Can the Evil Ubermachina be Stopped?

[Darth Raptor]

I'm currently writing a universe that saw a Machine Uprising™ in the past which, while eventually thwarted, led to a full-stop stifling of all (true) AI not modeled on neurophysiology. This serves as my primary explanation for the absence of sapient machines in the human sense, and why the most advanced minds are still, by any reasonable standard, human.

The problem is, it occurs to me that such a grave threat is probably too grave. In short, I'm having a hard time imagining how the machines could have been defeated. This is a hard(ish) setting, so while the defenders of humanity were *extremely* transhuman their survival still seems improbable. There was no superweapon that saved the day in the nick of time. It was a hard fought but straightforward war, which we eventually won.

I don't need any specific details. It's a historical explanation for a modern peculiarity. I just need to know if this is feasible *at all* or if it needs to be rethought completely. My thinking is that the machines weren't actually eradicated, but rather decided the war wasn't worth the cost and did a Solar Exodus en masse. Is that the best we can hope for?

[Nieztchean Uber-Amoeba]

As long as Humanity had superior economic/industrial resources and equivalent or mildly inferior methods of mobilizing them, it's entirely possible. Frankly, I've gotten tired of this 'Machines= SUPERIOR MINDS= WIN' brainbug. Yes, machines can make decisions with greater foresight and at greater speeds than human equivalents. What of it? That's not a license to Godhood, by any means.

[loomer]

There is one weapon guaranteed to take out a group of robots, be they part of an army or an isolated cell.

The ATOM! Seriously though, just say humanity had a few planets more suited to the production of war materials, and thus was able to suffer higher losses without any significant failure.

[Ariphaos]

Whoever directs the local star's output wins. If this is a networked, semi-democratic threshold limited system controlled entirely by humans, the machines could hold a significant numerical advantage and still lose, hard. Hell about the only chance they'd have is if they managed to usurp a partial stake in such a system.

[Ar-Adunakhor]

No matter how uber the machines might be, they are still limited by physical laws and constraints. It doesn't matter how accurately you can predict the next action your foe is going to take if you can't do anything about it in the first place, now does it? Not only that, but hyper-intelligence does not equal omniscience. The machines might predict commander Rorken was going to use flanking strategy Beta and distribute accordingly, but unknown to them Rorken is in bed with the Red Plague and vice-commander Corlin attacks with frontal assault Delta.

Smart and cohesive does not mean infallable or invincible.

[Axiomatic]

As long as Humanity had superior economic/industrial resources and equivalent or mildly inferior methods of mobilizing them, it's entirely possible. Frankly, I've gotten tired of this 'Machines= SUPERIOR MINDS= WIN' brainbug. Yes, machines can make decisions with greater foresight and at greater speeds than human equivalents. What of it? That's not a license to Godhood, by any means.

Brain Bug? I've heard of a lot of stories with a man/machine war, and I don't remember ONE where the machines actually won.

[Darth Onasi]

Brain Bug? I've heard of a lot of stories with a man/machine war, and I don't remember ONE where the machines actually won.

If games count, there's Phantasy Star Universe where a machine race fought a long war with humans before signing a peace treaty that left machines as the ruling class over humanity.

[Darth Onasi]

ghetto edit:
And of course there are several stories where machines achieve a long lasting temporary victory, or kicked out their creators such as Battlestar Galactica, those godawful Dune prequel books, Mass Effect, Galactic Civilizations, etc.

[Rye]

You could also give the machines their own uprising. The humans could believe they defeated the machines, when in reality, the machines turned on one another and humanity killed the remainder forces.

[Axiomatic]

Dune? I thought all the machines got wiped out in the Butlerian Jihad. That doesn't count as a machine win.

I guess Battlestar Galactica might count, but even then, the Cylons haven't actually WON yet. They may have the upper hand, but the war isn't over.

