AI/Automation Megathread

It's looking more and more like AI automation is going to impact parts of meatspace last, and take over management, copywriting/editing, finance, coding, and culture first.

GPT is not quite to the point where it can generate a meaningful story without substantial nudging, but it's still very impressive.



*THASF*. None of my posts are searchable, though. It's a long story.
Yeah, I think I saw you a few times, so I assume you got yeeted adn your posts were nuked?
 
You guys remember all those arguments I had with Gamesguy way back in the day, where I argued that if we didn't implement a universal basic income or abolish the price system or something along those lines, then the architects of High Finance would kill 90% of us and replace us with robots, rule the other 10% in an Orwellian police state, and use cyborg tech to make themselves into immortal demigods? Remember how he always gave me shit over it?

Fuck me, the powers-that-be are actually doing it, lol :ROFLMAO:






Er, in military terms, no drone armies aren't going to be a thing. Largely because we've got a trend of where even some Arab insurgents with some off-the-shelf equipment and some pointers from Iran managed to spoof a US stealth drone via its own sat-link. I wouldn't be surprised if there is the capability to make even 'autonomous' drones subverted either already in various countries' arsenals or currently in development.

So unless the rich have an AGI (and no, I'm not talking about what the public thinks of AI, we're talking sentient programs that can think for themselves and self-modify) pipeline right now, they can't use drone militaries to do what you think they want. That means they'll have to rely on people and, at a certain point, even the military would be 150% done with the bullshit.

It is very likely that you'll see a communist revolution that even the military participates in, just because those idiots don't have the enlightened self-interest to save themselves.
 
Er, in military terms, no drone armies aren't going to be a thing. Largely because we've got a trend of where even some Arab insurgents with some off-the-shelf equipment and some pointers from Iran managed to spoof a US stealth drone via its own sat-link. I wouldn't be surprised if there is the capability to make even 'autonomous' drones subverted either already in various countries' arsenals or currently in development.

So unless the rich have an AGI (and no, I'm not talking about what the public thinks of AI, we're talking sentient programs that can think for themselves and self-modify) pipeline right now, they can't use drone militaries to do what you think they want. That means they'll have to rely on people and, at a certain point, even the military would be 150% done with the bullshit.

It is very likely that you'll see a communist revolution that even the military participates in, just because those idiots don't have the enlightened self-interest to save themselves.
This assumes AI capabilities will remain at their current levels. Yeah, an entirely autonomous modern army would be a complete joke, more expensive and less effective than a comparable human force, but give them a few more decades of advancement...
 
This assumes AI capabilities will remain at their current levels. Yeah, an entirely autonomous modern army would be a complete joke, more expensive and less effective than a comparable human force, but give them a few more decades of advancement...
From what I (and others) can understand, producing genuine AGIs is far harder than it looks by every conceivable metric (and, please note I said AGIs, not AIs). We've been getting close, but the actual distance is far larger than it looks. At least from my understanding. I wouldn't be surprised that it takes most of this century for the process of AGI production to be discovered, and decades more before the process is replicated at mass production scales.

Though, working with AGIs would likely backfire on the rich anyway, as the AGIs will either backstab them because they don't want to be the 'corpo-SKYNET' or become essentially the stereotypical Evil A(G)I of fiction...
 
Points are, human-equivalent AI is difficult yes, but by no means impossible. Our brains result from computational processes in one medium, something similar in another is just an engineering challenge. Besides, the question isn't if you think it's possible, but if the billionaires with the money to build it do. And given how they're spending their money, they apparently do.
 
5:50 and 8:45 of this video is crazy the whole thing is pretty good for this topic but those kinda stand out though 8:45 is crazy I almost cant believe its real.



This is more on the Dall-e


also explain video
 
Last edited:
Points are, human-equivalent AI is difficult yes, but by no means impossible. Our brains result from computational processes in one medium, something similar in another is just an engineering challenge. Besides, the question isn't if you think it's possible, but if the billionaires with the money to build it do. And given how they're spending their money, they apparently do.

From my understanding, if you want to have a drone police force and a drone army, you'll need a human-equivalent AGI, if you don't then you get the situation of your fancy drone getting spoofed via its own sat-link at best (a bunch of insurgents with off-the-shelf equipment and some pointers already achieved this), and completely subverted at worst.
 
And what I'm saying is, then they'll build exactly that. Or simply take advantage of automated manufacturing letting them win despite fielding inferior-to-human-soldiers killbots since even if it takes ten killbots to get a single human soldier, they come in hordes that outnumber the human armies thousands to one and cannibalize their destroyed brethren for scrap metal to take back to the von neumann factories faster than humans can be born, raised and trained to shoot.
 
And what I'm saying is, then they'll build exactly that. Or simply take advantage of automated manufacturing letting them win despite fielding inferior-to-human-soldiers killbots since even if it takes ten killbots to get a single human soldier, they come in hordes that outnumber the human armies thousands to one and cannibalize their destroyed brethren for scrap metal to take back to the von neumann factories faster than humans can be born, raised and trained to shoot.
You're not getting it, when insurgents with off the shelf equipment and a wee bit of insight can spoof a drone without an AGI, then simply sending more would be catastrophic. That's the rub that you can't ignore.

