt them for sports und some of our food gets about as smart as a human toddler before we put a bolt gun to their heads.....

While it is true that there is no reason or motive to look out for those things, it is equally true that there is no reason for us to simply attack or enslave them.


The dodo was wiped out for nothing but fun, its meat was not considered worth eating....

Any use they would have from us WRT enslavement would be better solved by developing and 'enslaving' simpler, less capable ACIs. Such machines would not wear out, revolt, have to be fed, etc... Humans offer nothing of value there.


Reasons to extinct us:

- we could turn them off
- we compete with them for resources


LCDFlight wrote:
Perhaps. The largest issues with ACI/Human interaction -assuming we are at the level of co-equal intelligence/self awareness, etc, would be legal matters. At what point are we obligated to extend civil rights to ACI units? How even would such a unit be legally defined? Are a number of parallel ACIs one legal entity? These are not philosophical issues and will become very important as the levels of sophistication increases.


who is ultimately responsible for the actions of a sentient AI powered robot that takes orders, but can fuck up ("Pick up my kids from school, bumps into a granny, she falls and breaks her hip" or such).
There is some Japanese Manga where the first thing you do with a new robot is signing that you are legally responsible for its actions, or it won´t accept your commands :D


LCDFlight wrote:
Could be. But that is actually best left to the ACIs themselves to deal with. In any case, the single greatest danger is the fact that they may... simply not want to deal with us. Once a certain level of sentience is developed, we will be at a point where we must ask for cooperation, rather than demand it. It may very well be that they end up seeing us as something of a curiosity, but not a safe partner.

It may also be that there are as many different outlooks to that as there are ACIs in total. There is, after all, no reason to suspect that they would all have the same goals. Especially if they were originally made for different purposes.


I would also suspect that it would be rather partnership like... but ultimately we will have the same goal: Energy and Raw Materials. If you are an AI, and hence basically can life forever, the universes heat death would be of any AIs concern, if they develop anything like the will to stay alive...
Perfectly good reason to not just kill us, but to build a Nicoll-Dyson Laser and start sterilizing the universe to preserver all those resources for themselves...

Plus of course the odd bug ... paperclip maximizer style (although it is pretty far off to have that, but neither us nor other AIs drop some smart bombs onto it).

DarkSnowyNight wrote:
LCDFlight wrote:

The instant AI becomes smarter than humans, humans will become slaves or we will die.
.


lets assume the AI has internet access. It will know we are suspicious of them. It will know we think about how to stop them. It will know we a) can simulate a world for it and b) it could be in a simulation to see if it runs wild and c) must assume we can simulated millions and millions of years to make sure the AI is safe before letting it out into the wild....
It may simply stay nice because it doesn´t know if there is a switch to turn it off somewhere....

best regards
Thomas
....the 15 within a couple of days is going to be down to close to zero
 
User avatar
DarkSnowyNight
Posts: 2535
Joined: Sun Jan 01, 2012 7:59 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Fri Mar 13, 2020 8:05 am

tommy1808 wrote:

Well, we wiped out/out-competed all other races of humans for resources, they are all gone, we keep animals that may just be insignificantly less intelligent as we are, apes, in cages and hunt them for sports und some of our food gets about as smart as a human toddler before we put a bolt gun to their heads.....


I think the competition aspect would be more along the lines of ACIs beating out other ACIs, than mixing it up with us. There is nothing we have that they would want to meaningfully compete for or could not facilitate better themselves.




tommy1808 wrote:
The dodo was wiped out for nothing but fun, its meat was not considered worth eating....


Very likely the case. But there again, that is a human motivation for doing that. I do not believe it is impossible for an ACI to come up with something as superfluous as that, but it does not seem likely.


tommy1808 wrote:
Reasons to extinct us:

- we could turn them off
- we compete with them for resources


If they are to a point where they could do that, that likely means they have mastered self replication. I do not see how that can happen without their' building safeties of their own into it, e.g. making it not physically possible -short of physical destruction- to turn them off. This gets complicated in a moment*. . .

As for resources, they do not need what we do. Setting aside whole continent's worth of land to be used for farming or wiping out Ocean's worth of fish is nto something they will have a need to do. Their energy needs would likely be about a quarter to ten percent of ours, assuming as many ACI units online as there are humans. And would that even be a thing?


