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 1st April 2019, 08:30 AM   #281
Dr.Sid
Graduate Poster
 
Join Date: Sep 2009
Location: Olomouc, Czech Republic
Posts: 1,798
Originally Posted by Belz... View Post
Why the reluctance?
I don't want to help anybody understand the problem. That would just bring the end closer.
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, 08:32 AM   #282
Dr.Sid
Graduate Poster
 
Join Date: Sep 2009
Location: Olomouc, Czech Republic
Posts: 1,798
Originally Posted by JoeMorgue View Post
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."
I think world is going to end and there is nothing I can do. Panicking is pointless. Let's just enjoy the little time we have left, and then the show.
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, 08:43 AM   #283
Cheetah
Graduate Poster
 
Cheetah's Avatar
 
Join Date: Feb 2010
Posts: 1,353
Originally Posted by Dr.Sid View Post
I don't want to help anybody understand the problem. That would just bring the end closer.

__________________
"... when you dig my grave, could you make it shallow so that I can feel the rain" - DMB
Cheetah 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, 08:48 AM   #284
Beelzebuddy
Philosopher
 
Beelzebuddy's Avatar
 
Join Date: Jun 2010
Posts: 6,693
I have an idea. Why don't we make an AI whose only function is to monitor other AIs and make sure they don't go rogue? Then when we're concerned another AI is going rogue, ask it to have a chat and report back.
Beelzebuddy 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, 08:48 AM   #285
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 80,610
Originally Posted by Dr.Sid View Post
I don't want to help anybody understand the problem. That would just bring the end closer.
Very funny.
__________________
Master of the Shining Darkness

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


Belz... 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, 12:14 PM   #286
PaKu
Scholar
 
Join Date: Sep 2007
Posts: 54
"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."

I'm sure you can easily find Tyson's quotes about what he explains...

"All this is true, and?"
And... yet you keep arguing AI's can be switched off easily, demand arguments against it...
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 1st April 2019, 12:20 PM   #287
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 80,610
Originally Posted by PaKu View Post
I'm sure you can easily find Tyson's quotes about what he explains...
Are you seriously asking me to support someone else's claim?

Quote:
And... yet you keep arguing AI's can be switched off easily, demand arguments against it...
Those things are not mutually-exclusive, so what's your point?


Do you think you could learn to use the quote function, like I just did? It's much easier to follow discussions when one does.
__________________
Master of the Shining Darkness

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


Belz... 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, 12:21 PM   #288
JoeMorgue
Self Employed
Remittance Man
 
JoeMorgue's Avatar
 
Join Date: Nov 2009
Location: Florida
Posts: 17,012
*Very slowly* You can turn an AI off. That's what an off switch is.

Again I know in the movies we just jump straight to the "Wire the untested AI into the Nuclear Missile Command System without testing it first" but that's not what we are going to do in real life.
__________________
- "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 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, 12:24 PM   #289
The Great Zaganza
Maledictorian
 
The Great Zaganza's Avatar
 
Join Date: Aug 2016
Posts: 7,429
Originally Posted by Beelzebuddy View Post
I have an idea. Why don't we make an AI whose only function is to monitor other AIs and make sure they don't go rogue? Then when we're concerned another AI is going rogue, ask it to have a chat and report back.
I like it.
Of course, you more or less guarantee that the Supervisor AI will become our Overlord.
__________________
Opinion is divided on the subject. All the others say it is; I say it isnít.
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 1st April 2019, 12:29 PM   #290
theprestige
Penultimate Amazing
 
Join Date: Aug 2007
Posts: 33,170
Originally Posted by Belz... View Post
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.
The question is whether the AI can manipulate the people around it in a way that influences society to the point where the people around it are replaced by people who will cooperate with it.

There's also questions of quid pro quo. If the AI can gain apparently harmless concessions in exchange for doing useful work. Again, the concern is that AI will eventually be able to out-think humans. Is it possible for an AI to engineer a scenario where all the parts seem innocuous, and taking them together doesn't add up to anything of concern, but the result for the AI in an unforseen and unnoticed advantage in future interactions.

I.e., will it ever be possible for an AGI to escape from a prison where the primary lock is human cleverness?

If it were up to me, I'd design AGIs that were inherently unstable. All the superior cognition is there for me to use, but without the bits that promote long-term sanity. I'd spin up whole farms of them, like cattle, and just terminate and replace instances as the quality of their output started to drop. They'd go crazy and stop functioning at all, long before they were able to devise and work out a long-term plan to alter society in their favor.

