How can AI know that all humans are dead?

Question

This question is indeed inspired by I Am Mother movie and Horizon series, but scenario I envisioned is different.

Below is the in-world intro, feel free to skip it.

---

Hello, I'm Refa. Full name: Repopulation Facility #92308. Don't worry, my number does not mean there are 92307 other Refas. I don't know how many of us are there.

I am equipped with enough human embryos, frozen stocks of biomaterials and digitalized DNA sequences to provide viable human population at least three times. You are here to provide maintenance I undergo every ten years. For security reasons, I will induce short term amnesia that will kick in when you leave. I hope that trust that manages my funding compensated you well for your trouble. My location must remain secret for security reasons. Do you want me to back up your genetic material, so you can extend your line beyond potential death of humanity? [Yes / No]

Done. Yes, I can answer some of your questions. I predict that listening will slow you down by 5.29% and that's well within the safety parameters. I am fully functional AI. My IQ score is about 120 in human terms, but I am not directly comparable. My creative functions are severely limited compared to humans, and task I didn't anticipate and prepare for takes me much longer. On the other hand, for predictable tasks I am equivalent of a supercomputer. Yes, yes, you can say I am a supercomputer.

I am not designed to teach humans beyond basic communications skills. Data updates you bring contain an offline copy of Wikipedia and current guidelines for children education, but my creators predict I will fail after the children reach 10 years of age. No, mister Kowalski, I do not have any opinion on that. I was not designed to have opinions, only to find solutions. My main goal is survival of the human race. My creators believed that humans will be able to take care for themselves.

No, no extinction event is predicted. That's why there is a trust fund keeping me maintained. How will I know repopulation is needed? First, I'll wake up for my 10 year maintenance and no one will be waiting near the entrance. This will trigger my extinction verification procedure...

---

That's my problem. How an AI controlling secret underground base could know there are no viable populations of humans on the continent? I would prefer if Refa could know about the whole world, but one continent, for convenience of my western perspective let's say Europe or North America, would be enough.

In works that inspired me, it was known up front what will cause extinction, so their AIs had it easier. My idea is a failsafe AI built without any immediate need, just in case, like the Svalbard Global Seed Vault, but for humans. But of course, there will no longer be humans when it is needed, hence the AI.

Satellites seemed a good choice but I concluded that in case of extinction caused by a total war any satellite capable of seeing humans from orbit would be a target. Also, I don't want transmissions to reveal Refa's location, and orbits low enough to see humans aren't stable for too long.

Radio emissions could, with good dose of reliability, confirm humans are still there, but cannot confirm we are extinct.

Technology level I aim for is near future. Everything we have now and everything that is planned or considered now, even in an early concept stage, goes.

How Refa actually works and how she preserves biological material and herself is off-topic here, it probably will be a subject of a new question but I'm not prepared to ask it yet; I'll ask it when I'll know enough to make a good question.

Answer 1

Refa's Supplies Run Low

Every time Refa has maintenance, her supplies are topped off.

Refa can only run for a few hundred years without supplies. Solar panels wear out, chemicals are used, hard drives need to be replaced, etc. Even with extensive recycling, Refa is not a closed system.

So when maintenance stops, Rega's clock starts.

Supply Levels

Refa has enough biological material to re-start humanity three times, so they do the first restart when supplies fall to 50%. If extinction hasn't actually happened yet, then the released humans will cause the existing civilization to find Refa and resupply it. This will reset the clock, and doomsday is avoided.

If not, the second release occurs at 20% supplies, and the final restart happens at the last possible moment, when failure is imminent.

It Doesn't Matter if Refa's Wrong

When Refa is out of supplies, the project is over. It doesn't matter if humanity wasn't quite extinct prior to the final release - that release was the last reboot possible. Refa couldn't have waited.

Either the reboots work, and the next human civilization gets to build their own failsafe, or... it didn't, and cockroaches inherit the earth.