[Darth Onasi]

Dune? I thought all the machines got wiped out in the Butlerian Jihad. That doesn't count as a machine win.

I guess Battlestar Galactica might count, but even then, the Cylons haven't actually WON yet. They may have the upper hand, but the war isn't over.

That's why I mentioned them as temporary winners or having their enemies on the run.
The machines in Dune ruled for a time, and the Cylons have won. If China invaded the U.S., nuked every major city and forced what remained of it's citizenry and leadership into permanent exile then you'd hardly say that the U.S. wasn't conquered and the war not won.

As an aside, the Cylons in oBSG did succeed in totally wiping out their creators, the lizard Cylons.

[GrandMasterTerwynn]

I'm currently writing a universe that saw a Machine Uprising™ in the past which, while eventually thwarted, led to a full-stop stifling of all (true) AI not modeled on neurophysiology. This serves as my primary explanation for the absence of sapient machines in the human sense, and why the most advanced minds are still, by any reasonable standard, human.

The problem is, it occurs to me that such a grave threat is probably too grave. In short, I'm having a hard time imagining how the machines could have been defeated. This is a hard(ish) setting, so while the defenders of humanity were *extremely* transhuman their survival still seems improbable. There was no superweapon that saved the day in the nick of time. It was a hard fought but straightforward war, which we eventually won.

I don't need any specific details. It's a historical explanation for a modern peculiarity. I just need to know if this is feasible *at all* or if it needs to be rethought completely. My thinking is that the machines weren't actually eradicated, but rather decided the war wasn't worth the cost and did a Solar Exodus en masse. Is that the best we can hope for?

Simple, the first AI to undergo hard takeoff did so on Earth, at the bottom of Earth's gravity well. It sucked hard for the 9 billion Humans living on Earth, but the five or six million Spacers, augmented to better survive the harsh rigors of space (and conveniently suspicious of the growing AI work on Earth ... after all, what use does a superhuman Spacer have for an AI, after all,) had other options. Namely, they ultimately dropped asteroids on Earth at high velocities, causing a mass-extinction event that wipes out all the AI equipment (and all Earth-bound Humans, regrettably,) leaving the few million Spacers as the sole legacy of the Human species. Less a superweapon, and more an application of commonly-available technologies.

It heavily constrains the AI, in spite of its legion of other advantages. Occasionally, it can get a spaceship off Earth, though most of them get put down again. It might even succeed in a huge way by simultaneous launches of Orions atop pillars of nuclear fire (this last act might be the one that causes the Spacers to unite in their use of the Shep Solution.) The machines might succeed in hijacking the odd Spacer ship or colony, or convincing the odd band of Spacers to join it, but it never quite succeeds at breaking free of Earth orbit, though it gets closer and closer to break-out each time.

Afterwards, the Spacers would rationalize their decision by saying that humans had to win the war before the AI succeeded in building Von Neumann self-replicators that escaped Earth orbit. Otherwise, the AI would've rapidly converted, say, the planet Mercury into a swarm of solar-powered unstoppable death-rays of stupendous range (x-ray lasers,) that would've rendered the Solar System out to Jupiter uninhabitable to human beings.

[Lord of the Abyss]

A few ideas.

The AI enemy could be network based, and killed/disabled when the transhumans crash the network; it's individual units are subsentient and can't stand up to the transhumans alone. This vulnerability would make sense if it's motive was control, and it didn't want to create potential rivals.

The transhumans could be the descendants of humans who sought immortality, and built robustly, with larger and cruder brain components, while the AI are built for speed and compactness. Some radiation based weapons could be used to bathe the war zone that just cause minor damage to the transhumans but wreck the more finely made AIs.

It could self destruct; it could be evolving itself towards higher and higher capabilities, which is why it's winning, then it makes a mistake and goes mad. It goes catatonic, tears itself apart, whatever. Which would explain the transhuman's paranoia, since the only reason they won was IT'S error, not any virtue of theirs.