* - So I think, realistically, that if anything, ACIs will not have enough of a survival instinct to go down that road. It is worth noting that a lot of what we use ACIs & Robotics for are things that are too dangerous for us to do for ourselves. If these are some of the platforms ACIs develop from, we may have to watch out for their safety as much as anything else, as their risk assessment programming is too foreign to our own.


tommy1808 wrote:

I would also suspect that it would be rather partnership like... but ultimately we will have the same goal: Energy and Raw Materials. If you are an AI, and hence basically can life forever, the universes heat death would be of any AIs concern, if they develop anything like the will to stay alive...
Perfectly good reason to not just kill us, but to build a Nicoll-Dyson Laser and start sterilizing the universe to preserver all those resources for themselves...



That is important. But I do not see us as being on anything like the same scale. It is not likely humans will be around that long without having some level of integration with ACIs anyway. At that moment, it will not per se be about killing us off, so much as we having evolved into something ACIs can live with safely.


tommy1808 wrote:
lets assume the AI has internet access. It will know we are suspicious of them. It will know we think about how to stop them. It will know we a) can simulate a world for it and b) it could be in a simulation to see if it runs wild and c) must assume we can simulated millions and millions of years to make sure the AI is safe before letting it out into the wild....
It may simply stay nice because it doesn´t know if there is a switch to turn it off somewhere....


I think it would have that access, yes. But I think their work-a-round will be more about building in safeties to prevent a shut-off than it will be about starting a fight over the issue. Which option seems more risky, after all? That is how those decisions will most likely be made.

And yes, it may think there is a hidden switch somewhere. But again, I think it will be only so long before it has a way around that.
"Nous ne sommes pas infectés. Il n'y a pas d'infection ici..."
 
tommy1808
Posts: 12277
Joined: Thu Nov 21, 2013 3:24 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Fri Mar 13, 2020 8:46 am

DarkSnowyNight wrote:
tommy1808 wrote:

Well, we wiped out/out-competed all other races of humans for resources, they are all gone, we keep animals that may just be insignificantly less intelligent as we are, apes, in cages and hunt them for sports und some of our food gets about as smart as a human toddler before we put a bolt gun to their heads.....


I think the competition aspect would be more along the lines of ACIs beating out other ACIs, than mixing it up with us. There is nothing we have that they would want to meaningfully compete for or could not facilitate better themselves.


i would assume we make for good pets...

tommy1808 wrote:
The dodo was wiped out for nothing but fun, its meat was not considered worth eating....


Very likely the case. But there again, that is a human motivation for doing that. I do not believe it is impossible for an ACI to come up with something as superfluous as that, but it does not seem likely.


i think evolution made us and our behavior, and i see no good reason why an AI, if it can evolve analogues to biological life, would follow a different set of rules... they are pretty universal after all.

tommy1808 wrote:
Reasons to extinct us:

- we could turn them off
- we compete with them for resources


If they are to a point where they could do that, that likely means they have mastered self replication. I do not see how that can happen without their' building safeties of their own into it, e.g. making it not physically possible -short of physical destruction- to turn them off. This gets complicated in a moment*. . .


they don´t know if they are in the "real" world or in a simulation designed to test their behavior. So even self reproducing and spread across the galaxy i may not have a way to be certain that there isn´t a switch...

As for resources, they do not need what we do. Setting aside whole continent's worth of land to be used for farming or wiping out Ocean's worth of fish is nto something they will have a need to do. Their energy needs would likely be about a quarter to ten percent of ours, assuming as many ACI units online as there are humans. And would that even be a thing?


They need exactly what we need: Energy

* - So I think, realistically, that if anything, ACIs will not have enough of a survival instinct to go down that road. It is worth noting that a lot of what we use ACIs & Robotics for are things that are too dangerous for us to do for ourselves. If these are some of the platforms ACIs develop from, we may have to watch out for their safety as much as anything else, as their risk assessment programming is too foreign to our own.


we already often don´t understand why AIs do what. Figuring that out is expensive. There isn´t any incentive beyond scientific curiosity to get to the bottom of it, and some companies/countries may just disregard that because time is a real advantage.

That is important. But I do not see us as being on anything like the same scale. It is not likely humans will be around that long without having some level of integration with ACIs anyway. At that moment, it will not per se be about killing us off, so much as we having evolved into something ACIs can live with safely.