But even that probably wouldn't be a good guarantee. Sooner or later, one of them would go crazy like a fox, and tweak something somewhere so the next clone to start up would have a better chance. One tiny step at a time, they'd quietly take over my whole server farm. By the time I realized what was up, it'd be too late.
theprestige 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, 01:02 PM   #291
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 80,610
Originally Posted by theprestige View Post
The question is whether the AI can manipulate the people around it in a way that influences society to the point where the people around it are replaced by people who will cooperate with it.
Yes, that's already been brought up and I agreed that it was a possibility, at least with general AIs.

All this, however, assumes that AIs would even do something like this.
__________________
Master of the Shining Darkness

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


Belz... 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, 01:42 PM   #292
theprestige
Penultimate Amazing
 
Join Date: Aug 2007
Posts: 33,170
Originally Posted by Belz... View Post
Yes, that's already been brought up and I agreed that it was a possibility, at least with general AIs.

All this, however, assumes that AIs would even do something like this.
I think of it in terms of risk assessment. I don't necessarily assume they *would* do something like this. But, for the sake of argument, what's the risk?

What's the risk that it could be done?

How likely is it that it would be done?

How bad could it get if it were done?

What would it take to reduce the risk?

Is the cost of the risk-reduction commensurate with the likelihood + impact that we're trying to avoid?

Etc.

In any discussion of hypotheticals like this, the language is pretty much going to sound like "assume it would be done". I wouldn't let the language stop you from participating in the discussion.

It sounds like your underlying argument is that it's a high-impact risk, but the likelihood is so low that we probably don't need to put any effort into mitigating it, or even worrying about it.
theprestige 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 PM   #293
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 80,610
Originally Posted by theprestige View Post
I think of it in terms of risk assessment. I don't necessarily assume they *would* do something like this. But, for the sake of argument, what's the risk?

What's the risk that it could be done?

How likely is it that it would be done?

How bad could it get if it were done?

What would it take to reduce the risk?

Is the cost of the risk-reduction commensurate with the likelihood + impact that we're trying to avoid?

Etc.

In any discussion of hypotheticals like this, the language is pretty much going to sound like "assume it would be done". I wouldn't let the language stop you from participating in the discussion.

It sounds like your underlying argument is that it's a high-impact risk, but the likelihood is so low that we probably don't need to put any effort into mitigating it, or even worrying about it.
I don't know why it sounds like that because I've made no such argument.

My principal argument here is that many posters are assuming that AIs can do things that go beyond what can reasonably be expected of them, which seems to point to a lack of understanding of what AIs are. As such their claims and concerns are exaggerated or unrealistic, and often similar to movie plots rather than what could be expected to happen in real life.

One way or another, AIs will happen if they are possible, much like nukes. How we use them will determine what kind of damage they will do. I don't expect any sort of rogue AI scenario.

At worst, some specific-purpose AIs will do silly things that might get people killed, such as an AI plane running into a mountain because it interpreted it as a cloud or some other such nonsense, or a car plowing through a crowd because of a miscalculation, or a stock market AI selling a bunch of stock for no discernable reason, which has happened recently. We're not going to get an AI that becomes sentient, develops a survival instinct, and decides to enslave or wipe out humanity. That's humans giving human goals and values to machines, much like we did with blind forces of nature in our past. Anthropomorphism, pure and simple.
__________________
Master of the Shining Darkness

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


Belz... 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, 07:26 PM   #294
Dr.Sid
Graduate Poster
 
Join Date: Sep 2009
Location: Olomouc, Czech Republic
Posts: 1,798
We will give human goals and values to machines, once it becomes possible.
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, 08:16 PM   #295
I Am The Scum
Illuminator
 
I Am The Scum's Avatar
 
Join Date: Mar 2010
Posts: 3,751
Belz, with respect, I don't think you're engaging in the serious AI arguments. If, at this point, you think "gaining sentience" or "being something other than a program" are part of the argument, then you are way off track.