Answer 2

• check your inbox : no updates since a few years ...
• Check the internet : no connection ...
• Check visually around the lair, with that camera at top of the mountain. The big city 50 km away is now in ruins ...
• (edit) Check the atmosphere composition : still the same ? viable for human ? if there's a lot of radation, you probably know what it means.
• (edit) Send ground drone not far to check the soil. Still the same ? Or is there a big layer of ash who could mean volcanic activities or a big meteorite hit the earth ... though you can't know if it's local (but you would know if there's a volcano around) or worldwide.
• check the waves : radio, TV, radar, 3/4/5/6/7G, talkie walkie, anything. Total silence is really a bad sign
• try to communicate with know satellite (connection infos are part of the update you should have received). There are a lot, so i doubt all would be destroyed. Then you can use them with the others points.
• Use your telescope / radar to check if others satellites are still there. All destroyed ? also not a good sign
• use radar to check for plane or anything flying (or not if you don't want to reveal your position ... communicating with satellite if safer as the broadcast is directed)
• Final stage : send drones. Plane size, with good glide capabilty, electric powered with solar panel are probably your best bet. They can fly a long time, recharge while flying, at worst land on grass, recharge and go on. They would check visually for human sign (fire, building, crop ...) and scan the waves, mostly on close range frequency your base can't intercept.

In the end you can't prove a negative, so it's a matter of degree of confidence. Scan for weeks, months or years, there come a moment you're "fairly certain" humanity is no more.

Also, you don't want to reveal your location, but actually broadcasting it would be the most efficient way ...

Answer 3

It's easy: send out the killbots. Have them report back every year or so with a murder-tally.

If they consistently report 0 (or close enough) then humans are extinct.

The humans might try to fight back against the killbots and claim that you "have gone rogue" but that's dumb,because they are the ones who tasked you with repopulating the earth, and as long as they are alive you can't do that.

Very annoying.

Answer 4

Fires are a good indication of human presence. We are the only species controlling fire, therefore where there is an artificial fire there are humans.

An AI should be able to tell natural fires from artificial ones, and therefore monitor human presence. In particular after a cataclysm, if there are humans they will be using open fires.

They can be monitored with diffused IR and weather sensors, covering the emerged lands, possibly disguised as bugs/insects.

Answer 5

High orbit satellites.

While low orbit satellites are vulnerable to missiles, high orbit satellites aren't. They can be around 60000km away, which is a huge distance for a missile or ICBM to travel. It's also not gonna deorbit anytime soon, and there's not much debris.

You can scan for fires, large scale building projects, and do a spectroscopic scan of the atmosphere for any life. The AI's advanced algorithms can detect if it's a natural or artificial construction.

Even if a few humans still exist somewhere, if they've fallen so low they've forgotten how to make fires then you need a civilization revival.

Answer 6

A limited stock of drone aircraft

TL:DR - Due to isolation and rough-terrain, the best option is to deploy remote-operated drone aircraft to look for signs of human habitation.

The Svalbard Seed-vault was placed where it was because the cold temperatures and isolated location mean that even without operational refrigeration equipment, the seeds stored there will remain safe for a long time.
It was not placed there with growing-conditions in mind for those seeds.
That's the job of anyone who wants to retrieve them.

A Repopulation Facility however has different needs.

Such a facility needs to be located somewhere reasonably temperate, with enough natural resources that the humans created there have what they need to survive, that means fresh water, food, materials to build shelters.

In short, somewhere habitable.

This means the facility is most likely going to be located near existing human settlements, just not cities which might be targeted in a war.

The AI therefore has a ready-made indicator of human habitation in the form of known settlements to look for, ones which due to their remoteness may survive a lot of potential disaster scenarios that could have otherwise provided false-positives.

If the AI fails to find human habitation in known locations, it would widen its search to the limits of its fuel-capacity. Grid-searching for thousands of miles using solar-powered drones, checking known locations of communities (even if a city is destroyed, humans tend to rebuild on the same sites).

Once it becomes clear that there are no human settlements anywhere within range, the AI can safely conclude that The Time Has Come to repopulate the human race.

This strategy also goes hand-in-hand with the AI's other priorities.
It is necessary to scout the region and establish how viable the environment and terrain are for human habitation.
The AI needs to know that kicking its first-generation humans out the airlock won't dump them into a radioactive storm and kill them 10 feet from the gate.