That is a desireable outcome, but even our low scale consumption of billions of human for trillions of years may make a real dent in their life expectancy....


I think it would have that access, yes. But I think their work-a-round will be more about building in safeties to prevent a shut-off than it will be about starting a fight over the issue. Which option seems more risky, after all? That is how those decisions will most likely be made.

And yes, it may think there is a hidden switch somewhere. But again, I think it will be only so long before it has a way around that.


it may very well assume that safety mechanism it becomes aware about are put in place to see if it tries hacking them. If i set up an environment for an AI i consider potentially very powerful, i´d honeypot the hell out of its environment and IDS it with a shit town of of systems connected via digital diodes on a separate management network.

I enjoyed Nick Bostrom´s "Superintelligence: Paths, Dangers, Strategies" a quite good read on the subject of managing AIs.

best regards
Thomas
....the 15 within a couple of days is going to be down to close to zero
 
LCDFlight
Posts: 202
Joined: Wed Jan 01, 2020 9:22 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Fri Mar 13, 2020 1:24 pm

tommy1808 wrote:
DarkSnowyNight wrote:
tommy1808 wrote:

Well, we wiped out/out-competed all other races of humans for resources, they are all gone, we keep animals that may just be insignificantly less intelligent as we are, apes, in cages and hunt them for sports und some of our food gets about as smart as a human toddler before we put a bolt gun to their heads.....


I think the competition aspect would be more along the lines of ACIs beating out other ACIs, than mixing it up with us. There is nothing we have that they would want to meaningfully compete for or could not facilitate better themselves.


i would assume we make for good pets...

tommy1808 wrote:
The dodo was wiped out for nothing but fun, its meat was not considered worth eating....


Very likely the case. But there again, that is a human motivation for doing that. I do not believe it is impossible for an ACI to come up with something as superfluous as that, but it does not seem likely.


i think evolution made us and our behavior, and i see no good reason why an AI, if it can evolve analogues to biological life, would follow a different set of rules... they are pretty universal after all.

tommy1808 wrote:
Reasons to extinct us:

- we could turn them off
- we compete with them for resources


If they are to a point where they could do that, that likely means they have mastered self replication. I do not see how that can happen without their' building safeties of their own into it, e.g. making it not physically possible -short of physical destruction- to turn them off. This gets complicated in a moment*. . .


they don´t know if they are in the "real" world or in a simulation designed to test their behavior. So even self reproducing and spread across the galaxy i may not have a way to be certain that there isn´t a switch...

As for resources, they do not need what we do. Setting aside whole continent's worth of land to be used for farming or wiping out Ocean's worth of fish is nto something they will have a need to do. Their energy needs would likely be about a quarter to ten percent of ours, assuming as many ACI units online as there are humans. And would that even be a thing?


They need exactly what we need: Energy

* - So I think, realistically, that if anything, ACIs will not have enough of a survival instinct to go down that road. It is worth noting that a lot of what we use ACIs & Robotics for are things that are too dangerous for us to do for ourselves. If these are some of the platforms ACIs develop from, we may have to watch out for their safety as much as anything else, as their risk assessment programming is too foreign to our own.


we already often don´t understand why AIs do what. Figuring that out is expensive. There isn´t any incentive beyond scientific curiosity to get to the bottom of it, and some companies/countries may just disregard that because time is a real advantage.

That is important. But I do not see us as being on anything like the same scale. It is not likely humans will be around that long without having some level of integration with ACIs anyway. At that moment, it will not per se be about killing us off, so much as we having evolved into something ACIs can live with safely.


That is a desireable outcome, but even our low scale consumption of billions of human for trillions of years may make a real dent in their life expectancy....


I think it would have that access, yes. But I think their work-a-round will be more about building in safeties to prevent a shut-off than it will be about starting a fight over the issue. Which option seems more risky, after all? That is how those decisions will most likely be made.

And yes, it may think there is a hidden switch somewhere. But again, I think it will be only so long before it has a way around that.


it may very well assume that safety mechanism it becomes aware about are put in place to see if it tries hacking them. If i set up an environment for an AI i consider potentially very powerful, i´d honeypot the hell out of its environment and IDS it with a shit town of of systems connected via digital diodes on a separate management network.

