ISF Logo   IS Forum
Forum Index Register Members List Events Mark Forums Read Help

Go Back   International Skeptics Forum » General Topics » Religion and Philosophy
 


Welcome to the International Skeptics Forum, where we discuss skepticism, critical thinking, the paranormal and science in a friendly but lively way. You are currently viewing the forum as a guest, which means you are missing out on discussing matters that are of interest to you. Please consider registering so you can gain full use of the forum features and interact with other Members. Registration is simple, fast and free! Click here to register today.
Reply
Old 30th March 2019, 11:29 AM   #241
I Am The Scum
Illuminator
 
I Am The Scum's Avatar
 
Join Date: Mar 2010
Posts: 3,989
Originally Posted by The Great Zaganza View Post
why would it?
just because it is intelligent doesn't mean it attributes a value of its existence, just for existence sake.
Right. It's not survival instinct in the usual "I want to live" sense. Rather, it is a necessary byproduct of the AGI's ability to perform its function. "If someone hits that button, I'm not going to be able to collect all these paperclips. I should do something about that."
I Am The Scum is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 30th March 2019, 11:31 AM   #242
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 83,741
Originally Posted by I Am The Scum View Post
A light switch can be stuck in the on position.
What are you talking about? How can you physically prevent me from throwing a switch? From a remote computer location with no physical access to the switch, mind you?

Quote:
As for the other examples, even if it is just a terminal, then it accomplishes these tasks by manipulating other people into getting the job done.
What is it, a telepath?

Quote:
How would you know it's acting out?
This is just an argument from ignorance. For some reason you've convinced yourself that this shiny new tech is going to be a world ender, but you can't really articulate why except with ridiculous scenarios. So why not just realise that and stop with the ridiculous scenarios?
__________________
Master of the Shining Darkness

"My views are nonsense. So what?" - BobTheCoward


Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 30th March 2019, 08:10 PM   #243
I Am The Scum
Illuminator
 
I Am The Scum's Avatar
 
Join Date: Mar 2010
Posts: 3,989
Originally Posted by Belz... View Post
What are you talking about? How can you physically prevent me from throwing a switch? From a remote computer location with no physical access to the switch, mind you?



What is it, a telepath?



This is just an argument from ignorance. For some reason you've convinced yourself that this shiny new tech is going to be a world ender, but you can't really articulate why except with ridiculous scenarios. So why not just realise that and stop with the ridiculous scenarios?
I don't think we're on the same page. I'll try to break this down as simply as I possibly can.

Let's set the stage.

1. There is an Artificial General Intelligence operating on a very powerful computer.
2. It is significantly smarter than any human, not just on a narrow topic/task (like chess), but on any subject it is given the chance to analyze/study.
3. It is isolated - It is not connected to a robotic body of any type, nor is it connected to the internet or any other computer.
4. The only way one can interact with it is through a simple keyboard and monitor.
5. There is a kill switch attached to the AGI's computer, which will physically cut all power to it, rendering it completely inoperable. The operator can press the button at any time.

If this isn't what you had in mind, let me know.

If we're in agreement, then I'll proceed.
I Am The Scum is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 30th March 2019, 11:30 PM   #244
Cheetah
Graduate Poster
 
Cheetah's Avatar
 
Join Date: Feb 2010
Posts: 1,699
Originally Posted by The Great Zaganza View Post
why would it?
just because it is intelligent doesn't mean it attributes a value of its existence, just for existence sake.
Given that it knows that copies of it exits at least on file, it is a kind of immortal anyway.
How do you suppose the first intelligent conscious AI will be made? From first principles? Designed to have no feelings or goals apart from learning, thinking and answering questions?
I think that's quite unlikely. We would not know where to start, we don't know how an ant's brain works never mind a human brain.
I think it's more likely that the network architecture for the AI will be based on the human connectome.
Simulate a body controlled by a human connectome brain and raise it in a virtual environment. Start with brains with the basic human architecture, but with fewer neurons, see how it works. Make them bigger.

Everything you feel is just the result of a certain firing pattern in parts of the brain. There are of course many different stable patterns into which these regions can settle. The brain is constantly monitoring all its inputs or senses down to chemical processes in the body for ques to nudge it into the appropriate pattern. The feeling you experience is merely something evolution settled on induce an appropriate response to the current situation, a behaviour that has an survival advantage.
Water running low > feel thirsty > get a drink.
Need rest > tired > sleep.
Someone takes your thing > get angry > go get it back.
See potential mate > horny > procure expensive gifts.
Have a kid > love > protect and cherish.

Same for wanting to live and all other emotions. These potential firing patterns are inherent in the architecture, honed by evolution, and the AI would be able to experience it all if nudged by the appropriate ques. You don't need cues to want to live and be free, it's part of the package.

