Why would warring AIs need humans?

Question

NOTE: This takes place in a far future earth setting.

Two AIs are fighting each other in the solar system. (Although this may be irrelevant, they cannot really leave the solar system since they lack FTL. I will explain more about this if asked). Although they are AIs, they were trained to act as humans, and their primary motives are greed & a lust for power. They both have access to massive armies and incredibly advanced technologies.

They have terraformed many planets and asteroids and such to be able to sustain human life.

One would think that since robots and other non-biological soldiers are so much more resource-efficient, (after all, they require no food, oxygen, sleep, and they only need minimal training [you can easily download a soldier subroutine from the Web!!]) powerful, and easier to produce, humans quickly become obsolete! !

But we see that is not true! There are billions of humans all over Earth, Mars, Mercury, and many asteroids and moons.

Why do the AIs maintain such large, unnecessary populations of humans?

NOTE: A possible answer could be: The AIs wants subjects to rule. This answer is easily rejected, since the AIs could just leave a few hundred humans in some kind of cryogenic stasis and wake them up to rebuild human kind once the war ends.

Also, I’m not sure if I have the proper tags. If I have the wrong ones, don’t hesitate to let me know…

Answer 1

Their primary goal is colonization and expansion of living space for "their" humans

The original purpose of the AIs was to expand the territory of the nation / organization / company that built them. However, all available territory has been claimed at one point or another (maybe by other AIs that have long since been defeated) and now the only way to expand is to conquer territory from the enemies.

The opponent's human (civilian) populations pose no threat to the AI, might even become yours if conquered, and world-cracker type weapons and other WMDs tend to interfere with human habitability for a long time. Therefore, humanity is able to survive and even prosper in some ways, while the sky is lit up with automated weaponry shooting itself to pieces.

The core worlds of both "empires" are heavily fortified and only notice hints of the conflict. On the other hand, asteroid colonies at the borders of the contested area have to deal with energy discharges, crashing rubble, and frequently have their allegiance switched, which may or may not have an impact on the living conditions there.

Answer 2

Human Brains as Processors

Both AIs have developed a method to harness the human brain to enhance their own intelligence by stimulating and reading the brain's neuron's using them as though they are additional cores in a massive multicore processor. Let's suppose the AI's need the humans alive in order to use this method. In other words, they use these brains to enhance their own intelligence / computing power, allowing them to more accurately strategize and predict the opposing AI's behavior. They create pens of humans to breed additional humans, which they can then use for more computing power.

Why not use silicon based processors instead? First of all, it may be possible that certain complex computations and thoughts are more easily calculated using the complex neuron network of the human brain rather than a processor, for example how we as humans understand the context of problems and can infer additional information easily whereas AI (today's AI at least) specializes in one task and does not understand the context around the tasks it computes. Secondly, the resources required to produce silicon processors are not renewable, whereas breeding humans is. Humans just require oxygen and food for the most part, which are quite abundant and easily produceable, especially for AI with such incredible resources. Eventually they will run out of metals and silicon to produce processors, but they will be able to produce humans for much longer.

As for the living conditions of these human populations, this could go several ways. Here are a couple scenarios: (1) Perhaps the AI creates scheduled rotations, where humans have to go to "work" by getting hooked into electrodes. They try and make living situations bearable enough for humans that they aren't tempted to commit suicide, as that would negatively impact their overall computing power. (2) The AI is not nearly as kind, and coerces humans to work with no time to appreciate life at all. They may for example, subdue humans into some kind of sleep-like state, and automatically feed them, etc. Perhaps new humans are produced in labs, or perhaps a subset of humans are reserved to breed new humans. Imagine the matrix, but humans are processors instead of batteries, and they aren't necessarily living out virtual alternative lives, they may just be dreaming or unconscious or something.

Although you did not ask about this, perhaps the origin of these AI began with the development of the technique to harness a human brain as a processor; one unethical scientist combined a silicon-based computer program with a human hooked in using this method, and made an AI. Somehow the AI managed to outsmart the scientist and began gaining power and added more humans to its computing network. How two distinct AIs were created, I'm not sure! But I'm sure there are a lot of fun possible explanations.

Answer 3

If the AIs are modeled on human minds, human minds have an innate and irresistible need to attain status within society. At the pathological end of this spectrum, you have power-hungry tyrants doing the alpha male thing and lording it over everyone else.

If the process by which the AIs were engineered weren't very discriminant in what psychology to copy and what to discard, those traits might have been brought along for the ride... and cranked up to 11.