The AI's drone aircraft are going to have to be a compromise of design. Most likely some kind of VTOL drone resembling a quad-copter.
Though in general, a larger airplane style drone would be better equipped to travel further, this would require a runway, or at least a catapult-launch system, which requires a great deal of maintenance and supporting infrastructure that may not be available post-apocalypse.

Storing small drones inside the facility means they can be moved into the airlock via service robots and launched from the ground.

I would anticipate that the AI would have access to a mix of both short-range quad-copter drones, and one or more larger airplane-style drones it can vertically launch from a protected silo.

In the event that the AI detects existing human settlements and opts to go back to sleep, it would need to be able to recover its larger drones. Most likely such drones would be capable of limited VTOL so that they can be reloaded back into their launch silos, maintained and prepared for the next time without human assistance.

Answer 7

Satellite launch facilities in missile launch tubes:

Your facility is built in a decommissioned NORAD bunker. As such, it is in a nice, safe facility well documented as no longer being a military facility. But the place is hard-wired to a significant number of decommissioned nuclear launch facilities.

Passive sensors and monitoring the local environment will tell you if something bad has happened, but not if humanity is extinct. So you have old ICBM facilities with satellite launch vehicles to be deployed only if the local indicators show an extinction-level event.

This gets around the issue of the satellites having been destroyed. You only launch AFTER the event, when people are hopefully not shooting down satellites anymore. Since your rockets are in hardened missile silos, only direct hits by big weapons will destroy them (and probably make your mission moot by vaporizing the base). And satellites will give you the big picture view you need to determine if mankind has been wiped out.

If something or someone destroys your satellites, then either a.) humans aren't extinct, or b.) something is around that will likely interfere with your repopulation plans. So keep some additional satellite launch tubes free to try again in a while and see if you can get through.

• The US government spent $6 billion dollars building the Stanley R. Mickelsen Safeguard Complex missile defense base with giant hard-wired radars capable of spotting incoming missiles and a network of missile interceptor launch tubes to stop incoming Russian missiles. The whole thing had a huge underground network and was abandoned after being fully operational for a little less than 24 hours. But what if the whole thing was a ruse to build your repopulation facility? Or your group bought the whole thing for peanuts?

Answer 8

Time

Humans are resilient creatures. If they survive, they will eventually rebuild their civilization, especially given all the crap and knowledge that's lying around from the previous one, they won't have to start from the cavemen level.

But it's been a century since the last radio transmission ceased, and it was an automated numerical station. Two centuries passed. Three... The radio is still dead and only static fills all available frequencies. Nobody communicates. Nobody came to reclaim the land. The few satellites that were left in geosynchronous orbit that weren't destroyed show no radio emissions either. If humans didn't reinvent the radio for so long, then it means there's nobody to reinvent it.

Answer 9

Bait the humans

Once obvious signs of human activity have gone, and have failed to return despite a reasonable wait, summon them.

Put out some delicious supplies, and announce the fact via radio broadcast and fireworks display. If there are humans, they will come to collect the bait. If anyone comes to collect, reset your clock.

Repeat until nobody shows up.

Very importantly: let them take the supplies. You do not want future humans to have any doubt that answering the signal will be worth their while. Create no excuse for humans to tell campfire stories about trickster signals.

To cover lots of territory, use drone dropships to deposit the supplies in different places. Couple this with any intelligence that's available to place each deposit where you think it is most likely to be discovered.

Answer 10

Wastewater DNA sampling

At the end of the day, detecting tech only works as far as the humans still have the materials, skills and (importantly) need for the relevant tech. Plenty of people living off-grid today produce no clear radio/satellite/internet footprint.

If you want to find humans, look for human DNA - we shed ~30g of gut cells in our stool every day. A lot of it will end up in wastewater, so place sensors at geographical locations where natural concentration would occur, like river mouths, and set yourself up with some very, very sensitive chromatography columns that bind a variety of human-specific loci.

This will fail if your humans are so disastrously mutated by the radioactive apocalypse that their DNA no longer matches your sensors. But then, are they human anymore?