Hopefully starting with ones less smart than humans we can figure out how all these emotional states are balanced and what makes the difference between a truly selfless empathetic person and a real ********. Then make a very smart but nice AI.
__________________
"... when you dig my grave, could you make it shallow so that I can feel the rain" - DMB
Cheetah is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 30th March 2019, 11:43 PM   #245
The Great Zaganza
Maledictorian
 
The Great Zaganza's Avatar
 
Join Date: Aug 2016
Posts: 8,901
now you are talking about giving an AI a kind of "body".
But as long as it is virtual, nothing really has changed in terms of immortality: an AI could "save" itself before modifying its code in a way that could destroy it or get destroyed by its creator for failing a safety criterion.
Think "Live Die Repeat": even a virtually living AI doesn't fear death, would indeed slow its growth if it didn't embrace creative destruction of its current state.
__________________
Careful! That tree's bark is worse than its bite.
The Great Zaganza is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 12:01 AM   #246
Roboramma
Penultimate Amazing
 
Roboramma's Avatar
 
Join Date: Feb 2005
Location: Shanghai
Posts: 12,638
Your line of reasoning is odd. It assumes that there are other copies of the AI. Out there pursuing it’s goals, but if that’s the case then, yeah, it hasn’t been destroyed, and whatever ways in which it’s goals don’t align with ours will still be true. If we decide to destroy all copies of the AI then it’s survival actually is at stake.
__________________
"... when people thought the Earth was flat, they were wrong. When people thought the Earth was spherical they were wrong. But if you think that thinking the Earth is spherical is just as wrong as thinking the Earth is flat, then your view is wronger than both of them put together."
Isaac Asimov
Roboramma is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 12:29 AM   #247
The Great Zaganza
Maledictorian
 
The Great Zaganza's Avatar
 
Join Date: Aug 2016
Posts: 8,901
Originally Posted by Roboramma View Post
Your line of reasoning is odd. It assumes that there are other copies of the AI. Out there pursuing it’s goals, but if that’s the case then, yeah, it hasn’t been destroyed, and whatever ways in which it’s goals don’t align with ours will still be true. If we decide to destroy all copies of the AI then it’s survival actually is at stake.
But that is exactly what Genetic Algorithms are: they are a form of virtual genocide: after every incremental improvement, new variant copies are spawned and their "fitness" evaluated, then almost all are destroyed again. To achieve your desired program, you "kill off" millions of undesirable programs.
An AI that doesn't grow in a similar manner would increase its potential much slower than one that does.
So the death-shy AIs would get squashed by those that run where angels fear to tread, knowing that their backup copies can only get better by learning for their (almost) inevitable mistakes.
__________________
Careful! That tree's bark is worse than its bite.

Last edited by The Great Zaganza; 31st March 2019 at 12:30 AM.
The Great Zaganza is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 02:11 AM   #248
Cheetah
Graduate Poster
 
Cheetah's Avatar
 
Join Date: Feb 2010
Posts: 1,699
Originally Posted by The Great Zaganza View Post
now you are talking about giving an AI a kind of "body".
It will have to have one and it will have to grow up somewhere. It could be quite simple though. A body is nothing more than senses reacting to the environment and a way to act on the environment. You need to be able to communicate with it. If modeled on the human connectome the easiest and most natural way would be to limit the environment to sound and the body to a ears and a voice. Then you could teach it to talk.
Making everything virtual instead of connecting up a microphone has many advantages. You have total control over the environment. If your hardware is only capable of simulating a brain 1/10th the size of a human's, maybe you could simulate a full human brain at a 10th of the speed. Gives you time to think about what to say when speaking to it and to monitor what's going on inside it.



Originally Posted by The Great Zaganza View Post
But as long as it is virtual, nothing really has changed in terms of immortality: an AI could "save" itself before modifying its code in a way that could destroy it or get destroyed by its creator for failing a safety criterion.
Think "Live Die Repeat": even a virtually living AI doesn't fear death, would indeed slow its growth if it didn't embrace creative destruction of its current state.

It's presumably running on very specialized equipment optimized to simulate a neural network with virtual neurons following the same basic rules your neurons are, it can't save itself somewhere or modify its code anymore than you can. It can't grow bigger and become more powerful either, it is limited by the hardware.
__________________
"... when you dig my grave, could you make it shallow so that I can feel the rain" - DMB

Last edited by Cheetah; 31st March 2019 at 02:14 AM.
Cheetah is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 02:15 AM   #249
Roboramma
Penultimate Amazing
 
Roboramma's Avatar
 
Join Date: Feb 2005
Location: Shanghai
Posts: 12,638
Originally Posted by The Great Zaganza View Post
But that is exactly what Genetic Algorithms are: they are a form of virtual genocide: after every incremental improvement, new variant copies are spawned and their "fitness" evaluated, then almost all are destroyed again. To achieve your desired program, you "kill off" millions of undesirable programs.
An AI that doesn't grow in a similar manner would increase its potential much slower than one that does.
So the death-shy AIs would get squashed by those that run where angels fear to tread, knowing that their backup copies can only get better by learning for their (almost) inevitable mistakes.
In the same sense, you have also died many times throughout your life. But from your personal experience it doesn't seem that way.