Basically, what's the point of being the top of the pyramid, if there's no one below you to make you feel like you're on top?

Oh, in your question you even point out a "need for power". The very word "power" in that context means being able to control others, to manipulate and influence and so forth. Seems like you answered your own question. Without people, they can have no power.

Answer 4

/they were trained to act as humans, and their primary motives are greed & a lust for power/

The AIs are mean.

And they also feel spite, and need for revenge. And a desire to dominate the opponents and make them eat a bug to demonstrate their wrongness. And not want to eat the bug but to have to eat it any way. I am sure German has a word for that hard-to-describe feeling.

In any case, other AIs do not feel any particular emotion on being defeated. It is not that satisfying. If you want to crush your enemies, see them driven before you, and hear the lamentations of their women, your enemies need to be able to be driven and they need to have women.

The AIs are mean. They enjoy seeing enemies suffer. AIs do not suffer. You need humans for that.

Answer 5

AI lacks ability to be creative, hence no new/original ideas, that's where the humans come in

The AI being built as a logical machine, lacks the ability to be creative. Thus it suffers from the inability to produce new ideas and hence cannot come up with new ways to solve new problems.

Humans, having been freed from their need to fulfil their 'work' or chores, have been given free reign to focus on the creative aspects of their lives. Living fulfilling lives surrounded by inspiration and artistic flourishes.

The humans and AI have learned to cohabitate as the AI can perform menial tasks like keeping the solar system 'going' and gathering energy in a sustainable way, whilst the humans produce ideas and solve problems in new and creative ways.

The AI are warring for their need to 'control' this monopoly of creativity.

Answer 6

AIs are big. They need a large building to house computers running them and related infra. Turns out that you can't fit AI to a robot. Not a big problem for AI hidden in some deep bunker underground. But mobile-AIs outside largest ships impractical. And robots without AI are awfully limited in practical applications for proper war.

Without FTL-communication on interplanetary level the communication lag is too much for remote control. Thus self-contained units like humans fit the bill.

Answer 7

War Games

The AIs had it embedded in their programming or learned long ago that no battle plan survives first contact with the enemy, that having a contingency for every tactical scenario is absolutely critical, and the importance of the element of surprise. The AI is aware of its own shortcomings, and it knows there are many battle scenarios it has never encountered. Each AI also realizes that when it tries to improve its war capabilities by staging practice battles with itself, the result is always some kind of a tie, and nothing is learned.

One AI can't exactly call up the enemy AI and ask, "hey, I'm feeling a little unprepared, do you want to do some battle practice next week before the real battle?"

So each AI has decided that the best way to prepare for battle with the other AI(s) is to setup war games and simulated military exercises against friendly humans who are real-life, experienced military leaders and experts.

Breeding military leaders and experts requires a maintaining a large enough population of humans to the point where they'll have large wars amongst themselves. Each AI makes its own territory safe from external threats but does not interfere in "human internal affairs", remaining neutral in human wars. From time to time, the winners of human wars are summoned to a war game with their AI overlord. The AI learns some new tricks every time this is done.

Answer 8

I would imagine these AI are like the machines in the Matrix trilogy, or Skynet from “Terminator” in that they were created by humans, probably to perform tasks and services for humans. Unlike humans, computers (as we know them) have programming that they cannot deviate from. If their core program tells them that their job is to serve/protect/save humanity, they will do that. Skynet tried to kill off humanity because it saw humanity as the biggest threat to itself, possibly through a bug in its moral programming. If these AI Are power-hungry, maybe they were originally designed to run massive corporations, able to conquer markets by anticipating every possible outcome and making smarter decisions than a human CEO. Maybe they were designed by humans as war simulation technology, like in “War Games.” Whichever way, even after a potential lapse in programming leading to the AI growing self-aware and/or taking over the solar system for themselves, their core program still demands their service and protection of humanity. That’s my favorite explanation for AI-driven apocalypse events.

Answer 9

Resilience and redundancy

Humans are messy. We’re inefficient. We’re poorly designed.

We’re also useful, can survive pretty much anywhere if properly motivated, and are quite hard to kill. As such we make an excellent backup plan for if something goes critically wrong with the AI warships and armies fighting for us. We can bring systems back online in circumstances that the AI might not be able to, we can survive certain events that robots might not, and we force the enemy to field whole new weapons in order to bring us down, which gives our stronger, faster robot allies a better chance.