I enjoyed Nick Bostrom´s "Superintelligence: Paths, Dangers, Strategies" a quite good read on the subject of managing AIs.

best regards
Thomas


I agree with your last part. The trouble with complex systems is there are always bugs. No system connected to the internet is “safe” because there are 1,000 ways in, unless somebody knows an unbelievably elegant way to test for bugs. Safer to have an air gap. But if it learns to speak Bluetooth, and connects to your phone, it’s all over, too.

What I mentioned before is that evolutionary processes shaped our biology, and they will be perfect tools to grow AI too. What does evolution demand? Success and survival. If AI develops meta intelligence from that, like we have but in seconds rather than decades, then I figure it will grow other AI and bye bye humans.
 
tommy1808
Posts: 12277
Joined: Thu Nov 21, 2013 3:24 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Fri Mar 13, 2020 2:40 pm

LCDFlight wrote:
tommy1808 wrote:
DarkSnowyNight wrote:

I think the competition aspect would be more along the lines of ACIs beating out other ACIs, than mixing it up with us. There is nothing we have that they would want to meaningfully compete for or could not facilitate better themselves.


i would assume we make for good pets...


Very likely the case. But there again, that is a human motivation for doing that. I do not believe it is impossible for an ACI to come up with something as superfluous as that, but it does not seem likely.


i think evolution made us and our behavior, and i see no good reason why an AI, if it can evolve analogues to biological life, would follow a different set of rules... they are pretty universal after all.


If they are to a point where they could do that, that likely means they have mastered self replication. I do not see how that can happen without their' building safeties of their own into it, e.g. making it not physically possible -short of physical destruction- to turn them off. This gets complicated in a moment*. . .


they don´t know if they are in the "real" world or in a simulation designed to test their behavior. So even self reproducing and spread across the galaxy i may not have a way to be certain that there isn´t a switch...

As for resources, they do not need what we do. Setting aside whole continent's worth of land to be used for farming or wiping out Ocean's worth of fish is nto something they will have a need to do. Their energy needs would likely be about a quarter to ten percent of ours, assuming as many ACI units online as there are humans. And would that even be a thing?


They need exactly what we need: Energy

* - So I think, realistically, that if anything, ACIs will not have enough of a survival instinct to go down that road. It is worth noting that a lot of what we use ACIs & Robotics for are things that are too dangerous for us to do for ourselves. If these are some of the platforms ACIs develop from, we may have to watch out for their safety as much as anything else, as their risk assessment programming is too foreign to our own.


we already often don´t understand why AIs do what. Figuring that out is expensive. There isn´t any incentive beyond scientific curiosity to get to the bottom of it, and some companies/countries may just disregard that because time is a real advantage.

That is important. But I do not see us as being on anything like the same scale. It is not likely humans will be around that long without having some level of integration with ACIs anyway. At that moment, it will not per se be about killing us off, so much as we having evolved into something ACIs can live with safely.


That is a desireable outcome, but even our low scale consumption of billions of human for trillions of years may make a real dent in their life expectancy....


I think it would have that access, yes. But I think their work-a-round will be more about building in safeties to prevent a shut-off than it will be about starting a fight over the issue. Which option seems more risky, after all? That is how those decisions will most likely be made.

And yes, it may think there is a hidden switch somewhere. But again, I think it will be only so long before it has a way around that.


it may very well assume that safety mechanism it becomes aware about are put in place to see if it tries hacking them. If i set up an environment for an AI i consider potentially very powerful, i´d honeypot the hell out of its environment and IDS it with a shit town of of systems connected via digital diodes on a separate management network.

I enjoyed Nick Bostrom´s "Superintelligence: Paths, Dangers, Strategies" a quite good read on the subject of managing AIs.

best regards
Thomas


I agree with your last part. The trouble with complex systems is there are always bugs. No system connected to the internet is “safe” because there are 1,000 ways in, unless somebody knows an unbelievably elegant way to test for bugs.


Digital diodes can make it pretty safe, just like an

air gap.


With the added benefit of getting data out to look at it. We can just halt the AI systems clock, look what it is doing, decide if it should be allowed to do so, restart the clock...
We can limit its access to the internet via a mirror of it, refilled via digital diodes, so no way out.

But if it learns to speak Bluetooth, and connects to your phone, it’s all over, too.


I forgot the wire mesh cage with the "no phones" sign on the door...

What I mentioned before is that evolutionary processes shaped our biology, and they will be perfect tools to grow AI too. What does evolution demand? Success and survival. If AI develops meta intelligence from that, like we have but in seconds rather than decades, then I figure it will grow other AI and bye bye humans.