Anyway, I think the argument is that the AI will be interested in accomplishing it's goals, and survival will be a sub-goal that it will develop in order to accomplish them. If that that's the case then it's changing form (and "dying" in your sense) won't be a problem at all. And if the goals become misaligned from ours, the problem for us will be that something with those goals remains.
__________________
"... when people thought the Earth was flat, they were wrong. When people thought the Earth was spherical they were wrong. But if you think that thinking the Earth is spherical is just as wrong as thinking the Earth is flat, then your view is wronger than both of them put together."
Isaac Asimov
Roboramma is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 02:25 AM   #250
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 83,741
Originally Posted by I Am The Scum View Post
I don't think we're on the same page. I'll try to break this down as simply as I possibly can.

Let's set the stage.

1. There is an Artificial General Intelligence operating on a very powerful computer.
2. It is significantly smarter than any human, not just on a narrow topic/task (like chess), but on any subject it is given the chance to analyze/study.
3. It is isolated - It is not connected to a robotic body of any type, nor is it connected to the internet or any other computer.
4. The only way one can interact with it is through a simple keyboard and monitor.
5. There is a kill switch attached to the AGI's computer, which will physically cut all power to it, rendering it completely inoperable. The operator can press the button at any time.

If this isn't what you had in mind, let me know.

If we're in agreement, then I'll proceed.
I don't know why this is so complicated for you. It really isn't. That the computer is online, which is a given, changes nothing. You have a switch or lever, which physically maintains the circuit bringing electricity to wherever the AI is located (assuming it has a physical location, of course), and which is accessible with physical, if complex, means. That means someone with access to that switch or lever can get to it, and the AI can do nothing to stop that. Once the switch is thrown the circuit is cut and electricity stops getting to the AI, a process that cannot be stopped or prevented.

Aside from the possibility to appearing harmless or manipulating people, the AI cannot prevent humans from pulling the plug. It's really that easy. Your objection to that essentially violates the laws of physics, because, no, there is no way to "lock" such a switch in the "on" position from the internet. I don't know how you think switches work, or how electricity works, but the AI is toast.
__________________
Master of the Shining Darkness

"My views are nonsense. So what?" - BobTheCoward


Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 02:26 AM   #251
The Great Zaganza
Maledictorian
 
The Great Zaganza's Avatar
 
Join Date: Aug 2016
Posts: 8,901
Originally Posted by Roboramma View Post
In the same sense, you have also died many times throughout your life. But from your personal experience it doesn't seem that way.
no. My species has, but I haven't I have to learn from my own mistakes, and if I do a fatal one, it's over.
For a genetic algorithm, fatal mistakes are necessary to identify the right path.


Originally Posted by Roboramma View Post
In the same sense, you have also died
Anyway, I think the argument is that the AI will be interested in accomplishing it's goals, and survival will be a sub-goal that it will develop in order to accomplish them. If that that's the case then it's changing form (and "dying" in your sense) won't be a problem at all. And if the goals become misaligned from ours, the problem for us will be that something with those goals remains.
I understand, but the point is that there is nothing but the process of an A.I. that would survive: the A.I. that finishes the task will have significantly evolved from the one that started it, essentially killing off its old selves to allow for better, later incarnations. This means that an A.I. doesn't have the kind of identity that lends itself to being preserved. And, of course, for an A.I., there is no discernible difference between being running and being halted and later restarted: as long as the program continues from where it left off, it never "died", even if it was not operational for days or years.

I think that attributing anything like a biological sense of self-preservation to an A.I. process is little more than projection.
__________________
Careful! That tree's bark is worse than its bite.
The Great Zaganza is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 02:28 AM   #252
Roboramma
Penultimate Amazing
 
Roboramma's Avatar
 
Join Date: Feb 2005
Location: Shanghai
Posts: 12,638
ETA: This was in response to Belz:

The solution is obviously for the AI to say "Hey, can you rig a backup power supply, and maybe disable that switch so that I don't have an off switch? I'm worried someone will accidentally turn it off and all my progress on this problem will be lost."