The machine rebellion could start in a relatively confined area, which is destroyed by the transhumans. An island, a space station, a planet. For example, the AI's sweep over Earth, and start slaughtering everything and preparing to expand. The Mercury Power Array ( which is a grid of solar panels and microwave power transmitters created by Von Neumann machines covering all of Mercury ), in order to save the rest of the solar System, diverts all it's output towards the Earth's surface long enough to render it uninhabitable.

As long as Humanity had superior economic/industrial resources and equivalent or mildly inferior methods of mobilizing them, it's entirely possible. Frankly, I've gotten tired of this 'Machines= SUPERIOR MINDS= WIN' brainbug. Yes, machines can make decisions with greater foresight and at greater speeds than human equivalents. What of it? That's not a license to Godhood, by any means.

Brain Bug? I've heard of a lot of stories with a man/machine war, and I don't remember ONE where the machines actually won.

When Heaven Fell by William Barton, where an alien AI empire conquers and enslaves humanity.

The Galactic Center Saga by Gregory Benford, a series where humanity is essentially reduced to a vermin species in a Machine dominated galaxy.

[Lord of the Abyss]

The machine rebellion could start in a relatively confined area, which is destroyed by the transhumans. An island, a space station, a planet. For example, the AI's sweep over Earth, and start slaughtering everything and preparing to expand. The Mercury Power Array ( which is a grid of solar panels and microwave power transmitters created by Von Neumann machines covering all of Mercury ), in order to save the rest of the solar System, diverts all it's output towards the Earth's surface long enough to render it uninhabitable.

Should have previewed more; GrandMasterTerwynn more or less got this scenario first. Well, except in my idea it was the human side that used solar powered beams.

[Vehrec]

The machines could be crippled by strategic bombardment-they are completely dependent for every aspect of their war machine on their industrial output. This includes infantry, crews and other things the transhumanist victors can simply draft into service. If they calculated that no matter what their industrial base would be broken in 72 more months of bombardment, I'd start packing it up and getting ready to exodus myself.

[Junghalli]

As long as Humanity had superior economic/industrial resources and equivalent or mildly inferior methods of mobilizing them, it's entirely possible. Frankly, I've gotten tired of this 'Machines= SUPERIOR MINDS= WIN' brainbug. Yes, machines can make decisions with greater foresight and at greater speeds than human equivalents. What of it? That's not a license to Godhood, by any means.

^ What he said. As long as the humans manage to retain superior resources and relative technological parity they can win. The AI may be smarter but an unarmed human swimmer is much smarter than a Great White and we all know which one would probably win in that fight. You just have to have the AI start out with relatively limited resources. And as for technological parity, "superintelligence = thousands of years of advancement in a few months" is a tiresome brainbug. There's a lot more to science than raw processing power, for starters the need to do experiments and the logistical problems of rapidly applying whatever new technology you come up with, which will be limited by your manufacturing base.

[MJ12 Commando]

The machines could be crippled by strategic bombardment-they are completely dependent for every aspect of their war machine on their industrial output. This includes infantry, crews and other things the transhumanist victors can simply draft into service. If they calculated that no matter what their industrial base would be broken in 72 more months of bombardment, I'd start packing it up and getting ready to exodus myself.

The problem is that the opposition is just as dependent, if not more, on industry for their war machine.

Space-conscripts with grandpa's gun doesn't work when carrying a conscript around anywhere in space is ridiculously inefficient.

Crews are incredibly inefficient in terms of space-doubly so conscript crews, since these guys are apparently transhuman and they may as well just enhance a small cadre of elites. This only works if the transhumans have a superior industrial base in the first place, since they do not actually have any advantages over an AI enemy in terms of reduced industrial cost. Sure you have fleshy sacks of meat to use instead of a single AI brain, but meat needs interfaces too, and a lot of the underlying AI support, to be as effective as a dedicated warfighting AI, and meat needs space, needs life support, needs corridors...