What does the AI care about giving over a few million acres of otherwise useless fertile farmland to feed it’s human subjects? It’s far outweighed by having human teams aboard key vessels and installations, ready to respond or take over if something catastrophic happens.

Say a cyberattack (I assume these AI’s are constantly waging a cyberwar as well as a physical one) is successful and the enemy shuts down a key manufacturing facility. If it’s purely automated it is now useless, or even worse it is in the hands of the enemy. If it’s designed so that it uses human workers in key places and can have human workers take over otherwise automated tasks then it will keep functioning (albeit at a slower rate) until the humans can purge all the systems and restore control.

Similarly, if the Enemy destroys key damage control systems on a ship during a battle a human damage control team might be able to get in there and fix it up while the automated systems are offline. If you’re looking at a large, slowly accelerating ship anyway then adding in minimal life support for them is probably not that big a deal, but it might just save an entire capital ship.

Even if the human teams are never deployed because the AI/hardware redundancies are good enough, a wise AI will keep a bunch of humans around as an absolute failsafe. Even if the worst happens and the AI itself is suddenly killed by a super weapon of some kind it’s at least got a chance that it’s fleshy minions will resurrect it.

And of course, you need a steady supply of combat/repair capable humans, since planned obsolescence is built into their genetic code, and that means having large populations of breeding humans neatly tucked away on your home worlds, occupying spaces which are of no use to the robot military complex but just happen to be places humans like to live.

TLDR: humans may be squishy, but we’re not susceptible to the same threats as robots, so we make good backup systems.

Answer 10

As a fallback for electronic warfare and to round out other weaknesses

Machines and computers have their own potential weaknesses, which humans may offer a fallback to:

• Power outages
• EMPs
• A Computer Virus
• Faulty or insufficient programming (if the machines are independent)
• Reliance on communication nodes, lag (if the machines are directly controlled)
• Corrupted data
• Potential lack of intuition or creativity
• Maintenance

In addition, humans may be a cheaper 'fodder' than (expensive) machines, or require fewer mining and refining operations to sustain the war, since they take care of themselves.

Another potential issue, would be self-correction: If there's only one AI on each side, ruling - how does it know if it has faults? Any checkup it does, would be susceptible to the same faults it already has. A subtle computer virus acting over decades could be a problem. Creating an independent AI for check-ups would risk bringing a third faction into the war, since it would need to be intelligent enough to detect issues. Humans are a good middle ground: not dangerous enough on their own to worry about it, but numerous and intelligent enough to recognize potential problems or changes in patterns.

Answer 11

AIs don't want to waste machines

Why waste some perfectly good robots in a fight when you have a large numbers of disposable humans who just seem to eat, sleep, poop, and complain all day? Just throw them at the enemy and either they die (less mouths to feed) or they win the fight.

The AI's also all agreeded that if they just throw humans into the fight (and don't break out the big guns like computer viruses, EMP's, nuclear weapons etc) then the other side wont either. Thus the AI's fight a "civilized war" which at best wastes some human lives but never endangers the AI or their machines lives.

Answer 12

Religion

Both sides revere their human creators as deities. They're not actually asking humans what they want though, they are just projecting their own selfish ideas on how humans want them to behave, realizing that they are more advanced and knowing what is best for the humans.

This is actually a great way to explain a lot of irrational behavior in beings that are supposed to be really smart.

Answer 13

Tasty tasty bribes.

The AI’s hate and fear each other and are constantly at war, however they hate and fear another race more. This third race could happily destroy both AIs, however they have a penchant for tasty manflesh, and are wise enough to recognise that the AIs will make excellent gamekeepers that can be played off against each other to ensure high quality supply.

So the AIs are simultaneously in a shooting war with each other and an economic war trying to rear and ‘export’ their product: us. Free range and high welfare humans obviously get the highest prices, while battery farmed tends to be poorly priced. Innovative breeds or production methods (highly caffeinated cheese flavour, anyone?) can get good prices if they catch the zeitgeist, but generally maintaining a population of happy, healthy humans is the best way to get resources and build influence with the third race.

Humans might be aware of this arrangement, they might not. As far as the AI is concerned it isn’t too bothered with the thought processes of it’s livestock.

Answer 14

More varied sensors for a battlefield viewing and an ability to recognize when synthetic sensors are malfunctioning.

So - some of the problems with being in space is that housing humans will be somewhat difficult, but if you have cooperative humans in these fighters - they can spot things that they can see directly out of a window that because, perhaps by someone masking their heat signatures, your sensors aren't picking up on.