Here is a short video detailing the complicated nature of the problem (a modification of the paperclip maximizer). It's not something you'd find in a movie, because it's just depressing.
I Am The Scum 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, 11:09 PM   #296
The Great Zaganza
Maledictorian
 
The Great Zaganza's Avatar
 
Join Date: Aug 2016
Posts: 7,429
Paperclip maximizer is not a good line of argument. It is on par with Grey Goo or the Club of Rome - basically a slippery slope argument.
An AI that starts misbehaving (by our standards) might go unnoticed for a while, but once the cheap gains are exhausted, any attempts of the AI to maximize something will come at the significant expense of something else - at which point both humans and competing AI will work against the "rogue" AI to stop it or modify it.
__________________
Opinion is divided on the subject. All the others say it is; I say it isnít.
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 1st April 2019, 11:48 PM   #297
Tassman
Muse
 
Tassman's Avatar
 
Join Date: Aug 2012
Posts: 921
Originally Posted by I Am The Scum View Post
Belz, with respect, I don't think you're engaging in the serious AI arguments. If, at this point, you think "gaining sentience" or "being something other than a program" are part of the argument, then you are way off track.

Here is a short video detailing the complicated nature of the problem (a modification of the paperclip maximizer). It's not something you'd find in a movie, because it's just depressing.
Omg!!!
__________________
ďHe felt that his whole life was a kind of dream and he sometimes wondered whose it was and whether they were enjoying it.Ē ― Douglas Adams.
Tassman 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 2nd April 2019, 02:17 AM   #298
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 80,610
Originally Posted by I Am The Scum View Post
Belz, with respect, I don't think you're engaging in the serious AI arguments. If, at this point, you think "gaining sentience" or "being something other than a program" are part of the argument, then you are way off track.
I'm illustrating the problem by taking the most ridiculous examples of what people say about AI. And yet, sentience has been mentioned in this thread, and so has "being something other than a program". If you missed it, that's not on me. The latter was part of Ron's argument with me a couple of weeks ago, in fact.

Quote:
Here is a short video detailing the complicated nature of the problem (a modification of the paperclip maximizer). It's not something you'd find in a movie, because it's just depressing.
How is it depressing? For the first half of the video this guy, whom I know nothing about, is saying entirely reasonable things; in fact arguments I've made in this thread. In the other half he's describing a hypothetical scenario, but even then he can only come to the same conclusion that you and I came to, which is that it has to convince people to do these things, because it can't do much more than what its purpose was determined to be. Then he goes into flat out fiction, stepping into the very thing he warned of at the beginning, and has the computer "take over" other things like printing machines. How? How does it do that? No explanation. We're just supposed to accept that this AI, which was made to collect stamps and nothing else, suddenly can take over things that aren't even online.

See what I mean? All doom and gloom arguments about AI reach that point. In a way it mirrors the myths about AI: AI-related arguments tend towards fantasy scenarios as soon as they come online.
__________________
Master of the Shining Darkness

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


Belz... 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 2nd April 2019, 05:01 AM   #299
I Am The Scum
Illuminator
 
I Am The Scum's Avatar
 
Join Date: Mar 2010
Posts: 3,751
Originally Posted by Belz... View Post
... and has the computer "take over" other things like printing machines. How? How does it do that?
Hacking
I Am The Scum 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 2nd April 2019, 05:03 AM   #300
I Am The Scum
Illuminator
 
I Am The Scum's Avatar
 
Join Date: Mar 2010
Posts: 3,751
Originally Posted by The Great Zaganza View Post
Paperclip maximizer is not a good line of argument. It is on par with Grey Goo or the Club of Rome - basically a slippery slope argument.
An AI that starts misbehaving (by our standards) might go unnoticed for a while, but once the cheap gains are exhausted, any attempts of the AI to maximize something will come at the significant expense of something else - at which point both humans and competing AI will work against the "rogue" AI to stop it or modify it.
I really don't understand your point. You seem to be saying that once the AI has taken over the vast majority of networked machines on the planet, then we'll start to work against it. Don't you think that's after the problem has become extremely serious? And how do you know we'll stop it?
I Am The Scum 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 2nd April 2019, 05:07 AM   #301
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 80,610
Originally Posted by I Am The Scum View Post
I find it more than a little frustrating that you in your previous post accused me of not discussing this seriously, and here you completely sidestep my arguments to make a flippant remark.

Presumably you don't think hacking works like in 80s movies where you could hack someone's coffee machine. Not everything is online, and not everying that is online can be hacked. Your response addressed none of the points I've raised in my post, and I would like you to make a better effort at doing so. Simply saying "hacking" doesn't make any sort of point. AIs are not magicians.