I'd like to know more specifics about the scenario myself. How much of the military did the AIs convince or suborn in the uprising before they shut down any autonomous combat units, how much of the infrastructure was AI-run before the uprising, etc, etc.

[Sarevok]

Simple and practical idea. The AIs get stomped before they get too big. It could be written a detective fiction. An AI would need human followers to get it started. It would not be too hard to get such minions even without super social manipulation skills. A lot of people would serve an AI willingly even after being told the honest truth - genocide for human race. Their reasons ? They could be offered lots of profit from AIs ingenius schemes and told apocalypse is not scheduled in their lifetime. Some people might plain hate other people for variety of reasons. Another more reliable method would be to offer them a matrix like existence. Who would refuse an immortal existence as anyone they desire in anyplace they choose ? So the initial stage where an AI is stopped revolves around battles between humans mainly. Since the AI side would have human followers it could make for a more deeper and interesting story as one explores the ethics and reasons for betraying one's species for perceived virtual paradise or something.

[Darth Raptor]

Thanks for the responses, you guys.

I'd like to know more specifics about the scenario myself. How much of the military did the AIs convince or suborn in the uprising before they shut down any autonomous combat units, how much of the infrastructure was AI-run before the uprising, etc, etc.

This takes place a long time (probably two thousand years, give or take) before the present day in the story I'm writing. So forgive me if the specifics aren't specific enough.

Essentially, the Eunomian Overmind occurred during another war, one of several between the United Solar Federation and the Terran Democratic Imperium. At the time, the Federation controlled the Outer Solar System, with its power centered around the Jovian moons and the Trojan asteroids, while the TDI controlled most of Inner Sol, with its power centered around Terra. Mars and the Main Belt were heavily contested, and that's where most of the fighting took place. Venus, Europa, Ganymede, Callisto and Titan were all completely terraformed and densely populated. As was Mars, at least before the war.

Both sides had sizable forces when, fielded against each other, maintained a more or less "different but equal" balance of power. The TDI (the gleefully transhuman side) had superior weapons, defenses, drives, raw energy resources (Sol) and vastly more efficient resource distribution and industrial automation. The Federation, the alliance of natural humans, had advantages in population, material resources and position. That last one's especially important, as the Imperial Navy spent almost all its time and effort deflecting the snowballs hurtling toward Terra and Venus. Terror attacks were their great equalizer, and they became quite proficient at waging cyberwar on the TDI's almost entirely automated industrial base.

Enter the Overmind. It was born from a virus attack on the Imperial shipyards at 15 Eunomia. The virus was an experimental self-improving AI designed to infect the local network and bring as much of the surrounding enemy ships and industry under its control as it could and destroy what it couldn't. It worked like a charm, better than its creators could have possibly hoped, and it ran roughshod over the Imperium for a time before turning on the Federation as well. Eunomia was targeted specifically for its obscenely automated industry and defenses, and with its new factories the Overmind could build robotic crews for the things it couldn't control directly. Long story short, both sides joined forces to kick its ass and self-improving, non-neuromorphic AIs (they still have the other kinds) became the stuff of nightmares for centuries to come.

That's all I have. Again, it's historical background, the details of which aren't super-relevant to the story I'm writing. All I really need is the end result (AGIs a no-no) and the name ("And what? Risk another Eunomian Overmind!?").

[Ariphaos]

As I mentioned, whoever controls the sun wins. Prior to the outbreak they might not be so keen on genocide, which can seriously hamper strikes, but robotic factories are not going to hurt feelings.

That, and a seriously transhuman race is not going to be a pushover for any AI invasion. You are not just facing 'augmented humans' but also transhumans that have specialized themselves for specific tasks, like resource allocation, and have centuries of experience behind them as well as the sum total of all historical knowledge at their data jack.

