Hiding the plan of a group from an intelligence that reads their minds

Question

A lot of background first:

Humans have been collected by various star-spanning empires. They have been somewhat modified and are now used as cheap cannonfodder and basically sensors.

While far superior machines and weapons exist the modified humans have two advantages. They are build out of cheap easy to get materials (for their tech) and upon their death the entire consciousness of that human can be caught and placed into a new 3D printed body*. Humans are used to deplete ammo and to scout for the advanced warmachines by dying to them, or potentially getting lucky and destroying one.

An AI oversees multiple star systems and gives orders to several Machine Intelligences (MI) per system which coordinate the details of the battle. The MI's also use the humans as willess slaves with an implant for a constant indirect control.

Here the story kicks in: one nation has voted for their biological soldiers to have their own free will, as long as they dont start losing too much. It both means more efficiency as no implant needs to be build and it soothes the consience of the nation's inhabitants. The MI's are tasked to keep monitering the human's minds upon death to judge their efficiency both individually and as a whole. Badly performing humans are warned and eventually terminated and replaced. If the humans dont perform well enough the entire population is turned to willess slaves again.

The humans do their thing and eventually have to defend the internal structure of an MI against being hacked and taken over, which would turn an entire section of the army to the opposing side. Another MI guides the defending forces so any surviving troops can attempt to destroy the MI in the event it is hacked.

After the successful defense they hatch a plan: should they have to defend an MI again, they want to simultaneously defend it, steal the equipment their opponents bring in to hack the MI and hack it themselves to free a portion of humanity and hope they can carve out their own nation somehow. Anyone involved assumes they will be made willess or terminated regardless of a success or failure as they will be guided by another MI.

The problem: the MI's will essentially read the mind of anyone who dies for intel on both the current battlefield situation and their performance. So how would the humans hide their plans from the MI and simultaneously coordinate the equipment theft and hacking of the MI they defend?

*if you had access to their body as they were build, stealing a consciousness is difficult.

Answer 1

Fair Warning: This question has the basic form of "how do I defeat my godlike character?" or "I've created an impossible-to-defeat situation, how do I defeat it?" That's impossible (and boring). It always requires your godlike character or impossible situtation to have an exploitable weakness (aka, an Achilles' Heel). So while I'm going to give you a couple of ideas concerning how humanity might do what you're looking for, you need to realize that the preferred solution is for you to create that weakness in your MI such that your humans can exploit it. You'll see why as you read my suggestions.

---

Don't Die: The movie Edge of Tomorrow has, if you think about it, the same premise as you do. Tom Cruise' character has the power to bring down a godlike force so long as one thing happens... he doesn't die. Your humans have the same requirement, anyone who knows enough about the plan can't die. How you write that into your plot is up to you (or, they die just in time for everything to turn out all right).

Compartmentalize: Humanity does its level best to coordinate this attack in a way such that no single human knows enough about it to tip-off the MI when they die. In other words, 99% of the plan is being carried out by soldiers who don't know anything more than the one duty they've been assigned to do. In this case, misdirection will be your friend because, for example, having one soldier (the most likely to die) capture the equipment to deliver it to the MI for disposal but another (the least likely to die) to intercept that first soldier to redirect the equipment somewhere more useful would be good. Especially if the intercept can be done in a way that leaves the first soldier dead and still believing that nothing's wrong. This comes from the movie The Guns of Navarone where the the injured character Roy Franklin was given bad information about the approaching British invasion of Navarone to misdirect the Nazis who captured and cared for him.

Guarantee the Brain is Gone: If a soldier who knows too much does die, the brain must be destroyed in such a way and at such a time that the MI can't read it. (You didn't mention, but I'm going out on a limb to believe that the MI can't read a bucket full of brain soup.) You didn't explain how the MI scans the human minds. If they can do that at a distance, that's impressively godlike. If the human must be transported to a processing center, then we have an exploitable weakness. Humans have another implant (a bomb) that detects brain death and vaporizes the head. MI might catch on eventually, but that adds to the plot.

---

Why are none of those particularly satisfying? Because your godlike machines can pretty much circumvent all of them. The machines have certainly had to deal with rebellions before, so they'd have conditions to stop all of this. They'd be watching for patterns in behavior that suggest rebellion or a threat against themselves. They'd be monitoring individuals that have a lower likelihood of death to ensure they don't develop ways to defeat the MI. Yadda, yadda, yadda.

Which is why you need to create an exploitable weakness in your MI. The idea of having to ship people to a processing center to be scanned is a start (you have the time between death and processing to ensure the brain can't be processed), but the MI would be watching for unprocessable brains as part of its, "dang, there's a rebellion brewing" protocol.

What this would mean is that your soldiers need to die in an unprocessable way that's expected by the machine. In other words, your characters need to take advantage of attributes of the war to "dispose" of the bodies (was there a recent nuclear explosion? Vaporize the body and modify the paperwork so the soldier appears to have died in that attack.).

Another possibility is that the MI can't actually read every aspect of a mind (that's really godlike). Limit what kind of information or how much information the MI can access so that plans can be developed around that weakness.

Or, you could ignore all of this (and your question) by adopting a familiar Sci-Fi trope: you ignore the fact that your godlike creation is godlike. This gets you an ending like the one in Tom Cruise's Oblivion where a spaceship enters the space of a presumably godlike AI that, for reasons unspecified, is unable to scan said ship to determine that the passenger isn't a woman, but rather Morgan Freeman carrying a honking big bomb. (Kinda like most Sci-Fi, which assumes all aliens don't have thermal imaging tech or, if they do, don't know how to use it...)