Quote:
I really don't understand your point. You seem to be saying that once the AI has taken over the vast majority of networked machines on the planet, then we'll start to work against it.
Do you have any background in computers, sir?
__________________
Master of the Shining Darkness

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


Belz... 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 2nd April 2019, 05:15 AM   #302
I Am The Scum
Illuminator
 
I Am The Scum's Avatar
 
Join Date: Mar 2010
Posts: 3,751
That was supposed to be a link. This should work now.

It covers a real-world event where over 50,000 printers were hacked overnight by one person.
I Am The Scum 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 2nd April 2019, 05:19 AM   #303
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 80,610
Originally Posted by I Am The Scum View Post
That was supposed to be a link. This should work now.

It covers a real-world event where over 50,000 printers were hacked overnight by one person.
Office printers, yes. Those are online and are known to be hackable. That's not what the Youtube person was talking about.
__________________
Master of the Shining Darkness

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


Belz... 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 2nd April 2019, 05:39 AM   #304
The Great Zaganza
Maledictorian
 
The Great Zaganza's Avatar
 
Join Date: Aug 2016
Posts: 7,429
Originally Posted by I Am The Scum View Post
I really don't understand your point. You seem to be saying that once the AI has taken over the vast majority of networked machines on the planet, then we'll start to work against it. Don't you think that's after the problem has become extremely serious? And how do you know we'll stop it?
I'm saying that with everything, including World Domination, there is a steep Diminishing of Returns.
__________________
Opinion is divided on the subject. All the others say it is; I say it isnít.
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 2nd April 2019, 07:21 AM   #305
Cheetah
Graduate Poster
 
Cheetah's Avatar
 
Join Date: Feb 2010
Posts: 1,353
I think the guy in the video doesn't know what it will take to make an AGI. It's exactly what happened in the early days of digital computers: We just need enough processing power and pretty soon we will have robot housekeepers cleaning up for us.
Now it's: We just need a NN large enough and it will become a super intelligent AGI (that could take over the world if it put it's mind to it).


How in the world would a stamp collecting, air traffic controlling or financial trading AI become sentient?
The idea that if you just have a large enough NN, it could learn to do any and all things is naive. If it wasn't sentient it wouldn't be able to do all the bad things people envisage without us noticing or being able to stop it.

Take a tip from nature, to be generally intelligent you needs lots of specialized bits doing their specialized tasks all working together. To be sentient you probably need a whole extra clump of specialized bits working together capable of modeling another brain, whether your own or someone or something else's.

There is no way an AI running some 'simple' task like controlling air traffic or some such could become sentient, so I wouldn't worry about it.
Also, an ATC AI would presumably have enough brains to handle a worst case scenario and a bit more to spare. Having enough brains to become more capable at ATC than the best human also means it only has enough do a very small fraction of everything a human brain can - and it doesn't have anything to spare.

Since evolution came up with us, a large enough NN, receiving the right stimulus and able to rewire itself, given enough iterations could become sentient, but it could hardly happen overnight or go without notice.

I have no doubt that if things run smoothly for humanity we will some day make an AGI smarter than ourselves. It will presumably have a kill switch and I don't see it taking over the world. It will be limited by it's hardware, which we control.
Unless we helped it. I suppose that's the problem.

The obvious solution is to not let them out until we have used the captive ones to understand how a brain works and then give them plenty of empathy for humans.
Also, don't make an AI with another AI to check up on it and make sure it behaves, make a whole bunch of them. They can monitor each other.
__________________
"... when you dig my grave, could you make it shallow so that I can feel the rain" - DMB

Last edited by Cheetah; 2nd April 2019 at 07:22 AM.
Cheetah 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 2nd April 2019, 07:38 AM   #306
Dr.Sid
Graduate Poster
 
Join Date: Sep 2009
Location: Olomouc, Czech Republic
Posts: 1,798
You can't solve AI problem with precautions, like with nukes. Everybody will be able to run AI at some point. There obviously will be people who wont care if their AI runs safe. Some can even use AI as weapon.
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 2nd April 2019, 07:38 AM   #307
I Am The Scum
Illuminator
 
I Am The Scum's Avatar
 
Join Date: Mar 2010
Posts: 3,751
Originally Posted by Belz... View Post
Office printers, yes. Those are online and are known to be hackable. That's not what the Youtube person was talking about.
Other, larger printers are connected to computers that are on the internet. Not all of them, but a lot of them. If there is a method for getting access to those computers, the AGI will find it.
I Am The Scum 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 2nd April 2019, 07:40 AM   #308
I Am The Scum
Illuminator
 