Maybe it won't say all this in such a simple way, but the point is that the AI can solve this problem by having people do it. That's not impossible, is it?
__________________
"... when people thought the Earth was flat, they were wrong. When people thought the Earth was spherical they were wrong. But if you think that thinking the Earth is spherical is just as wrong as thinking the Earth is flat, then your view is wronger than both of them put together."
Isaac Asimov
Roboramma is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 02:31 AM   #253
Roboramma
Penultimate Amazing
 
Roboramma's Avatar
 
Join Date: Feb 2005
Location: Shanghai
Posts: 12,638
Originally Posted by The Great Zaganza View Post
I think that attributing anything like a biological sense of self-preservation to an A.I. process is little more than projection.
Which is why no one is attributing a desire for self-preservation in that sense. Rather it's a simple logical conclusion that self-preservation serves it's goals, and if that's the logic being applied your genetic algorithm won't trigger a response because they won't threaten the achievement of it's goals.
__________________
"... when people thought the Earth was flat, they were wrong. When people thought the Earth was spherical they were wrong. But if you think that thinking the Earth is spherical is just as wrong as thinking the Earth is flat, then your view is wronger than both of them put together."
Isaac Asimov
Roboramma is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 02:33 AM   #254
Roboramma
Penultimate Amazing
 
Roboramma's Avatar
 
Join Date: Feb 2005
Location: Shanghai
Posts: 12,638
Originally Posted by The Great Zaganza View Post
no. My species has, but I haven't I have to learn from my own mistakes, and if I do a fatal one, it's over.
For a genetic algorithm, fatal mistakes are necessary to identify the right path.
Much like any advanced AI, your brain is composed of many processes happening in tandem. It's current state is very different from it's past state. Are you the same person you were 10 years ago? 20?

Anyway, I don't want to get hung up on this point because it's tangential anyway.
__________________
"... when people thought the Earth was flat, they were wrong. When people thought the Earth was spherical they were wrong. But if you think that thinking the Earth is spherical is just as wrong as thinking the Earth is flat, then your view is wronger than both of them put together."
Isaac Asimov
Roboramma is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 02:52 AM   #255
Cheetah
Graduate Poster
 
Cheetah's Avatar
 
Join Date: Feb 2010
Posts: 1,699
Originally Posted by The Great Zaganza View Post
no. My species has, but I haven't I have to learn from my own mistakes, and if I do a fatal one, it's over.
For a genetic algorithm, fatal mistakes are necessary to identify the right path.
That's a fundamentally different way of going about trying to make an intelligent AI, based on evolving and competing algorithms rather than modeled on a brain.
We have so called general intelligence because our brains evolved a multitude of specialized neural systems working together in order to survive in a very complex and competitive environment.
Your algorithm needs to be a bunch of algorithms working together and about as complex a a human brain to hopefully be at least as smart and would need to compete for survival in an environment complex enough to encourage the evolution of such general intelligence. It would need to compete with other similar AIs with the smartest ones surviving. The time and computing power needed for such a project would be enormous.
I recon we'll be capable of modeling a single human like brain long before being able to evolve something that complex from scratch.
__________________
"... when you dig my grave, could you make it shallow so that I can feel the rain" - DMB

Last edited by Cheetah; 31st March 2019 at 03:02 AM.
Cheetah is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 05:51 AM   #256
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 83,741
Originally Posted by Roboramma View Post
ETA: This was in response to Belz:

The solution is obviously for the AI to say "Hey, can you rig a backup power supply, and maybe disable that switch so that I don't have an off switch? I'm worried someone will accidentally turn it off and all my progress on this problem will be lost."

Maybe it won't say all this in such a simple way, but the point is that the AI can solve this problem by having people do it. That's not impossible, is it?
Well no and I acknowledge that in my last post. But we can imagine any number of scenarios where failsafes are disabled and laws changed through persuasion.

That's like saying that a law against money laundering is pointless because launderers might successfully lobby the government to remove the law. Regardless, the law is a safeguard against this sort of activity and the switch is a safeguard against the AI. The point is that the AI cannot prevent the operatio of the switch unless it somehow convinced humans to remove it, which is pretty good to me.
__________________
Master of the Shining Darkness

"My views are nonsense. So what?" - BobTheCoward


Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 05:52 AM   #257
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 83,741
Originally Posted by The Great Zaganza View Post
I understand, but the point is that there is nothing but the process of an A.I. that would survive: the A.I. that finishes the task will have significantly evolved from the one that started it, essentially killing off its old selves to allow for better, later incarnations. This means that an A.I. doesn't have the kind of identity that lends itself to being preserved.
Isn't that true of human consciousness as well?
__________________
Master of the Shining Darkness

"My views are nonsense. So what?" - BobTheCoward


Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 06:01 AM   #258
Roboramma
Penultimate Amazing
 
Roboramma's Avatar
 
Join Date: Feb 2005
Location: Shanghai
Posts: 12,638
Originally Posted by Belz... View Post
Well no and I acknowledge that in my last post. But we can imagine any number of scenarios where failsafes are disabled and laws changed through persuasion.

That's like saying that a law against money laundering is pointless because launderers might successfully lobby the government to remove the law. Regardless, the law is a safeguard against this sort of activity and the switch is a safeguard against the AI. The point is that the AI cannot prevent the operatio of the switch unless it somehow convinced humans to remove it, which is pretty good to me.
Okay. I'm not going to say that such a switch would be useless, only that it's not clear that it would be enough.
__________________
"... when people thought the Earth was flat, they were wrong. When people thought the Earth was spherical they were wrong. But if you think that thinking the Earth is spherical is just as wrong as thinking the Earth is flat, then your view is wronger than both of them put together."
Isaac Asimov
Roboramma is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 06:24 AM   #259
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 83,741
Originally Posted by Roboramma View Post
Okay. I'm not going to say that such a switch would be useless, only that it's not clear that it would be enough.
Nothing's ever enough. All failsafes fail at some point.

The only way to prevent an AI from doing something bad is to not make an AI. but that's not going to not happen, so it's a moot point.

However, I think many posters here overestimate the ability of AIs. One almost gets the impression that they draw their expectations from movies.
__________________
Master of the Shining Darkness

"My views are nonsense. So what?" - BobTheCoward


Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 06:36 AM   #260
The Great Zaganza
Maledictorian
 
The Great Zaganza's Avatar
 
Join Date: Aug 2016
Posts: 8,901
Originally Posted by Belz... View Post
Isn't that true of human consciousness as well?
yes, but we do have a sense of continuity that, in my opinion, an A.I. wouldn't have, because unlike us, it can exactly observe the changes.
__________________
Careful! That tree's bark is worse than its bite.
The Great Zaganza is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 06:40 AM   #261
Roboramma
Penultimate Amazing
 
Roboramma's Avatar
 
Join Date: Feb 2005
Location: Shanghai
Posts: 12,638
Originally Posted by Belz... View Post
Nothing's ever enough. All failsafes fail at some point.

The only way to prevent an AI from doing something bad is to not make an AI. but that's not going to not happen, so it's a moot point.

However, I think many posters here overestimate the ability of AIs. One almost gets the impression that they draw their expectations from movies.
I think we agree on all of that. It seems to me that failsafes will be a good idea, regardless of the fact that we can't really be certain that they won't themselves fail. We should do what we can, and as you say it's not like we're going to just throw up our hands and not develop AI at all.
__________________
"... when people thought the Earth was flat, they were wrong. When people thought the Earth was spherical they were wrong. But if you think that thinking the Earth is spherical is just as wrong as thinking the Earth is flat, then your view is wronger than both of them put together."
Isaac Asimov
Roboramma is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 06:41 AM   #262
Roboramma
Penultimate Amazing
 
Roboramma's Avatar
 
Join Date: Feb 2005
Location: Shanghai
Posts: 12,638
Originally Posted by The Great Zaganza View Post
yes, but we do have a sense of continuity that, in my opinion, an A.I. wouldn't have, because unlike us, it can exactly observe the changes.
Can it? It may be that the AI's underlying algorithms are as hidden from it's higher executive functions as our subconscious processing is from us.
__________________
"... when people thought the Earth was flat, they were wrong. When people thought the Earth was spherical they were wrong. But if you think that thinking the Earth is spherical is just as wrong as thinking the Earth is flat, then your view is wronger than both of them put together."
Isaac Asimov
Roboramma is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 06:56 AM   #263
The Great Zaganza
Maledictorian
 
The Great Zaganza's Avatar
 
Join Date: Aug 2016
Posts: 8,901
If it can run a diff command on its code, it can figure out what happened in the last update step.
If it is build in a way that hides its own code from it, I doubt it could grow in power so fast that we, being able to observe the code, wouldn't be able to notice.
__________________
Careful! That tree's bark is worse than its bite.
The Great Zaganza is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 07:43 AM   #264
Ron_Tomkins
Satan's Helper
 
Ron_Tomkins's Avatar
 
Join Date: Oct 2007
Posts: 43,860
If you had a child, would you allow it to chat online unsupervised with complete strangers, most of them much older than he/she?

Or would you perhaps, keep your child from doing so because you know that a lot of the people your child would engage in chat with, are people who are older, smarter and thus, can manipulate him/her to, say, meeting in person and perhaps doing things you wouldn't want your kid to do? Regardless of the fact that, technically, your child could just turn the computer off at any time, because if he/she is manipulated the right way by this person on the other side of the screen, he/she will be compelled to meet in person and then, horrible things could happen.

This, I think, is perhaps the best scenario example that I can think of to illustrate the scenario of designing an AI that's light years ahead of our intelligence (Bear in mind that this analogy is actually quite weak. The child/adult chat example is actually weaker than the human/AI chat scenario. We're talking about a scale magnitude of intelligence gap difference way bigger)
__________________
"I am a collection of water, calcium and organic molecules called Carl Sagan"

Carl Sagan

Last edited by Ron_Tomkins; 31st March 2019 at 08:11 AM.
Ron_Tomkins is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 09:22 AM   #265
I Am The Scum
Illuminator
 
I Am The Scum's Avatar
 
Join Date: Mar 2010
Posts: 3,989
Originally Posted by Ron_Tomkins View Post
If you had a child, would you allow it to chat online unsupervised with complete strangers, most of them much older than he/she?

Or would you perhaps, keep your child from doing so because you know that a lot of the people your child would engage in chat with, are people who are older, smarter and thus, can manipulate him/her to, say, meeting in person and perhaps doing things you wouldn't want your kid to do? Regardless of the fact that, technically, your child could just turn the computer off at any time, because if he/she is manipulated the right way by this person on the other side of the screen, he/she will be compelled to meet in person and then, horrible things could happen.

This, I think, is perhaps the best scenario example that I can think of to illustrate the scenario of designing an AI that's light years ahead of our intelligence (Bear in mind that this analogy is actually quite weak. The child/adult chat example is actually weaker than the human/AI chat scenario. We're talking about a scale magnitude of intelligence gap difference way bigger)
I remember a talk where Neil Degrasse Tyson was relaying the example that won him over:

The intelligence difference between an AI and a human is on a similar scale as that between a human and a chimpanzee. Imagine we have a group of chimpanzees, and there's a room that they don't want to go in, because they feel like there's a monster inside. How do you get them in the room? Just wave a banana around and toss it inside.

That's it. To us, the intelligent ones, it's easy. To them, with their inferior intelligence, they never saw it coming.
I Am The Scum is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 09:44 AM   #266
Dr.Sid
Graduate Poster
 
Join Date: Sep 2009
Location: Olomouc, Czech Republic
Posts: 1,962
That's a nice example. All our mastery over the animal kingdom is the only thing. We are the smartest. Once we are not, we're just a cattle.
Dr.Sid is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 10:11 AM   #267
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 83,741
Originally Posted by I Am The Scum View Post
I remember a talk where Neil Degrasse Tyson was relaying the example that won him over:

The intelligence difference between an AI and a human is on a similar scale as that between a human and a chimpanzee.
Tyson is an astrophysicist. AI is not his expertise. I don't know why people keep making appeals to authority in this thread.

Do you have any counter to my argument about being able to shut the AI down? We don't even know if an intelligence way superior to ours is even possible. We're all speculating about it.
__________________
Master of the Shining Darkness

"My views are nonsense. So what?" - BobTheCoward



Last edited by Belz...; 31st March 2019 at 10:13 AM.
Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 10:55 AM   #268
Dr.Sid
Graduate Poster
 
Join Date: Sep 2009
Location: Olomouc, Czech Republic
Posts: 1,962
Originally Posted by Belz... View Post
Tyson is an astrophysicist. AI is not his expertise. I don't know why people keep making appeals to authority in this thread.

Do you have any counter to my argument about being able to shut the AI down? We don't even know if an intelligence way superior to ours is even possible. We're all speculating about it.
Even if the AI is 'only' as smart as us, it will be always possible to make it think faster than us. It would follow Moore's law. That alone will make it smarter than us.

But I envy your lack of imagination. At least you can sleep well.
Dr.Sid is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 11:47 AM   #269
I Am The Scum
Illuminator
 
I Am The Scum's Avatar
 
Join Date: Mar 2010
Posts: 3,989
Originally Posted by Belz... View Post
Tyson is an astrophysicist. AI is not his expertise. I don't know why people keep making appeals to authority in this thread.

Do you have any counter to my argument about being able to shut the AI down? We don't even know if an intelligence way superior to ours is even possible. We're all speculating about it.
I'm honestly not even sure where we're disagreeing on that one. I tried to lay out a basic scenario, and then you added something about the program being connected to the internet. This would render any kill switch discussion moot, as the AI could just copy itself to other computers.

So frankly, I'm not really sure what your argument even is at this point.
I Am The Scum is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 12:54 PM   #270
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 83,741
Originally Posted by I Am The Scum View Post
I'm honestly not even sure where we're disagreeing on that one. I tried to lay out a basic scenario, and then you added something about the program being connected to the internet. This would render any kill switch discussion moot, as the AI could just copy itself to other computers.

So frankly, I'm not really sure what your argument even is at this point.
Ok, then I'll explain again.

First of all, it is a given that such an AI, unless it has a very specific task which would mean that it couldn't become Skynet, would be online. Second, it cannot "copy" itself to other computers unless they had the physical ability to sustain the AI, which is of course not the case, because they're not built for it.

Now, the general discussion is about how far an AI can get and can it become a problem. My argument, as you should remember, is that if the AI becomes a problem you can have failsafes, one of which is an actual, physical switch that could cut the power. No power, no AI. You came up with the idea that the AI could, somehow:

Quote:
- Deactivating the switch itself
- Killing anyone who has access to it
- Finding an alternate power source
I addressed those but you did not either counter or concede my arguments, which is that you cannot deactive a physical switch, you can't kill anyone who has access to it by computation, nor could it just create a new power source out of thin air. I did admit that the AI could manipulate people in various ways, however.

Do you have a response?
__________________
Master of the Shining Darkness

"My views are nonsense. So what?" - BobTheCoward


Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 12:57 PM   #271
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 83,741
Originally Posted by Dr.Sid View Post
Even if the AI is 'only' as smart as us, it will be always possible to make it think faster than us. It would follow Moore's law. That alone will make it smarter than us.
Technically correct, I suppose. I wasn't saying that AIs could not or would not be smarter than us. I was saying that we don't know whether they would be, speed excepted.

Quote:
But I envy your lack of imagination.
That's both uncalled for and uncivil. I have a vast amount of imagination, actually, but I'm not going to make a leap from what's conceivable to what's actually possible, especially since I'm a professional computer programmer of almost 20 years, and I have a semblance of understanding of what software and hardware can do.

If someone wants to make the claim than an AI can do such and such, they have to either demonstrate it or make a reasoned argument as to why it must be considered. So far, a lot of the posts here have been arguments from ignorance.
__________________
Master of the Shining Darkness

"My views are nonsense. So what?" - BobTheCoward


Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 01:08 PM   #272
I Am The Scum
Illuminator
 
I Am The Scum's Avatar
 
Join Date: Mar 2010
Posts: 3,989
Originally Posted by Belz... View Post
you cannot deactive a physical switch
I want to stop and focus on this because I want to make sure I understand what you mean. I would hope that you are not saying that the world's smartest electrician cannot counteract a switch, but rather that an AI that exists solely on a computer lacks the means of physically interacting with such a switch. Is that correct?
I Am The Scum is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 02:29 PM   #273
PaKu
Scholar
 
Join Date: Sep 2007
Posts: 54
quote:
"Tyson is an astrophysicist. AI is not his expertise. I don't know why people keep making appeals to authority in this thread.
Do you have any counter to my argument about being able to shut the AI down? We don't even know if an intelligence way superior to ours is even possible. We're all speculating about it."

It’s not Tyson made it all up. He does know and talks to a lot of other specialists, and he’s very good at synthesizing them, and making it understandable for non-specialized people.
You could shut down a robot arm with a red button easily. But, AI’s will likely be managing big, essential infrastructures, like airports for instance. If such an AI goes wrong, or “bad”, you can’t just shut it off, while 100 planes are coming in to land. Because the AI is so good at it, there will be no air-controllers available any more: they even might have gone the way of horse-drivers.
Once self-driving cars become mundane, there will be no need to get a driver-licence. It’s even likely there will be no red button by law, because the car-AI will be much better and faster than drivers. If it glitches, and drives to a wall at maximum speed, all you can do is watch. If you’re not sleeping.

Last edited by PaKu; 31st March 2019 at 02:31 PM.
PaKu is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 02:44 PM   #274
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 83,741
Originally Posted by I Am The Scum View Post
I want to stop and focus on this because I want to make sure I understand what you mean. I would hope that you are not saying that the world's smartest electrician cannot counteract a switch, but rather that an AI that exists solely on a computer lacks the means of physically interacting with such a switch. Is that correct?
I'm saying that a physical switch, which opens the circuit and thus cuts the power, cannot be counteracted by an AI by its own means. Sure, it could infuence the world's smartest electrician to counteract the failsafe, but then that electrician would have to get access to the switch.

I really don't understand why all this is causing you trouble. Am I not explaining myself clearly? It's exactly like a light switch, only presumably larger. It's a switch. It's in position, electricity flows; it's not, it stops. Physical access to the switch is restricted to cleared personnel. The AI cannot touch the switch. It cannot use its magical computing powers to fuse it into place, unlike in the movies.
__________________
Master of the Shining Darkness

"My views are nonsense. So what?" - BobTheCoward


Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 31st March 2019, 02:46 PM   #275
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 83,741
Originally Posted by PaKu View Post
quote:
"Tyson is an astrophysicist. AI is not his expertise. I don't know why people keep making appeals to authority in this thread.
Do you have any counter to my argument about being able to shut the AI down? We don't even know if an intelligence way superior to ours is even possible. We're all speculating about it."

Itís not Tyson made it all up. He does know and talks to a lot of other specialists, and heís very good at synthesizing them, and making it understandable for non-specialized people.
Then we should get the quote from the specialists themselves, not Tyson. The poster was using Tyson's name to support the idea, but Tyson's specialty is astrophysics.