Conclusion: For fun and more believable stories, avoid having godlike characters or impossible-to-defeat situations.

Answer 2

You don't need to hide the plan, you need to stay low

Have you perhaps heard about the game AI War? It's a strategy game purely focused on puny humans, vastly outnumbered by numerous AIs with superior technology. How do you survive in this game? You make yourself a lesser threat than the rest of the galaxy.

If the cost of the AI taking back the humans is higher than the benefit, they won't do it. Moving your troops away from the main frontlines and exterminating/subduing back all of them are costly. On the other side, gaining back a hundredth of the empire has no value. It's a bit like weeds : You'll never pull out all roots by hands, and with chemicals you'll ruin the whole garden 🥀.

And then came one frame-challenge! I highly doubt hacking a MI right-away is a good plan at all.

Why is it?

• First : Your fighting forces already has free-will! Yes, they can point their guns at you, but you have no explosive collar on your neck. You're not constrained into gaining back control over yourself right now, since you already have it!
• Then, if MIs are crucial to the war efforts, having one besieged again should be rare enough. And I certainly don't think that meatfodder and scouts -which should be on the frontline!- will be near the core outside of a last-ditch effort. That'd be a silly move from the MI, even if you move possible treason aside. Therefore, waiting for this opportunity to happen while knowing the AI can poke at your mind anytime is risky, at best.
• Moreover, if you do manage to fall in that situation, then you'll have to deal with the invading forces AND having no AI ally as the MI is being hacked AND dealing with many clueless people not knowing what to do as soon as it is hacked. It will be an extremely tough situation to handle, gathering people around you as you're fired at by the original invaders.

And most importantly: Having the AI know that you can take control of such key structure will make it enter DEFCON-1 immediately, blazing all troops at your location. It'll be even more prominent since the enemies they're fighting originally will be near, too! To sum up, don't poke the sleeping dragon with a wooden spoon, that's a very bad idea.

It would be much wiser to prepare a rebellion in your own sector to have a safe-space for bigger projects without any suspicious eyes. It's because that's the actual strategic goal of your rebels : Don't bring strategic interest onto you until you're ready to actually strike. The greedier you are in the early game, the more risks you'll face too much foes before you're prepared for it. So wait, gather people, weapons and anti-AI technology, and once your forces grew up enough, attack.

In case you have to hide something...

It's going to be an impossible task for the MI6, but there are some ways to alleviate dead-probing which can be useful both for you plan and my alternative :

Follow a pyramidal approach to your organization and information sharing. Have only the higher ranks hold data about the operation on a larger scale, and keep them wweellll behind the locations where they might die. For instance if rebels are ordered to scout a location before the plan is carried out, let the ones with critical data behind as tactical operators. The idea is to at most give unrelated pieces of puzzles instead of everything at once. It'll make everything slower to decypher.

Also, to create your ruses properly, you really need to have a strict control of information : Don't tell anyone what something is for, just what they have to do with it. If someone knows more than they should do, either raise them in power and keep them away from danger, kill and destroy them to prevent any consciousness gathering or give them amnesia pills like how it's done within the SCP foundation. In any case no leakage should be tolerated. If you're morally grey, give false data to some people, and send them dying so the AI can collect them; It'll confuse the peck of anyone of what's the true plan or not.

Answer 3

Exciting fiction!

The successful defense of the MI inspired multiple popular fictional works. The noble humans possessed of free will fight on behalf of their beloved MI to defeat the brainless legions of their enemies. Their MI (who looks like Aishwarya Rai in some versions, and Hrithik Rosha in others) nearly is hacked and is saved at the last minute. The grateful MI celebrates the free will of its human saviors. The music is great. The exact nature of the gratefulness of the MI hottie varies from version to version. In some versions the saviors of the MI are made even more free and the MI converts from overlord to protector, and sometimes lover. Sometimes the MI gets godlike powers and challenges the universe of god machines on behalf of its beloved free humans. Everyone dances!

A lot of the stuff in the movies really happened, more or less. People are proud of it and rightly so. The MI is not actually that hot but it understands the need for it to be personified and it perceives how these fictions and patriotic fictions in general help morale.

The actual plan is in one of the movies too. More than one. The music is just as good. The MI has watched all these movies in the minds of dead humans. Its favorite is the one where it turns godlike because of the 6 arms and the eyebeams its character gets. It does not pay close attention to the details of any of them. Although one of the tunes is super catchy, even for an MI.

Answer 4

Bury it deep

The Machine knows what the human was doing and thinking shortly before it died. This tends to be sensory data and broad emotional colours of the battle.

You cannot scan for the human's innermost thoughts. You cannot scan for the number of children the human had, anymore than you could ask them in the middle of the battle and get a coherent answer.

The Machine knows the human was disintegrated. The machine knows what the human was doing, what it saw, how terrified it was, and whether or not it was following orders. This is all the machine needs for tactical purposes.

Anything deeper or more subtle cannot be scanned so easily. So even if the human had rebellious thoughts while it was being gunned down, this will not show up on the scanner.

The scanner can only detect sedition if the human was either (a) focused on rebellious thoughts shortly before it died or (b) doing something obviously rebellious.

Case (a) is impossible to scan for as all humans already have rebellious thoughts, even the ones that will never actually rebel.

Case (b) is possible if the human is for example assaulting one of the machine's bases. But we know that without reading their mind since we know our base is under attack.