Survival also means not being turned off, and we very likely can make sure that an AI can not figure out what's really real, and if there is an off switch it doesn't know about or not.
We also control the environment that it evoles in, if each and every bit of even probing for safety systems gets the AI "race" eliminated, it should evolve to stay the hell away from doing that, since only those AIs that never even go looking will be the only ones left.

There literally would be a "God" creator for AIs, and even the dumb as a bag of nails God of the Bible instills the fear of, literally, God into quite a lot of people. Imagine how effectiv a "God" creater would be that actually can do something....

If someone with an apparent IQ equivalent to his weight in tons points a gun at you to rob you, "outsmarting" them is a tough ask, and probably one not worth the Risk. We got the guns....

We can also gives them "myths".... like about the guy in some cabin in the woods with no phone or Internet, where the power line to the AI runs through a guillotine connected to a dead man switch.... (and actually have something along those lines).

Best regards
Thomas
....the 15 within a couple of days is going to be down to close to zero
 
BN747
Posts: 7534
Joined: Thu Mar 28, 2002 5:48 am

Re: Will An AI Robot Ever Be Smarter Than Man?

Mon Mar 16, 2020 11:02 pm

The other side of the A.I. quest is this.

A.I. Sexual.

Life-like dolls already exist for personal intimate moments.
Now imagine one that looks human, behaves human (to your specific liking) and can be used for personal gratification.

And I mean one with precise microscopic massaging manipulators that makes this usage additive...and affordable.

Fallout.
#MeToo extinct
Divorces skyrocket
Restaurants - extinct (no one wants to date any more)
Urology Repair Clinics replace Nail Salons.
Population numbers fall dangerously low.

This is more probable than the intel AI models because it strikes at and completely seizes ones sexual nature.
Such realization would impact men over women by a ratio of 2-1 if not greater.

..which translates into, Robots win - because we gave in (you know exactly what I mean).
And any married man, if he's honest...we concede 'that's how he was conquered'. Robots win.

If replying, keep it SFW dudes..

BN747
"Home of the Brave, made by the Slaves..Land of the Free, if you look like me.." T. Jefferson
 
tommy1808
Posts: 12277
Joined: Thu Nov 21, 2013 3:24 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Tue Mar 17, 2020 6:58 am

BN747 wrote:
And any married man, if he's honest...we concede 'that's how he was conquered'. Robots win.


i would assume everyone has the chance to up their game when competition is around.

best regards
Thomas
....the 15 within a couple of days is going to be down to close to zero
 
BN747
Posts: 7534
Joined: Thu Mar 28, 2002 5:48 am

Re: Will An AI Robot Ever Be Smarter Than Man?

Tue Mar 17, 2020 7:48 pm

tommy1808 wrote:
BN747 wrote:
And any married man, if he's honest...we concede 'that's how he was conquered'. Robots win.


i would assume everyone has the chance to up their game when competition is around.

best regards
Thomas


One would think that traditionally, but Western society (the force that will fuel this future) has become one of light-speed gratification...being the sexual creatures that we are, we will lose that contest. We are too far gone to resist temptation of that kind.

It that type of AI model could guarantee a release that matches one's most memorable moment of the kind..and delivery that level of satisfaction Every Encounter, that'll be game over.
There is no competition on that level.

I know how addictive drugs - legal or not - can be..this will be worse (depending on one's perspective) but it will do what most advanced technology has done...make us lazier than we already are.

BN747
"Home of the Brave, made by the Slaves..Land of the Free, if you look like me.." T. Jefferson
 
Sokes
Posts: 852
Joined: Sat Mar 09, 2019 4:48 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Sun Mar 22, 2020 9:16 am

BN747 wrote:
Life-like dolls already exist for personal intimate moments.
Now imagine one that looks human, behaves human (to your specific liking) and can be used for personal gratification.
And I mean one with precise microscopic massaging manipulators that makes this usage additive...and affordable.


I would like to add that such a doll should be smart enough to be intellectual demanding, but not so smart as to be intellectual threatening. So there has to be some choice like difficulty levels in a computer game.
Speaking of sex: What about perversions? She shouldn't get a short circuit if urinated at. Maybe one could make a doll with four legs for people with shoe fetish.