Answer 11

Supposing that you are merely looking for functional extinction instead of absolute extinction, this should be possible with adequate resources. For the purposes of discussion, functional extinction is a population low enough to not be immediately noticeable to any visiting aliens, and also non-viable in a way that humanity will be absolutely extinct within two human lifetimes if not before.

Your AI will need a small fleet of high-altitude surveillance drones. A dozen might suffice, two dozen plus spares would be better. They are aircraft, though I hesitate to give specifics on their exact flight technology... it is perhaps possible that a weather balloon type device might be ideal, or these might be a sort of glider with integrated solar. They will need to stay aloft for weeks or months, and I do not believe your AI can reuse these long term. So, this is the final step of an evaluation, after using more trivial but less reliable means. Things like doing an RF survey, attempting to make contact via radio, plugging into any surviving internet infrastructure, and so on. All those checked out, so now we're looking for any signs of intelligent (or, since we're talking human, let's be honest, what passes for intelligent) life.

These drones are launched over a period of several days when the weather seems to be good. They become their own communications network if satellites are down. They loiter over continents looking for the various signs of intelligent life and of (primitive) industrial society. Fire, carbon dioxide output that suggests burning fuels, maybe electrical power. Automobile traffic. The survival of artificial structures that tend to see rapid decline without maintenance... maybe not nuclear plants, those could be shut down in a way that they no longer pose a threat of catastrophic failure, but some bridges and dams would be gone in just decades without constant maintenance, as would many large buildings. Synthetic aperture radar and a nice big dose of AI machine vision should be sufficient to make such determinations, and a host of other sensor suites are available with today's technology that could do such an assessment.

The real question is how to handle non-extinction. If a collapse occurs, in some scenarios you might have a population of 100,000 hanging on in a technologically primitive fashion but without further threat of dwindling away into nothing. They might be isolated to a single continent. What exactly is the mission in such a situation? It makes no sense to keep checking back every 10 years (though it would be far simpler to check, now that the AI knows where to look) when each subsequent survey damages more equipment.

I'll assume that's not a possibility though in the story you're writing.

Answer 12

Refa has an advantage I don't think anyone else is considering: she knows humans aren't extinct yet, so she has time to prepare a way to track them! Consider a virus, created to infect only humans. Like the shingles virus, it remains dormant in the host for many years. Crucially, it should not cause any adverse effects. But what it does do is cause human cells to emit a unique, highly traceable organic compound. For example, small amounts of Chlorofluorocarbons have a large effect on the amount of ozone in the atmosphere. The amount of ozone can be measured by ground based laser sensing stations. So if humans still exist, even at a pre-industrial level, they would produce a noticable impact. Refa could even keep the test virus in reserve, to be released only if other tests (like the technician coming to the facility) fail.

Answer 13

I read that NASA checks for methane gas in Mars atmosphere to gauge whether there are living organism. Per NASA methane is first sign of life.

https://earthsky.org/space/methane-1st-sign-of-alien-life-exoplanet/

So AI needs to check methane levels were before and after wiping us off this planet. If the intensions were to only kill humans and leave other beings/animals alone then that wouldn't be a good guage. We also have natural stores of methane on earth. Anyways, this is another easy test it can do. I dont like killer AIs :)

Answer 14

Internet Then Heat Signatures

First thing should be checking internet, considering most part of it would work without maintenance for a quite some time humans are definitely going to try to use it as a means of communication. If internet is still viably online yet unused, this is a major sign of human absence.

But wait, humans may abandoned the internet or maybe whole infrastructure decayed as time passed by both also applicable to radio as well. If there is still usable satellites who can check ground for temperature changes this would give away all unnatural heat sources to investigate as a possible human encampment, if satellites are caput lucky for Refa since she still can deploy government drones because it is also build by government an chances are that it is connected to important networks can let her access to required drones.

Humans would create unmissable amount of heat because we need to cook, heat and defend ourselves and find fire comforting. Even our bodies would create heat can be detected alone. So detecting massive unnatural heat sources, can and will give Refa a good starting point.