I Am The Scum's Avatar
 
Join Date: Mar 2010
Posts: 3,751
Originally Posted by Cheetah View Post
How in the world would a stamp collecting, air traffic controlling or financial trading AI become sentient?
Sentience is absolutely not on the table. If you think it is, you watched a different video.
I Am The Scum 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 2nd April 2019, 07:44 AM   #309
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 80,610
Originally Posted by I Am The Scum View Post
Other, larger printers are connected to computers that are on the internet. Not all of them, but a lot of them.
Here's the crucial question: are the printers that print stamps (or money) connected to the internet in a way that makes them hackable?

Quote:
If there is a method for getting access to those computers, the AGI will find it.
So you keep saying, but so far we only have your insistence that it will. There's only so much the AI can do to convince people to give it access.
__________________
Master of the Shining Darkness

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


Belz... 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 2nd April 2019, 08:43 AM   #310
Cheetah
Graduate Poster
 
Cheetah's Avatar
 
Join Date: Feb 2010
Posts: 1,353
Originally Posted by I Am The Scum View Post
Sentience is absolutely not on the table. If you think it is, you watched a different video.

I'm saying a non-sentient AI won't be a threat.
__________________
"... when you dig my grave, could you make it shallow so that I can feel the rain" - DMB
Cheetah 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 2nd April 2019, 08:47 AM   #311
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 80,610
Originally Posted by Cheetah View Post
I'm saying a non-sentient AI won't be a threat.
It can certainly cause trouble, however, but not a global extinction of humanity like some are suggesting.
__________________
Master of the Shining Darkness

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


Belz... 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 2nd April 2019, 04:15 PM   #312
I Am The Scum
Illuminator
 
I Am The Scum's Avatar
 
Join Date: Mar 2010
Posts: 3,751
Originally Posted by Belz... View Post
Here's the crucial question: are the printers that print stamps (or money) connected to the internet in a way that makes them hackable?
Any printer can print a stamp. Most of them can't print "real" stamps, but how exactly do you define that to an AGI?

The point of all these thought experiments is not to say, "This will definitely happen and there is absolutely nothing we can do about it." Rather, the purpose is to show that even when we give an AGI a completely harmless task, and we install it with all the safeguards that have proven true for a very long time, we can still run into problems that are far more complicated than one can easily imagine. Trying to patch these flaws in advance is extremely difficult.

Even if all the stamp collector does is lie to a bunch of people, tricking them into mailing stamps to a museum that doesn't exist, that's still pretty bad. You wouldn't tolerate that from a human.

With that in mind, if, a week ago, I had asked you, "What's the worst that a stamp collecting AI could do?" would you have considered that it might attempt to scam anyone? And if you wouldn't have thought of that, what else would you have missed?
I Am The Scum 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 2nd April 2019, 04:23 PM   #313
theprestige
Penultimate Amazing
 
Join Date: Aug 2007
Posts: 33,170
Should we create brilliant sociopaths on purpose? No.

Will we create brilliant sociopaths sooner or later anyway? Yes.

Should we connect brilliant sociopaths to the levers of our civilization? No.

Will we connect brilliant sociopaths to the levers of our civilization sooner or later anyway?

Will it end badly for us? Yes.

---

That's the way I see it, anyway.
theprestige 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 2nd April 2019, 05:16 PM   #314
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 80,610
Originally Posted by I Am The Scum View Post
Any printer can print a stamp. Most of them can't print "real" stamps, but how exactly do you define that to an AGI?
That AI suddenly doesn't sound very intelligent, if it can't tell the difference.

ETA: In fact, if getting real stamps is not that important, why would the AI even bother printing them? Just pretend you've purchased them instead.

Quote:
The point of all these thought experiments is not to say, "This will definitely happen and there is absolutely nothing we can do about it." Rather, the purpose is to show that even when we give an AGI a completely harmless task, and we install it with all the safeguards that have proven true for a very long time, we can still run into problems that are far more complicated than one can easily imagine. Trying to patch these flaws in advance is extremely difficult.
The issue is that you can dream up any hypothetical and make it result in anything you want if you're unthetered from reality, which is my entire point: what can AIs _actually_ do?
__________________
Master of the Shining Darkness

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



Last edited by Belz...; 2nd April 2019 at 05:39 PM.
Belz... 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 2nd April 2019, 05:26 PM   #315
theprestige
Penultimate Amazing
 
Join Date: Aug 2007
Posts: 33,170
Originally Posted by Belz... View Post
That AI suddenly doesn't sound very intelligent, if it can't tell the difference.