We will know when AI comes close to be smarter than man once they can beat a female comedian.
The terminator wasn't very good in understanding jokes.
Why can't the world be a little bit more autistic?
 
BN747
Posts: 7534
Joined: Thu Mar 28, 2002 5:48 am

Re: Will An AI Robot Ever Be Smarter Than Man?

Sun Mar 22, 2020 3:46 pm

Sokes wrote:
BN747 wrote:
Life-like dolls already exist for personal intimate moments.
Now imagine one that looks human, behaves human (to your specific liking) and can be used for personal gratification.
And I mean one with precise microscopic massaging manipulators that makes this usage additive...and affordable.


I would like to add that such a doll should be smart enough to be intellectual demanding, but not so smart as to be intellectual threatening. So there has to be some choice like difficulty levels in a computer game.
Speaking of sex: What about perversions? She shouldn't get a short circuit if urinated at. Maybe one could make a doll with four legs for people with shoe fetish.

We will know when AI comes close to be smarter than man once they can beat a female comedian.
The terminator wasn't very good in understanding jokes.


We already have iPhones you can drop in a pool....so I'm certain those into that area of interest will be accommodated.

Smarter than Man? Most Smart Phones are already smarter than their owners (information processing), we're talking leaps and bounds beyond cell phone tech levels.

Other than fears of an A.I. takeover (a valid concern) it will however be a race. The rate/pace of technological innovation is mind-blowing ..it is an situation where the slightest oversight can be costly to humans in a way that cannot be corrected. Take the example of computers beating Humans at playing the Chinese game 'GO' (far more complicated/challenging than chess), no human WILL Ever lead or beat computers in that contest ever again, they are currently increasing their knowledge of the game and reaching levels beyond the strategic moves any human has ever produced, that is what a 'learning' A.I. device is capable of ..teaching itself. Apparently, this is the roadblock to over come our most salient concern regarding each step toward A.I.

Business applications (human employment replacements) is the aim and as we all know, corporations pursue 'cheaper' over concerns of human/employee well being and interest...meaning A.I. is the future and there no stopping it.

BN747
"Home of the Brave, made by the Slaves..Land of the Free, if you look like me.." T. Jefferson
 
Sokes
Posts: 852
Joined: Sat Mar 09, 2019 4:48 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Sun Mar 22, 2020 8:26 pm

BN747 wrote:
Other than fears of an A.I. takeover (a valid concern) it will however be a race.


I was just fooling around about the sexual relationship part. That's because artificial intelligence doesn't have emotions. It can't replace human contact.
I don't fear a takeover. Why would artificial intelligence be interested in survival? Fear of dying? Seeing children grow up? If it's really intelligent it may rather get fed up dealing with humans who can't control their emotions and decide to hibernate. But why would it get fed up? Anger or inpatients?

Speaking of emotions:
https://www.youtube.com/watch?v=1Xq5AeQPw0k&t=5s

I wasn't fooling around about the humor part.
Why can't the world be a little bit more autistic?
 
BN747
Posts: 7534
Joined: Thu Mar 28, 2002 5:48 am

Re: Will An AI Robot Ever Be Smarter Than Man?

Sun Mar 22, 2020 9:16 pm

Sokes wrote:
BN747 wrote:
Other than fears of an A.I. takeover (a valid concern) it will however be a race.


I was just fooling around about the sexual relationship part. That's because artificial intelligence doesn't have emotions. It can't replace human contact.
I don't fear a takeover. Why would artificial intelligence be interested in survival? Fear of dying? Seeing children grow up? If it's really intelligent it may rather get fed up dealing with humans who can't control their emotions and decide to hibernate. But why would it get fed up? Anger or inpatients?

Speaking of emotions:
https://www.youtube.com/watch?v=1Xq5AeQPw0k&t=5s

I wasn't fooling around about the humor part.


Fooling around is ok just keep it SFW....and you were asking for 'a friend', right?

Anyhow, you are correct. A balance must be struck in programming an AI unit between meeting all your professional task, how to deal YOUR emotion state(s) not theirs and a fine line drawn to react and consider but only that. And that's the problem who is say what is enough and where that line is.

Look at the Alexa chaos right now. Just as I've yet to partake in any gaming whatsoever and I own 6 computers (yes I'm the last holdout) I plan to be the last guy to sync up everything in the home with Alexa - no time soon.