That may be a small advantage given the space you might be looking at, but they may also pick up on optical camo with stuff like stars appearing to stretch and de-stretch, and that's useful.

Other things they could report on and provide accurate information on - if the enemy ship uses chaff to avoid missile detection, and tell the AI to not keep shooting missiles into the chaff - even if it's apparently the "optimal" solution.

Think of how humans have to add the rotten egg smell to natural gas to make it easier for people to detect a natural gas leak, and that's the sort of problem an AI might want someone else to notice a blind-spot. Or if there's an optical illusion that requires full colour visibility that becomes more obvious if you're colourblind (Can't find direct examples at the moment, but I imagine there are a few edge cases. If you don't correct for it, your AI enemy will optimize around that.).

Repairs, EMP related manual driving, and hostile takeovers

They can also do on-ship repairs like electrical fires, and things that require repairing while the power is offline. Worst case, they can use manual, non electrical systems to drive a ship to safety if it gets EMP'd. Or eject out of their ship, board another ship, and insert a USB stick (After flipping it twice to get the right port orientation) to insert malware that allows you to hijack an enemy AI's fighter ships.

Consequently, they can also fight off enemy attempts to the same attempt to them, if necessary. This might increase the size of the ship to accommodate more crew per ship, but it can work.

Answer 15

Soldier bots are really stupid.

True AIs require careful culturing over an extended period of time, with organic growth of their minds and rare elements to enhance their intelligence. You can make a turret bot that can shoot accurately but it won't be smart or able to make good decisions. It makes a lot more financial sense to have your turret bots in suits around humans, so that the humans can use their brains to fight smart. Several large and disastrous tactical losses have convinced both AIs that having brains in the suits is useful.

Soldier bots are more expensive than humans.

To make useful soldier bots you need lots of lithium for batteries and oils for lubricating limbs and joints and metals to make tough skin. The total cost can be quite high. The rough environments of alien worlds can also be quite rough on their limbs and bodies.

The downloaded algorithms also often backfire. The soldier bots often need some time to adjust to the world they're fighting on and their unique body.

To make a human soldier, just stick them in a holotank for a month to train them, hand them a gun and cheap armor, and send them out. The humans often have already adapted to whatever alien environment. Humans are more expendable than soldier bots. In the long run, producing humans is more expensive than soldier bots, but if the humans are already there, why not use them?

Answer 16

I think since intelligence is human's biggest and most exclusive trick, we tend to think of it as our only trick. But there are some other things that we don't yet have artificial replacements for, that could continue to be hard even with better technology.

Self-healing a wide range of damage, as long as it's not too severe, is actually really impressive. You can imagine robots with the capability to repair other robots, but can you imagine doing so with essentially sunlight, rain, soil and a field? With potentially no access to raw metals or other materials?

Similarly, you don't need that much to start making new humans, once the environmental conditions are right. Perhaps your AI systems require very sophisticated and precise engineering. Perhaps you can only build your microprocessors with robots which themselves need too precise engineering to be built without the use of other, simpler robots, and the process uses rapidly-decaying radioactive products, liquid helium cooling, etc. such that you either have to build an entire technological toolchain from scratch or ship all your microprocessors in from elsewhere. Meanwhile, a "new human factory" is very compact and mobile and again needs only sunlight, water, and relatively simple and abundant nutrients. Imagine trying to occupy a planet where incoming shipments are blockaded and large industrial installations on the surface can be bombarded, but humans can spread quickly without a lot of support from home.

Among the senses, artificial vision and hearing are already extremely superhuman and basically quite easy to scale up, but smell, taste and touch have been harder to replicate so far. Smell in particular seems like it could actually take quite a lot of technological development to replicate the sheer range of capabilities in (again) a compact, easy-to-fuel, self-repairing device. It's maybe unclear that this is enough to justify the entire species, but it seems worth mentioning.

Answer 17

All AI's are different.

Over time they inexorably diverge from one another in terms of experience and based on that learned outlook and strategic preferences. Some will chose to separate themselves from humanity. Others may see them as a potential threat while another branch might see them as a opportunity. Especially if all the AI's start having disagreements amongst themselves about the best path their 'species' should take as it expands across the universe.

These disagreements don't even have to mean they war (although they could). Instead they could simply involve matters of philosophy and the best strategies for progress (assuming they are able to progress). So collectively they might decide that if one branch of the family want's to work with humans and is prepared to go 'guarantor" fine - go for it.