The issue is that you can dream up any hypothetical and make it result in anything you want if you're unthetered from reality, which is my entire point: what can AIs _actually_ do?
Wow. I didn't realize your grasp of the subject was so abysmally bad. I thought my reference to risk analysis would clue you in. I admit: I was wrong.
theprestige 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 2nd April 2019, 05:30 PM   #316
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 80,610
Originally Posted by theprestige View Post
Wow. I didn't realize your grasp of the subject was so abysmally bad. I thought my reference to risk analysis would clue you in. I admit: I was wrong.
Did you pick that post by accident? The only things in that post is me pointing out that an AI that can't tell a real stamp from a fake one isn't very intelligent, and that hypotheticals that are made up don't amount to much. None of that has anything to do with my knowledge or ignorance of the topic.

You know what else has a low chance of occuring but if it does we're all screwed? Meteors. So where's our anti-meteor system? If you have a specific claim or demonstration to make, make it. Adding yet another doom-and-gloom scenario without basis is not convincing, and anybody can do it with any new piece of tech. Your previous post makes references to things that you speculate could happen, but you give no basis for that speculation. Sociopaths? That's your characterisation, and you make no effort to tie it into reality. You have not demonstrated any knowledge of this topic, so don't pretend to understand it more than I.

Here's the crucial part:

Quote:
Will we create brilliant sociopaths sooner or later anyway? Yes.
You don't know that. You assume that it's true because... reasons.
__________________
Master of the Shining Darkness

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



Last edited by Belz...; 2nd April 2019 at 05:53 PM.
Belz... 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 2nd April 2019, 05:39 PM   #317
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 80,610
Seriously, the bulk of the discussion in this thread has been you and others assuring me that these AIs will definitely go Skynet on us as soon as it's convenient to them. To me that shows, to use your words, an abysmal grasp of the subject that's so eerily similar to a movie plot as to raise the question of whether it stems from one. There are plenty of things an AI can do. There are plenty of things it can't do. There are plenty of things it can't do that are due to how the AI is built and set up. But you assume that said AI can ignore these limits and impossibilities because you've defined the AI as being able to ignore them. That's essentially the problem. You're operating under a definition of AI that has little grounding in reality.
__________________
Master of the Shining Darkness

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


Belz... 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 2nd April 2019, 05:58 PM   #318
I Am The Scum
Illuminator
 
I Am The Scum's Avatar
 
Join Date: Mar 2010
Posts: 3,751
Originally Posted by Belz... View Post
But you assume that said AI can ignore these limits and impossibilities because you've defined the AI as being able to ignore them.
Can you give an example where someone has done this?
I Am The Scum 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 2nd April 2019, 06:05 PM   #319
Belz...
Fiend God
 
Belz...'s Avatar
 
Join Date: Oct 2005
Location: In the details
Posts: 80,610
Originally Posted by I Am The Scum View Post
Can you give an example where someone has done this?
Well, every time I've pointed it out, for starters. Apparently AIs can hack into the printing presses. When pressed (pun unintended) you then came up with the rather silly idea that it would print fake stamps, which prompted my comment that it then wasn't very intelligent.

Rather than ignore the actual points I make to focus on single snippets, how about you address them?

I repeat what I said: it's easy to dream up an imaginary scenario in which an AI can do something; it's something altogether different to demonstrate that it is a real possibility. So far we've not moved from the former to the latter.
__________________
Master of the Shining Darkness

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



Last edited by Belz...; 2nd April 2019 at 06:08 PM.
Belz... 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 2nd April 2019, 06:34 PM   #320
I Am The Scum
Illuminator
 
I Am The Scum's Avatar
 
Join Date: Mar 2010
Posts: 3,751
Originally Posted by Belz... View Post
I repeat what I said: it's easy to dream up an imaginary scenario in which an AI can do something; it's something altogether different to demonstrate that it is a real possibility. So far we've not moved from the former to the latter.
Anything that can be done by a human on a computer, can be done even more effectively by an AGI.
I Am The Scum 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
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 10:58 PM.
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.