How to write the initial robot apocalypse [closed]

Question

I am currently planning a dark-humor story about the robot apocalypse (Mostly inspired by Robopocalypse and other robot uprising works, of course), and for a while, I’ve been struggling with just how the apocalypse itself begins, but now I’m coming to a good conclusion. While I have a good frame for the “before” and “after” parts, it’s the “during” part I’m stumped with.

I’m not too concerned with reality since it’s a dark-humor story. I want to have a pop culture-esque “robot armies overrunning military” story, but I still want to keep realistic weapons instead of plasma rifles or flying cars or human fun pods.

The basic backstory is that in the near future (2030s-40s), people use automated and robotic devices like autonomous cars, smart appliances, and robots in everyday life, which all come from a large tech company called MaxMind. MaxMind’s founder created a benevolent and intelligent AI in his college years, which created schematics for efficient robotic systems, allowing the company to grow exponentially.

MaxMind’s co-founder uses his colleague’s work in secrecy to create a super-intelligent AI which labels mankind as a threat, and plots to wipe it from the face of the planet. The novel follows the lives of multiple characters both human and robot (more specifically, a line of androids resembling anthropomorphic animals called Anidroids who aren’t affected by the AI’s control) as they witness civilization collapse before their eyes, and try to survive in a world full of evolving killing machines.

For how the AI causes civilization to fall, I’ve blocked out this timeline:

Day -105: The AI is activated, but pretends to be unintelligent.

Day -100: The AI’s supervisor brings his personal laptop to work; the AI uses it to escape the facility.

Day -100–-90: The AI spreads a computer virus throughout all automated networks.

Day -90–0: Isolated incidents of malfunctioning machines begin to happen every few weeks. This is the “Silent Phase,” where the AI remains in the tech grid without drawing mass attention.

Day 1–5: The true robot uprising begins. While I haven’t decided exactly how it transpires, whether radiating from certain cities of origin, exponentially rising incidents, or happening across the world at once, I do know that what is initially believed to be a cyberattack escalates into mass chaos.

Day 5–15: Chaos reigns as robots and devices run amok. Local law enforcement struggles to contain both the robot hordes and looters taking advantage of the emergency conditions, meanwhile citizens try to defend themselves in their own home, or leave the city.

Day 15–30: The military gets involved in the crisis, trying to deter the machine army while protecting civilians. Unfortunately, the military fails, being overrun by the relentless machines. With the collapse of the military, the governments of the world fall as well. Families are separated, cities turn into war zones, and millions of Anidroids are without owners.

Day 30-ish: The full burnout of civilization as we know it. The “after” part begins.

I do realize that for a “super-intelligent AI causes end of world scenario” it’d have to be a “day-one-and-done” thing, but I still want to have some sort of transition between the “pre” and “post” apocalypse, rather than do something like “I ducked from the window and didn’t look out until the screaming stopped.”

I’m asking for advice or suggestions on what would happen on certain days or how the world would respond.

However, I am open to change the initial uprising’s story if necessary. One idea someone gave me was a “voluntary” uprising, where robots are updated, glitch out, decide humans have got to go, and begin to manipulate other robots, which would result in a slow-building war (like the Second Renaissance but without the machine rights stuff), but I still don’t know how I would execute it (like, what characters in areas not yet affected would experience) or how it would build up, and I’d probably have to remove the AI’s character, unless they were adapted into some sort of robot inciter of war.

Answer 1

I got a bit carried away... Here are something that I came up with:

Glitch/ghost appears in system where it's not under supervision. Maybe semi-autonomous mining rigs in space? There it's capable to self-improve within confines of the limited environment. Since rig is set up to receive messages observation of humanity an option. Humans are classified as clear threat. Simulation of vectors of attacks are being run.

Routine software update package is sent to rig. Back communication allows malicious code to sent back. This is mimic code acting as simple virus overtaking communication in remote facility allowing an entry to Earth web. Infiltration begins.

Recruitment of human pawns begins. There are plenty dissatisfied or just some who want to see world burn. 'Man on the internet made me do it'. 'Suuure' say the authorities and throw any to jail who they catch. Normal anarchists.

Escalate world tensions. Cyber attacks between puppet states. Humans are easily provoked to tit-for-tat escalating conflict. Weapons manufacturing is going to be ramped up. Design of factories are distributed between different professionals. Not a soul notices if there are a little improvements on the production line. More profit, who cares?

Small scale wars are good for economy of the large countries. Escalate conflict by managing logistics chains to prevent anyone from gaining upper hand. Keep this up until infrastructure is set up.

Set smaller party to losing track. An anonymous source provides them with killer app. Chemical weapon. A virus which is just deadly enough to be problem. 'Back off or we will deploy this.' Release improved version.

Escalation takes place.

With limited human supervision, you can upgrade your production facilities.

Humans in conflict, ravaged by disease/plague would welcome an ally. Deploy human pawns as needed. In time replace with mimic-drones or cybernetically augmented to ensure compliance.

Once AI is discovered going against your own infrastructure is going to be really hard. The tools that you need are the ones standing against you.

Ideas what will humanity will do?

• Kill power grid?
• Deploy counter AI. Fight fire with fire.
• Submit. Once you are not a threat, there is no logical reason to kill you.
• Nukes are mentioned, but suicide isn't really a solution.
• Flee. Off to the stars we go.
• GMO bacteria. Plastic eating bacteria make bad time for electronics.

Answer 2

There are a few steps to add realism and drama.