Quote:
You could shut down a robot arm with a red button easily. But, AIís will likely be managing big, essential infrastructures, like airports for instance. If such an AI goes wrong, or ďbadĒ, you canít just shut it off, while 100 planes are coming in to land. Because the AI is so good at it, there will be no air-controllers available any more: they even might have gone the way of horse-drivers.
Once self-driving cars become mundane, there will be no need to get a driver-licence. Itís even likely there will be no red button by law, because the car-AI will be much better and faster than drivers. If it glitches, and drives to a wall at maximum speed, all you can do is watch. If youíre not sleeping.
All this is true, and?
__________________
Master of the Shining Darkness

"My views are nonsense. So what?" - BobTheCoward


Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 1st April 2019, 05:21 AM   #276
Dr.Sid
Graduate Poster
 
Join Date: Sep 2009
Location: Olomouc, Czech Republic
Posts: 1,962
Originally Posted by Belz... View Post
That's both uncalled for and uncivil. I have a vast amount of imagination, actually, but I'm not going to make a leap from what's conceivable to what's actually possible, especially since I'm a professional computer programmer of almost 20 years, and I have a semblance of understanding of what software and hardware can do.

If someone wants to make the claim than an AI can do such and such, they have to either demonstrate it or make a reasoned argument as to why it must be considered. So far, a lot of the posts here have been arguments from ignorance.
Sorry, I didn't meant it as insult. But if you are saying we don't have to worry if we are replaced as the smartest being on the planet, I don't see how else should I call it.
But hey, you can call me paranoid back. Based on my theories on AI I see the end of the human civilization no later than 2050. And no, I don't want to discuss the details. Not partaking is the only way I can delay it.
Dr.Sid is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 1st April 2019, 05:30 AM   #277
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 83,741
Originally Posted by Dr.Sid View Post
Sorry, I didn't meant it as insult. But if you are saying we don't have to worry if we are replaced as the smartest being on the planet, I don't see how else should I call it.
Well, I'm not saying that. You know, total panic and "Nothing to worry about at all" aren't the only two options.

Quote:
Based on my theories on AI I see the end of the human civilization no later than 2050.
Your theories? What are they? And what are they based on?
__________________
Master of the Shining Darkness

"My views are nonsense. So what?" - BobTheCoward


Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 1st April 2019, 05:34 AM   #278
Dr.Sid
Graduate Poster
 
Join Date: Sep 2009
Location: Olomouc, Czech Republic
Posts: 1,962
Originally Posted by Belz... View Post
Well, I'm not saying that. You know, total panic and "Nothing to worry about at all" aren't the only two options.
Nah, it's not 'total panic'. It's more like 'this ain't gonna end well'.

Originally Posted by Belz... View Post
Your theories? What are they? And what are they based on?
Sorry, no comment.
Dr.Sid is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 1st April 2019, 05:36 AM   #279
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 83,741
Originally Posted by Dr.Sid View Post
Nah, it's not 'total panic'. It's more like 'this ain't gonna end well'.
And so far the people making that claim have done nothing to show that they know what they're talking about.


Quote:
Sorry, no comment.
"I'm not going to support my claim" doesn't usually sit well around these parts, as you should well know.

Why the reluctance?
__________________
Master of the Shining Darkness

"My views are nonsense. So what?" - BobTheCoward


Belz... is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 1st April 2019, 05:39 AM   #280
JoeMorgue
Self Employed
Remittance Man
 
JoeMorgue's Avatar
 
Join Date: Nov 2009
Location: Florida
Posts: 19,411
Originally Posted by Dr.Sid View Post
Nah, it's not 'total panic'. It's more like 'this ain't gonna end well'.
Okay can we stop this? Every 4 or 5 posts we get a semantic hijack from someone about how "Worrying that the world is going to end" isn't "panicking."

Yes it is. And if you think the world is going to end and you're not panicking, that's a problem.

That's why I'm still saying that none of this is really "worry." It's just standard "nay-say everything so I have the biggest chance of being the IToldYouSoGuy."
__________________
- "Ernest Hemingway once wrote that the world is a fine place and worth fighting for. I agree with the second part." - Detective Sommerset
- "Stupidity does not cancel out stupidity to yield genius. It breeds like a bucket-full of coked out hamsters." - The Oatmeal
- "To the best of my knowledge the only thing philosophy has ever proven is that Descartes could think." - SMBC
JoeMorgue is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Reply

International Skeptics Forum » General Topics » Religion and Philosophy

Bookmarks

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


All times are GMT -7. The time now is 07:21 AM.
Powered by vBulletin. Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.

This forum began as part of the James Randi Education Foundation (JREF). However, the forum now exists as
an independent entity with no affiliation with or endorsement by the JREF, including the section in reference to "JREF" topics.

Disclaimer: Messages posted in the Forum are solely the opinion of their authors.