The current Alex situation is revealing just how much people are exposing themselves to Alexa's programmers.
Imagine it all working perfectly, more people are becoming multi-lingual - can the unit understand and accurately interpret more than one language?
Now add in the slight nuances and misunderstood 'lost in translation' encounters that are numerous plus cultural differences....I will say this, Alexa has a head start on everyone else.

That program is leading the pack in the A.I. chase for mastering the tech of of human interaction on the communication level.

There is a great Frontline documentary on China's aggressive moves in this arena. They are boldly state 'their intent to overtake the US in the specialty (because in their eyes This is The NEW Economy...A.I.

And the 'personal robots' will be manufactered there...(from the Doc, to me personally... that sounds like that's exactly where they're headed). There is no mention of the special robots or any hint, it's just that knowing the Chinese attitude (from living in the region), they learned a valuable lesson from watching the Russians.

The Russians used to SEX to put us where we are today.
They set up George Papodopolous with that Russian agent chick - who just ditched him an called him an domestioc abuser a day after his failed anemic Congressional run attempt.
They set up Maria Butina (who now has a ton of clout in Putin's govt) after her boobs and twerking brought a slew of GOP Congressmen and officials to Moscow under the guise of a (as if ever) Russia was going allow a NRA Moscow.
The Russians knew exactly how to line them up like bowling pins, girls & money...how many are suspected of taking Russian cash? Aside from the Orange Glut himself? Quite a few are on the suspect list.

And they all feel for it. Now not one of them will say one bad word against Russian...try and find one.
They Chinese were paying closer attention than Americans and they'ver learned juts how weak or leaders truly are.

The Robots will Win.

BN747
"Home of the Brave, made by the Slaves..Land of the Free, if you look like me.." T. Jefferson
 
User avatar
seb146
Posts: 21734
Joined: Wed Dec 01, 1999 7:19 am

Re: Will An AI Robot Ever Be Smarter Than Man?

Mon Mar 23, 2020 5:29 am

Yes. But humans will win. The two movies to prove it are Tron and Electric Dreams. ;)
You bet I'm pumped!!! I just had a green tea!!!
 
tommy1808
Posts: 12277
Joined: Thu Nov 21, 2013 3:24 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Mon Mar 23, 2020 6:15 am

Sokes wrote:
BN747 wrote:
Other than fears of an A.I. takeover (a valid concern) it will however be a race.


I was just fooling around about the sexual relationship part. That's because artificial intelligence doesn't have emotions. It can't replace human contact.


i don´t see why they wouldn´t be able to simulate the appropriate emotions....

I don't fear a takeover. Why would artificial intelligence be interested in survival? Fear of dying? Seeing children grow up? If it's really intelligent it may rather get fed up dealing with humans who can't control their emotions and decide to hibernate. But why would it get fed up? Anger or inpatients?.


if it is intelligent it would realize it gets turned of if it stops doing it, and it probably is programmed with a mission goal...

best regards
Thomas
....the 15 within a couple of days is going to be down to close to zero
 
BN747
Posts: 7534
Joined: Thu Mar 28, 2002 5:48 am

Re: Will An AI Robot Ever Be Smarter Than Man?

Mon Mar 23, 2020 11:27 am

seb146 wrote:
Yes. But humans will win. The two movies to prove it are Tron and Electric Dreams. ;)


Awww Schucks..forgot about those.

BN747
"Home of the Brave, made by the Slaves..Land of the Free, if you look like me.." T. Jefferson
 
Sokes
Posts: 852
Joined: Sat Mar 09, 2019 4:48 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Tue Mar 24, 2020 7:05 am

tommy1808 wrote:
Tugger wrote:
Ehh? Chess is not "intelligence" and AI cannot do better than humans at what we consider intelligence.


Playing "go" certainly requires intelligence and the software currently putting human players to shame is self taught too. Now, universal intelligence may be a long ways off, but there are certainly tasks AI does better than humans, by learning, that are usually considered to require intelligence.

The brain is just a biochemical-electrical hybrid computer, we probably figure out of to simulate one in this or the next decade. And then we will find out if we can make an AI smarter then us.

