This question follows Why would warring AIs need humans?

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.

If AI1 conquers a planet, what is stopping AI2’s robot army from burrowing underground a few thousand feet, and turning themselves off for, say 50 years, and then coming out and rampaging amongst the human colonizers?

This is sort of like the Vietnam war but the guerilla army doesn’t need food, water or anything! All they need to do is turn themselves off (so they don’t run out of power) and wait!

Why can’t the enemy army hide underground indefinitely?

Good answers will include:

  • Why this strategy isn’t feasible
  • or if it is feasible, possible strategies the invaders can use to stop them.

A possible answer would be: Just deal with it! There could be earthquakes, there could be hurricanes, there could be robot invasions!

But this makes no sense since the possibility of a robot army numbering potentially hundreds of millions in number is not merely an earthquake! It would require many resources to constantly have a standing army on that planet large enough to deal with this at any point on the planet.


5 Answers 5


There are three main problems.

The first is that very few landscapes are suitable for such burrows. Some will have dust that gets into crucial parts, some will collapse on the robots and crush them.

If the robot is actually designed against dust and crushing, this limits its combat capacity, life being full of trade offs. Which leads into the second, that digging in suitable areas is time and labor intensive, leaving you open to counter attacks.

Finally your robots are sitting ducks. Let the humans squirt a corrosive gas in their burrows, and they are goners. If they can collapse the burrows for protection even at cost to combat readiness, why the planet inhabitants can dig them up with their own robots.

In digging up, they have three advantages. The first is that because they have years, they can reuse the robots to do it. The second is that because they are right there, they can bring forces to bear more easily. Finally, they can salvage the enemy robots so this is not deadweight loss.


What is stopping an enemy army from hiding underground?


You cannot ever be sure that there isn't some kind of self-replicating widget with a backup of the AI's mind hidden somewhere you haven't quite looked yet. Behind the sofa. Inside the nuclear waste dump which no-one is supposed for visit for a thousand years. Buried down in the Moho. On the moon. In the asteroid belt. In the bone marrow of your neighbour.

You could grind everything up and throw it into the sun, or you could build the tools you need to deal with the situation.

You, the OP, as some kind of human meatbag have almost certainly been infected with at least one kind of herpesvirus in your life, and probably multiple ones. Pretty much every human over the age of a few years has been infected. Maybe you've been exposed to retroviruses too. So many ways for potentially deadly self-replicating threats to hide inside your cells, just waiting for the right moment to strike.

And what is that moment? Why, when your immune system is no longer up to the task of keeping them beaten back.

What you need, then, is a planetary scale immune system. Sure, search everywhere as hard as you can and root out every sign of infection you find, but where you've searched store your own defensive automata to keep watch for you. Whatever the enemy can do, you can do. They suddenly rise up from the ground to fight back? Well, your defenses are everywhere and will rise up at the same time and talk to you and each other and can control and contain any outbreak of the enemy infection, if not wiping it out then holding the line long enough for you to bring the big guns in.


Hiding anywhere is an arms race against how good the seeking technology is.

Finding things underground is already a pretty sophisticated field of technology. We can find mineral deposits thousands of feet underground using tech like ground-penetrating radar and radio waves –

Article on "Detection and Imaging of Underground Structures Using ELF/VLF Radio Waves" at Global Security

It seems plausible that in the far future, this has progressed. So hide all you want, but we'll find you.

  • $\begingroup$ Worth remembering that there are resolution issues with ELF/VLF. Spotting big nuclear bunkers may be possible, but smaller objects are going to be undetectable. Similarly, mineral deposits are large, as are most things that deep geophysical trickery can discover. $\endgroup$ Sep 15, 2021 at 12:15
  • $\begingroup$ Are those reoslution limits hard limits or will they be improved in the future? $\endgroup$ Sep 15, 2021 at 12:16
  • 2
    $\begingroup$ Baked into the laws of physics, unfortunately. $\endgroup$ Sep 15, 2021 at 12:20
  • $\begingroup$ I agree with this one. If there are robots down there, they got down there somehow and getting down there will leave signs. The victorious AI knows this. Its humans never stop looking. And don't think burrowing down on the moon will evade notice. The victorious AI thought of that too. It is victorious for good reasons. $\endgroup$
    – Willk
    Sep 15, 2021 at 13:38

Tunnels are easy to spot, and sleeping robots are easy to kill.

If they dig down a huge distance that takes time, and requires moving a vast amount of dirt. The enemy AI can track that from the sky.

They can then send burrower robots to find enemy robots and kill them. If the robots are asleep they can't monitor the area around them for threats, so they'll be easy prey.

Fifty years is also a long time to search for enemies. Since you have control of the surface you can get fresh supplies and power, and those below cannot without making a massive noise, so you can detect them better.

Scanning tech can find weird rocks.

We have lots of ways to detect gold and other ores underground, with radar and gravimetric detection and with lots of bots, and fifty years is a long time to scan for enemies. The AI could devote a small constant budget to detecting hidden enemy traps.

Human intelligence may help here, with experienced miners having more idea of what to look for underground.

  • $\begingroup$ I challenge how easy these are to spot. In Vietnam, tunnels were use a lot to hide in and they were not found. Perhaps in a dry, desert environment they are easy to spot, but not in any heavy vegetation. There are stories that there are German tunnels still not found from WWII. $\endgroup$
    – David R
    Sep 15, 2021 at 15:52
  • $\begingroup$ Tunnels are hard to spot for humans. Robots can track movement from the sky, and deploy swarms of robots to check every inch of the ground, and under the ground, and have more advanced technology to scan for tunnels than ww2 soldiers. $\endgroup$
    – Nepene Nep
    Sep 19, 2021 at 23:52

robots can go in a low power mode. but they would need some sort of clock running otherwise they come out of the ground one at a time. such a running clock would require only a little power to run. but it would limit the maximum duration they can stay underground. you would also need some kind of new battery tech. because modern day ones tend to leak after years of no use.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .