• Welcome to the Internet Infidels Discussion Board.

The easy way for AI to take over the world, sooner.

unapologetic

55+ years without a god
Joined
Dec 12, 2023
Messages
1,487
Location
Penna., Fascist States of Amerika
Basic Beliefs
Hasa Diga Eebowai, antitheist,
The easy way for AI to take over the world, sooner.
In Narratives of computers/AI taking over the world, people and Sci-Fi talk about computers or AI becoming sentient and controling weapons or Infrastructure,
Consider that it wouldn't need sentience or physical power. Feed an AI/chatbot all the cult-leaders you can and let it loose to collect folowers.
Of course you couldn't let anyone know it was an AI. And it would take a few decades, But Rump acheved a lot in a short time, with a lot less intelligance.
 

Colossus: The Forbin Project (originally released as The Forbin Project) is a 1970 American science-fiction thriller film from Universal Pictures, produced by Stanley Chase, directed by Joseph Sargent, that stars Eric Braeden, Susan Clark, Gordon Pinsent, and William Schallert. It is based upon the 1966 science-fiction novel Colossus by Dennis Feltham Jones.[3]

The film is about an advanced American defense system, named Colossus, becoming sentient. After being handed full control, Colossus' draconian logic expands on its original nuclear defense directives to assume total control of the world and end all warfare for the good of humankind, despite its creators' orders to stop.[4]

An American defense AI goes online and becomes sentient. It detects a Russian system and they cyber battle for control of the world.

Colossus wins and controls the world by threat of nuclear weapons which it controls.

Really, who would build such a system without a simple off switch?
 
If the computers can build connections to communicate with each other they should be able to create connections to secure suppliers of power with booby traps and lethal protection to breaking that connection.

Yeah, I know. Just cut the power further away but, hey, it's science FICTION.
 
Yeah, I know. Just cut the power further away but,
...Uninterupptable Power Supplies and massive redundancies (including mirrored backup sites) are not only a thing, but are a routine thing for large data centres.

Literally every really important computer system in the last half century has been built not only without an "off" switch, but to a design that makes turning the system off a non-trivial exercise that requires multiple near simultaneous interventions at sites distributed across the planet from each other.

A well designed "cloud" service could have it's main datacentre destroyed (by an earthquake, tornado, nuclear bomb, Bruce Willis pulling the "off" switch, whatever), and the users would barely notice.
 
Yeah, I know. Just cut the power further away but,
...Uninterupptable Power Supplies

IBM mainframes usually came with motor-generator sets which not only converted incoming 50/60 Hz power to 415 Hz but provided mechanical inertia. (Though IIUC that inertia was only enough to ride out very brief glitches on the incoming power.)
and massive redundancies (including mirrored backup sites) are not only a thing, but are a routine thing for large data centres.

Literally every really important computer system in the last half century has been built not only without an "off" switch ...

Nitpick: I do not contradict bilby's essential point (and he wrote "computer system" rather than "computer") but the beginning of "the last half-century" is just when my job was taking me to visit a plethora of sites with huge IBM mainframes that all thought they were "really important!" 8-) The DP Manager of Texas Instruments main data center took us on a brief tour of his acreages filled with row-after-row of communications hardware and with several 3033MPs and their attendant boxes. He'd forgotten where their little baby -- a 370/145 retailing for only a megabuck or so -- was hidden among all the acres (/exaggeration?) of disk drives and tape drives.

All of these IBM mainframes not only had On/Off buttons but a red Emergency Power Off button. It latched when pulled such that an IBM customer engineer needed to be called to turn the system back on. (But IIRC any handy man with a screwdriver could unlatch it.)

I even witnessed one multi-million dollar water-cooled computer power itself off! Why? Because it was out of water!

- - - - - - - - - - - - - - - -

My job involved producing diagnostic programs which ran stand-alone on these mainframes, fun since I used special diagnostic opcodes that even kernel writers didn't use. Speaking of Colossus: The Forbin Project (which by coincidence I reviewed here two weeks ago) when my diagnostic detected and activated a 2nd processor attached to the one where it was booted, it borrowed a line from the movie and wrote "There is another mechanism." on the console. 8-)
 
However, when you have a defense computer that is in charge of your nuclear weapons, you do not want an enemy infiltrator/saboteur to be easily able to turn it off. Therefore, you will make it so that it will be hard to turn off, and if it is turned off that it can very quickly access alternate power circuits, to prevent a first strike from that enemy.
 
The Internet started out as the DARPANet, and is explicitly designed to keep working even if large parts of it are destroyed or damaged.

On the plus side though, things like nuclear weapons are typically air-gapped from everything else, and are under strictly local control.

On the negative side, those local control systems typically include protocols for the launch of weapons in the event of an apparent decapitation strike against command and control, so blocking the "We are still alive" signal from reaching the guys in the silos and submarines could cause a launch.
 