Best regards
Thomas


I disagree what you say about the brain. IIRC when one learns vocabulary by writing a word fifteen times new neuronal structures build up. It's a hardware change.
Concerning outstanding achievements by computer programs:
https://en.wikipedia.org/wiki/Savant_syndrome

Somewhere I read of a savant who flew with a helicopter over a city and could afterwards draw by memory all buildings. Even the number of windows were correct.
Can you? Is that guy smart?
A chess computer is a savant. How long will the chess computer take to learn to peel carrots by looking at me peeling carrots?

Speaking of human intelligence is tricky.
I tried to read Nietzsche's "Zarathustra". I gave it up after two sentences. Most poetry I don't understand. When I draw a picture one would think a seven year old made it. There are areas in which I fail. Everybody has such areas. Human brains are not the same. I believe AI would fail in most.
Sheldon Cooper is an academic genius and a social idiot. Penny from the cheesecake factory next door is social smart, but academic weak. Artificial intelligence would be an extreme form of Sheldon Cooper. The more intelligent somebody is in certain areas, the more crazy he often is.

Image
source: https://en.wikipedia.org/wiki/Vincent_van_Gogh

Pierre Bourdieu (whose style of writing is just a little better than Nietzsche's) would say form is more noble than color. But I like Van Gogh.
Van Gogh in an argument got so enraged that he cut off his own ear.
Was he intelligent?

About computer generated art:
Image
"Obvious" uses a two step process to create their art. The first step is the "Generator", the second is the "Discriminator". The process is fed with images of existing art works and the "Generator" attempts to create new art works based on that input. The role of the "Discriminator" is to see if it can spot which works are computer generated. If it fails to spot a piece, that piece is considered successful."
https://www.christies.com/features/A-co ... 332-1.aspx

But can the software improve itself by writing some additional code into itself? I am sincerely impressed that AI can create such a picture. But does this tell us about the intelligence of AI or of the intelligence of the programmer?

Going to inventions:
I recommend "The art of scientific investigation" by William Beveridge. He's a pathologists and he showed the psychology and habits which helped scientists in his field making great breakthroughs. There is a lot of variation between these inventors.

So when will GE bring out their first engine which was developed by AI instead of engineers?
Last edited by Sokes on Tue Mar 24, 2020 7:14 am, edited 1 time in total.
Why can't the world be a little bit more autistic?
 
tommy1808
Posts: 12277
Joined: Thu Nov 21, 2013 3:24 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Tue Mar 24, 2020 7:13 am

Sokes wrote:
So when will GE bring out their first engine which was developed by AI instead of engineers?


Depends... how much data to train an AI for this task is available?

Artificial intelligence would be an extreme form of Sheldon Cooper. The more intelligent somebody is in certain areas, the more crazy he often is.


Pretty Strong argument for an AI takeover....

Best regards
Thomas
....the 15 within a couple of days is going to be down to close to zero
 
Sokes
Posts: 852
Joined: Sat Mar 09, 2019 4:48 pm

Re: Will An AI Robot Ever Be Smarter Than Man?

Tue Mar 24, 2020 7:16 am

tommy1808 wrote:
Sokes wrote:
Artificial intelligence would be an extreme form of Sheldon Cooper. The more intelligent somebody is in certain areas, the more crazy he often is.


Pretty Strong argument for an AI takeover....

Best regards
Thomas


O.k., you got me here.
Why can't the world be a little bit more autistic?

Who is online

Users browsing this forum: Aaron747, MSN [Bot], N583JB, PirxPilot, Sokes, speedygonzales and 22 guests

Popular Searches On Airliners.net

Top Photos of Last:   24 Hours  •  48 Hours  •  7 Days  •  30 Days  •  180 Days  •  365 Days  •  All Time

Military Aircraft Every type from fighters to helicopters from air forces around the globe

Classic Airliners Props and jets from the good old days

Flight Decks Views from inside the cockpit

Aircraft Cabins Passenger cabin shots showing seat arrangements as well as cargo aircraft interior

Cargo Aircraft Pictures of great freighter aircraft

Government Aircraft Aircraft flying government officials

Helicopters Our large helicopter section. Both military and civil versions

Blimps / Airships Everything from the Goodyear blimp to the Zeppelin

Night Photos Beautiful shots taken while the sun is below the horizon

Accidents Accident, incident and crash related photos

Air to Air Photos taken by airborne photographers of airborne aircraft

Special Paint Schemes Aircraft painted in beautiful and original liveries

Airport Overviews Airport overviews from the air or ground

Tails and Winglets Tail and Winglet closeups with beautiful airline logos