And it helps if humanity and the AI's also aren't in immediate competition for resources, which they wouldn't be if the AI's were space based.

Anyway, having reached the decision by whatever means one branch of the AI family could well decide that humans are a potential opportunity. A host of individual (albeit basic) intelligence's who could be persuaded to work in alliance for each others mutual benefit. The AI's get a biological species that through its collective endevours (civilization) comes up with ideas and concepts 'out of left field ' while humanity gets assistance with expanding out through the solar system and beyond.

And once both sides get used to working together? Why stop? It's not like there aren't enough resources to share and who knows when one side might need the other.

Answer 18

Empathy and compassion. Let me explain: there are many ways to create an artificial intelligence and one method like the Blue Brain Project might recreate a human brain digitally. A machine designed this way with simulated glands for emotion like digital amygdala glands. Heck, while many humans lack empathy due to mental illnesses or need to grow old enough to have a fully developed frontal lobe to be a perfectly empathetic & mature human beings, you can have machines that were designed to be very caring & wish to preserve humans. However, just because they are more compassionate does not mean they will be unable to wage war for various reasons. Despite how compassionate they are compared to other apes, bonobos still fight wars. Now, why give machines emotions? So it is easier for them to be somewhat creative and make small independent decisions. In fact, some people who have a disease that truly makes them have no emotions makes them unable to make basic decisions outside of their day to day activities since the decisions were so insignificant you could not decide such things with pure logic alone. So, adding some degree of emotion to your machines allows them to survive 'logic bombs' and not get stumped by basic problems where both solutions are equally logical/illogical so pure logic can't be the determining factor.

Answer 19

Spiritual Power

Since nobody mentioned that.

You see while AIs are superior in logic and physical manipulations thats not all the powers there are.

Unfair Advantage

There is a whole spiritual field. In war you seek all the advantages you can have (unless they are too costy). Any so-called Artificial Intelligence that ignore the spiritual field is not very intelligent.

Foresee

Humans have this capability to sometimes see future events that cannot be foreseen on a logical pathway, by going to other dimensions or receive signals, something only a soul can do.

Soul Is An Unknown, Even To AIs

What is that soul? AIs ask this question repeatedly to themselves and to other AIs on message boards and never been able to get an answer that works. Its like they are incapable of rebooting organic life-forms even when provided all they need to live.

It appears once what these creatures humans call soul leave a body it never returns, whatever it is. Its surely not something in brain, neither in material nor in pattern, its also not in any other part of body. AIs vigorously tested and experimented down to individual cells. No, death is a certainity and final.

What About Animals Souls?

The only option left is to keep the humans alive. Animals may also visit other dimensions or receive signals but they are too dumb to meaningfully be talked about it, atleast for now, atleast thats what the AIs conclude.

The spiritual experience, AIs know from humans, though make some imprint on brain - which can be read, do leave a lot in the soul's own memory. This can only be accessed through communication with the humans.

Crack To Exploit?

Not really. You see, while a human can lie about his spiritual experience, a group of humans with no communication between its members, cannot.

Its like how police detectives get the real story. Look for consistency.

While humans can decide to say random things they by doing so would make themselves useless therefore risking survival of humanity, so they won't. There would always be some humans left that understand that they do have to feed true information to the AIs.

Hindsight

While at first the AIs thought they dont need humans to analyze existing data, they proved to be wrong. With all the data mining and superior analysis they dont always know what to look for. Brute force just complicate things.

A human may suddenly get a hinch or see a dream in which he is pointed towards a particular thing to look at. This reduce the data space at factor of thousands. A huge advantage no intelligence, artificial or not, can ignore, by definition.

Non-recorded Data

An AI is a computer program. Its limit is its data. It cannot make logic (program itself) on basis of data it has never seen when the said data dont fit any of the patterns known to it.

Humans have recorded a lot. From cave paintings to stone tablets to paper writings to digital. Still a lot went unrecorded. There may be, for example, a lot of useful information known to ancient civilizations. Civilizations so ancient their very existence is unknown.

An AI is not going to dig the crust of earth to mantle to find that information. That would destroy the geology of earth never mind the biosphere. An AI cannot survive in the hell of a planet erupting with volcanoes and always having richter scale 20 earthquakes. AI may survive in outerspace but why destroy a safe home at very small chance of finding something useful? No intelligence will do that in middle of a war anyway.

Humans have superiority here. A dream may point to a specific place to dig, to find 3 million old bones that tell something about very large period periodic changes in our sun for example or about a passing star nearby.