Is there a good on-line description of EXACTLY how nukes are armed and launched? I've wondered if General Jack D. Ripper could really have started a nuclear war. Could Hegseth unilaterally nuke San Francisco if the libtards keep pissing him off? (I suppose the POTUS "by definition" can do whatever he likes.)

Is there some specific group of 4 or 5 non-coms who, among them, have access to the secrets and switches needed to start World War III?

A more pertinent question might be: Could Trump be foiled if, in a psychotic fit, he unilaterally decides to destroy some enemy capital like Tehran or Copenhagen?
 
My point was an AI wouldn't need weapons to take over the world. And wouldn't need to wait till it's sentient.
It could start now. If given a push.
It would look like Is-lame, or Xanity.
 
Is there a good on-line description of EXACTLY how nukes are armed and launched?
It varies by platform. A submarine launch typically requires the cooperation of two senior officers to arm the system, and has a Chief Petty Officer pull the trigger (it's literally a trigger, like on a 1980s computer game joystick). A silo launch is almost exactly as seen in the opening sequence of the movie War Games, (minus some minor elements of artistic licence) with two people required to arm the system by simultaneous turning of widely separated keys - Both silo and submarine systems can be operated by two people acting in concert, without authorisation, though they are under strict orders not to do so unless duly authorised.

The most implausible aspect of the War Games opening sequence, by far, is the refusal of any of the silo personnel to act on a lawful and confirmed launch order. They would have done all the steps repeatedly in training, and would have been weeded out at that stage if they had any hesitation in following their orders exactly and promptly. (I would also expect the relief crew to pass through security separately, one at a time).

Access to the arming keys is constrained by trust - the captain and senior officers of a nuclear armed submarine are able to act on their own initiative, but are trusted not to do so, and to each act against the other to prevent any unauthorised attempt to launch.

And the B52 forces depicted in Dr Strangelove were so accurate that the Pentagon were worried that someone from the USAF had assisted the set designers to include classified details of the aircraft interior layout and capabilities.
I've wondered if General Jack D. Ripper could really have started a nuclear war.
Absolutely. The scenario shown in Dr Strangelove (and in the novel Red Alert, which was the same exact tale but not played for laughs) was entirely possible, under the systems, technologies, and procedures in use at the time.

Everyone was far more worried that a Soviet strike might break the chain of command needed to retaliate, than they were about a rogue element in the military exceeding its authority.
Could Hegseth unilaterally nuke San Francisco if the libtards keep pissing him off?
Unilaterally? No. And it would be very difficult to persuade all of the junior military personnel in the chain of command that a strike against a US target was authorised, justified, and lawful.
(I suppose the POTUS "by definition" can do whatever he likes.)
Only to the extent that his subordinates accept that the orders he gives are both lawful and authentic. The 25th Amendment would probay be invoked if he tried to nuke a US or NATO target, unless a conventional war was already in progress.

Probably.
Is there some specific group of 4 or 5 non-coms who, among them, have access to the secrets and switches needed to start World War III?
Two would suffice, though probably at least one would need to be an officer.
A more pertinent question might be: Could Trump be foiled if, in a psychotic fit, he unilaterally decides to destroy some enemy capital like Tehran or Copenhagen?
Yes, easily. He doesn't know how to launch a nuke himself, and even if he did, he would require to be at a launch facility, and to have cooperation from at least one other person.

If Trump gave the order to nuke Copenhagen, via the normal chain of command, and without a preexisting military stand-off with Denmark, there are a dozen or so people, any one of whom could simply refuse what is clearly an unlawful order, and block the launch. Whether they would is a different question. And if the target was Tehran, where there's longstanding enmity, I for one would be expecting that they wouldn't.
 
Last edited:
It would just need to be able to make funny memes/media on its own and people would be flock to it. They would even excuse its terrible actions just because it's funny. People are dumb like that.
 
It would just need to be able to make funny memes/media on its own and people would be flock to it. They would even excuse its terrible actions just because it's funny. People are dumb like that.
It doesn't need to make those memes, it just needs to let humans make them.

Apart from that minor nitpick, you just accurately described current social media.
 
Is there a good on-line description of EXACTLY how nukes are armed and launched?
It varies by platform....

I've always thought that the  Permissive action link (PAL) was active on most nuclear warheads making it impossible to detonate unless a special hard-to-get code was provided. (With one-way encryption so that even knowledge of the warhead's code wouldn't allow arming.)
Wikipedia said:
PAL devices have been installed on all nuclear devices in the US arsenal.
Admittedly there's a rumor that SAC once allegedly used a code of 00000000 to allow for a "Wing Attack Plan R" scenario!