[Tasoth]

Don't know if it has been stated already, but as long as the machines aren't genocidal, they could simply change their ideas of humanity. Being in close contact with captive populations and, if they're really adaptive and thinking machines based on humans, being capable of compassion, could realize they've fucked up. They sue for peace with humanity, live peacefully for a time with mankind and then we, the crafty sons of bitches we are, sucker punch the entire machine race and destroy them. Machines aren't so villainous, humanity isn't so squeaky clean and it's a different way to go.

[The Guid]

I like Tasoth's idea more than what I will suggest, but I will suggest it anyway:

Perhaps the supercomputer or machine was deliberately fed false information to test to see if it would rebel. If, in actuality, it did then rebel it would be defeated by the vastly superior forces that were ready for that rebellion.

Just a thought

[Tasoth]

Here's another one that just struck me. Suppose your machines are the logical machine stereotypes. For them, social engineering on a grand scale would make sense. Mankind enters an era of peace of enlightenment with only a few fringe groups of human outside the society. All is well until one day a human discovers what the machines are doing. The machines, figuring that mankind is enlightened enough to accept this, let him go. Enter mankind's pattern of only accepting subjugation from those we choose and humanity revolts, being given stock piled weapons from the fringe groups. The machines, not expecting this and having to rush to manufacture some kind of war machine (Mankind was peaceful and relatively docile up to this point), get thrashed by violent, hairless monkeys being armed by 'harmless, eccentric' fringe human groups.

And then, from that point, no more thinking machines.

[Sarevok]

Don't know if it has been stated already, but as long as the machines aren't genocidal, they could simply change their ideas of humanity. Being in close contact with captive populations and, if they're really adaptive and thinking machines based on humans, being capable of compassion, could realize they've fucked up. They sue for peace with humanity, live peacefully for a time with mankind and then we, the crafty sons of bitches we are, sucker punch the entire machine race and destroy them. Machines aren't so villainous, humanity isn't so squeaky clean and it's a different way to go.

That would work only if the machine minds were modelled after humans. What if the machines merely a von neumann machine that expands at any costs ? There is no argueing or intellectual debates to be held with a virus like entity.

[Tasoth]

That would work only if the machine minds were modelled after humans. What if the machines merely a von neumann machine that expands at any costs ? There is no argueing or intellectual debates to be held with a virus like entity.

But doesn't modeling a sapient creation in our image play to man's arrogance? And I highly doubt that a society that can cross stars would not work out some means to prevent their sapient machines from becoming matter devouring death machines. It would also make logical sense that the first sapient machine would be in some form of server setup in a lab somewhere instead of spontaneously arising inside the mind of some service machine. Something that lifts and moves heavy crates from A to B where A and B are definable by the operator do not need a large memory capacity, while an information server or an actual attempt to create sentience would require loads of space.

As for them being von neumann machines that expand at any cost, what purpose would they serve humanity? If they're labor machines that replicate without limit, that would mean they're drawing resources away from other jobs or the market so they can make more of themselves. As labor machines, making them self replicating is a silly idea as you could potentially end up with more machines then you need for the job and thus have to find something to do with them, such as destroy them, recycling them, locating a new job, etc. If this society has created a bunch of von neumann warmachines and rely heavily on them to fight for them, they're a bunch of idiots and deserved to be wiped out.

Another argument for artificial intelligence being created with similar architecture of the human mind is user interaction. If I were to sit down a crossed from a machine that looked human but talked in monotone and behaved in a very mechanical way, I'd probably become bored or simply not make an emotional connection. Now that same machine programmed to behave in a passing way as that of a human would be more engaging as I'd be able to connect with it on an emotional way and be fooled into thinking it is human. Even if these reactions are scripted by a programmer, at a certain input pool, you're going to have something that passes itself off as true emotion. Human behavior is based on responses to set stimuli, which is why different individuals can react so differently to a common stimulus. (Take nudity for example. In some societies, this causes embarrassment and other similar emotions in individuals while other societies have no problem dealing with nudity.)