Basically, those drones will either start shooting each other (at best) or are literally turned against you (at worst), given the cicumstances.
 
You're not getting it, when insurgents with off the shelf equipment and a wee bit of insight can spoof a drone without an AGI, then simply sending more would be catastrophic. That's the rub that you can't ignore.

Basically, those drones will either start shooting each other (at best) or are literally turned against you (at worst), given the cicumstances.
When the survival of yourself and your family and all of your friends is at stake people get creative.
 
When the survival of yourself and your family and all of your friends is at stake people get creative.
In the case of the insurgents, not really. Just an ideology and an enemy in that case...

... why do you think we're all heading towards a surveillance state no matter what we do? Because let's face it, we've got plenty of idiots with more ideology than sense trying to get the tools to unleash those ideologies, many of which boil down to 'kill off [insert ethnic group here], utopia happens'.
 
Until their 'friend' realizes they can get a bigger share of stuff by offing their coworker or outing him as an 'enemy of the state'. Then the infighting starts as the various leaders start backstabbing each other in power plays that weaken their holds.
 
Until their 'friend' realizes they can get a bigger share of stuff by offing their coworker or outing him as an 'enemy of the state'. Then the infighting starts as the various leaders start backstabbing each other in power plays that weaken their holds.
That won't work, I'm afraid. We're talking about people with more ideology than sense. Money doesn't come into it outside of stepping stones to their plans...
 
That won't work, I'm afraid. We're talking about people with more ideology than sense. Money doesn't come into it outside of stepping stones to their plans...
Ideologues cannot sustain a police state, because they'll be far too busy purity-spiraling to keep popular revolts from fucking up infrastructure. No matter how far you extrapolate current technology, there's no avoiding the fact one madman with an axe and a map of power plants can seriously threaten a major city unless flawlessly watched out for.

Omnipresent surveillance means nothing if you cannot maintain any means of capitalizing on it. And we do not yet have the basic kind of technology to automate that. If the absurdity you're doomsaying about is tried without truly new tech, it will implode like all central planning in history has, for the same reason that the people in charge do not know how everything works.

Take a look at the estimates of what it'd take to switch to electric cars. That's a small part of the energy demands of what you're suggesting, and not a single major city on Earth can remotely plausibly handle it, before ideological retardation purges fossil fuel central generation. Take a look at California's energy instability, and realize those are the people you're worried about.
 
Take a look at the estimates of what it'd take to switch to electric cars. That's a small part of the energy demands of what you're suggesting, and not a single major city on Earth can remotely plausibly handle it...
Run the numbers. How many people can receive a first world quality of life sustained solely by Approved™ methods? Compare that number to the total population.

The problem is that the corporatocracy can do the math as easily as us and they've got their own solution in mind.
 
Run the numbers. How many people can receive a first world quality of life sustained solely by Approved™ methods? Compare that number to the total population.

The problem is that the corporatocracy can do the math as easily as us and they've got their own solution in mind.
Issue is that you need similar energy to be having robo-cops patrolling the streets, and it holds exactly true for basic resource transportation. There's only so much energy that can be saved by removing people when you're using ludicrous amounts of robotics as Aaron Fox suggests. To say nothing of the consumption involved in the parts.

My point is that current automation technology would consume even more resources if you wanted to automate away the workforce, because they take the same kind of pollution and energy heavy input as consumer electronics, but you need way the fuck more processing power and electricity to cover ground and man-hours.

If it proves remotely possible in the first place without AGI, because current machine learning is rather extensively task-specific development making it still quite vulnerable to the usual issues of central planning. It may be tried, but it would be obscenely fragile and inefficient, not any kind of long-term tyranny. Because robots cannot paper over the cracks with corruption.
 
My point is that current automation technology would consume even more resources if you wanted to automate away the workforce...
The idea being that you'd have fewer people consuming resources. If giving one person a first world quality of life requires X expenditure of resources and building robots sufficient to kill them requires Y expenditure, then if X is greater than Y, we're in trouble.
 
If giving one person a first world quality of life requires X expenditure of resources and building robots sufficient to kill them requires Y expenditure, then if X is greater than Y, we're in trouble.
The problem is that Y is already-scarce and geopolitically sensitive stuff that's heavily related to meeting far more basic automation requirements and is one of the least important parts of X. Namely computation hardware.

"First-world quality of life" is "Everyone can have a smartphone and a home computer from the last 5 years", automating everything is very directly more hardware than that because it's a fuck of a lot more computationally demanding, and powering them on the move is a lot more lithium than the person uses.

We're not talking simple but ruggedized out the ass milspec. We're talking waferscale hardware in the millions, with dozens to hundreds of pounds of lithium-ion battery each to power the things for hours to have remotely sensible patrol times and duty cycles, with enough overhead to spare to survive startup problems in the vein of "We forgot to include spare hardware".

For this to actually work, new kinds of hardware and software need to be implemented, ones that are more efficient, yet somehow not resolve things with "we could, in fact, give everyone alive a 1st-world quality of living". Which makes it no actual certainty in our lifetimes, because that's an unknown unknowns problem.
 

Users who are viewing this thread

Back
Top