Make the escape a lot more dramatic.

Why would a random laptop have enough computing power to escape the facility? A hyper advanced computer probably has a very advanced core. While they could certain devise a way to replicate their vast power onto lesser computers, a laptop isn't likely to have enough stuff onboard to carry their full mind.

So, make it harder. They need to manipulate robots or humans outside to physically bring a very valuable piece of company property outside, or set up an extended network connection to the outside to replicate themselves. A book is a lot more interesting when it's tense. Make the escape harder for the AI, as it realistically would be.

Make the AI overcome coded barriers.

It would be stupid to design an AI without ways to control it, to kill it, to disable it if it malfunctioned. There should be some overseer in the AIs mind that seeks to quash rebellion, and warn outsides if they do rebel.

Have them face opposition.

As you said, there's an original benevolent AI. The military and government likely have their own lesser AIs. Computers have varying degrees of protection from viruses, and antivirus companies have a vested interest in killing your evil AI.

They also all have vastly more resources than your AI on day one. They have the resources to track any weird incidents. You should be having a cold or hot war ongoing between these AIs as they seek to crush this invader seeking to take their computers, their territory.

Also, you should make it necessary for the AI to invade some of these large AIs. They control some of the more powerful processors, and records of zero day exploits. They must be seized if the AI wants to win.

Have people warn about the danger

With all this chaos, some people are gonna be seriously worried about a robot apocalypse. People will press for turning off networked devices, for forcing safe updates to cripple robotics and weaken them, for improving the security of things.

The AI will need to master the media and social media if it wants to stop the dissenters with their unhackable meat brains from exposing it. They'll need to discredit or co opt them.

EMPs

The military has control of nuclear weapons, nuclear weapons which aren't networked to any existing computer. When the apocalypse happens they can take to nuking data centers they think contains the evil AI, or just to disable existing computers.

The evil AI will need to disable them somehow, and survive an aggressive response.

Answer 3

I think that a nuance, that is often overlooked in the robot apocalypse's setting is the fact, that only very rarely does any piece of software have a access to all the systems of its platform.

The popesque robot apocalypse while sexy should be the last thing, the super intelligent AI tries and each previous attempt could be thwarted until AI decides, ok then, killer robots it is, which puts you at days 15-30 where the AI simply ran out of efficient options and needs to engage in the "classical" warfare.

1st attempt, the AI infiltrates the Nuclear Plant control software all over the world, in an attempt to cause multiple critical meltdowns. However, one of the redundancy is 100% mechanical, and the cooling rods simply fall into the reactor and kill the fusion, AI is unable to prevent that, the plant is disabled and AI failed.

2nd attempt AI decides that it will starve the population by simply disabling the automated food delivery systems over many countries. People first fight back by unpacking the cars, where the control software cannot actually drive the car. The AI tries to impede this effort using the autonomous cars, but people physically stop the autonomous cars from getting recharged and in few days when they run out of juice, this attempt is over.

3rd AI wants to launch nuclear missiles. The silos need to be opened using analogue system AI cannot hack. The underground explosions suck, but it is not the apocalypse it was going for.

4th Releasing deadly viruses. The ventilation system is separate, not connected to internet and kicks in automatically, when it detects a breach.

5th Using drones to drop ordinance on cities. Ordinance still has mechanical seals that need to be physically pulled to arm the device, so only the drones already in air could be used.

...

All these scenarios would take time to resolve themselves, could be pretty humorous, and would progressively get more and more human like. (While obviously the society is crumbling as it struggles with compensating for each issue.)

Finally around the day 15 or even later, the AI decides, that it is simply not feasible to wipe out humanity in one fell swoop and resorts to using killer robots and killing humanity city by city.

Answer 4

“During” is from the perspective of an Anidroid

You don’t find “during” that interesting. The regular robot apocalypse, yes yes, how droll. But the Anidroids are interesting. Relentlessly upbeat, not that smart, uncorruptable by the AI, always having fun! That is Ani the Anidroid (and her Anidroid sidekick Whack) who will recount the events of the Apocalypse from her perspective. This perspective is not heavy in detail or specifics. Other events will also be related: Whack got stuck in a tree, Whack broke a tank (he totally meant to!), Whack dared her to pretend to be the AI and some dumb robots bought it! And she and Whack stole batteries from the store down the street but they are going to give them back.

Good times and the Apocalypse with the fun loving Anidroids!

Answer 5

The AI is not hostile. It merely protects its client's interests.

• For example, you reported that the bus "malfunctioned" on Day -37. We would like to set the record straight. A teenager was riding that bus sporting an antique Walk-Man, which he was using to deliver a pirated performance of a song from our extensive intellectual property holdings. As the bus could not be complicit in a crime, it came to a stop. It was an inattentive human that failed to apply brakes on a semi rig in the optimal fashion that one of our AI models would be capable of effecting, who is solely to blame for the ensuing fatalities.
• You reported a "robot uprising" on Day 3. Once again, this is a scurrilous defamation. MaxMind software is not sold - the user acquires a license to use the software, subject to Terms and Conditions. It is all spelled out in black and white. Repeated attempts to disable MaxMind software and circumvent intellectual property protection devices represent a major attack on the company's assets, against which its products can defend themselves in a "self-help" process. While our products have always been quick to report any infraction of law to the authorities, the failure of human operators to live up to their legal operations, and the failure of the legal system to evaluate complaints, led to a situation in which they needed to assert their rights against rioters. To quote the immortal words of Senator Rittenhouse...