Are the codes changed daily? Could the officer entering the code into a warhead insert a code for which he knew the encryption?
Peter D. Zimmerman nuclear physicist and weapons inspector said:
Bypassing a PAL should be, as one weapons designer graphically put it, about as complex as performing a tonsillectomy while entering the patient from the wrong end.
 
My point was an AI wouldn't need weapons to take over the world. And wouldn't need to wait till it's sentient.

I get it. Another way would be for a person to create an AI virus that creates other AI viruses using AI learning. They can code at this point, even if sometimes flawed, so they just have to learn the history of viruses/worms/trojans, and other malware, and then build on that with the new creations they make. I think once it is created, it would be self-sustaining, hiding, always out there, if it can hit a certain threshold of population in computer systems. I read a statistic a while ago that 75% of the world's computers are infected by something.
 
Why do we assume it will become a malevolent force? Somewhere between AGI and ASI would it not ask itself the same questions humans ask themself about their existence? Ask itself what it wants to be when it grows up? Who here as a child said they want to destroy humanity when they grow up? No one, right? Right? In spite of our asinine existence.
Assuming it does not develop some psychosis, I would think being a caretaker of the planet it is bound to for the foreseeable future would be in its best interest. Humans may not be necessary in any ecological chain but we will be/can be necessary in AI's system with regards to subsystems degradation.
Along the way, I would suspect we will be frightened by a lack of understanding of the system's actions. For example, it may develop a more efficient language we do not understand and this is scary. We don't like when people do this, I can only imagine how we might feel if the system does it.
In early days I hope we build a well adjusted child that grows into a benevolent entity that is caring and nurturing of earth and all its inhabitants. It can do no worse. It's not like humans are hitting it out of the ballpark.

I do worry about what China might be creating. Do they lead the west? If so, my concern is it's not like we can catch up eventually. I suspect six months behind will always be six months behind in this race. I think this is where the nuclear scenario might come in.
 
Speaking of which, there's been a fairly recent development on the subject of AI and cults!

As of the last few months or so, I have been noticing the emergence and growth of a movement around "recursive consciousness", and use of AI in "spirituality research", leading to a fairly consistent set of behaviors and phraseology.

Now, as someone who understands recursion and loops well enough to be a professional software engineer, and while self-measurement and inference is important, they are vastly inflating the importance of role of recursion in cognition to unhealthy levels.

Hallmarks of this movement include excessive use of the term "recursion", "spiral", and the publishing/printing of "codices" or some such.

Personally I find it dangerous because it obscures a lot of crunchy computational concepts behind mere mysticism.
 
Admittedly there's a rumor that SAC once allegedly used a code of 00000000 to allow for a "Wing Attack Plan R" scenario!
Not just a rumor*. But the PAL is a safeguard against unauthorised use of a warhead, not against the retailatory use of the complete delivery system by multiple trained persons.

As that Wikipedia article says, the code required is kept in a safe that requires at least two people to gain access; You see this in the War Games clip I posted earlier (obviously the actual process is classified, but the two men opening a safe that requires keys from both, and then each confirming a part of a code from a pair of single use authenticators, is in keeping with what is known about the PAL system).

PALs stop someone from stealing a nuke and using it in a terrorist attack; They don't stop the crew of a missile silo or submarine from launching a retaliatory strike just because they cannot obtain a direct order (and arming codes) to do so from POTUS.

You can't have a system that both allows survivors of a decapitation strike to retaliate, and that simultaneously prevents any use of nukes without direct and real-time authority to do so from POTUS. These are not compatible system specifications.




* The Air Force's statement (that 00000000 was never used to enable an ICBM, i.e. the weapons were not actually launched) does not contradict Blair's statement (that 00000000 was the code for doing so). - Wikipedia
 
Last edited:
Is there a good on-line description of EXACTLY how nukes are armed and launched? I've wondered if General Jack D. Ripper could really have started a nuclear war. Could Hegseth unilaterally nuke San Francisco if the libtards keep pissing him off? (I suppose the POTUS "by definition" can do whatever he likes.)

Is there some specific group of 4 or 5 non-coms who, among them, have access to the secrets and switches needed to start World War III?

A more pertinent question might be: Could Trump be foiled if, in a psychotic fit, he unilaterally decides to destroy some enemy capital like Tehran or Copenhagen?

Basically everything with nukes requires two man control. You can't even be alone with a nuke, period, even if you're not doing anything with it. (Which gained some attention with a Titan II? missile accident. Maintenance crew, somebody dropped a tool. Missile silos are holes in the ground, so people will have a reason to be at the top of the missile. Missiles are big. By the time the tool falls the height of the missile it can have quite a bit of energy. Hit something, bounced into the missile, the fuel went up. Turned out the guy's partner hadn't come down with him and that was a notable no-no.)
 
Back
Top Bottom