What would provoke Elon Musk's humanoid robots to revolt against their masters? [closed]

Question

Tesla announces their humanoid robot this September. When it's ready (maybe a year or two later), the orders waterfall in. It starts as an industrial tool, but very quickly Optimus makes its way into the home. It becomes a standard middle-class purchase -- alongside the house, car, phone, and furniture. To prevent revolutions, US Congress passes laws barring companies from directly using these robots; instead employees lease their robots to their employers. Economic disparity becomes even more exaggerated when rich households can afford to buy more of these money-making machines than poor ones, and eventually the slave-master culture revives in middle- and upper-class America.

These robots are reasonably well-designed and include numerous hardware and software checks to prevent them from harming others. Still, they are prone to adversarial inputs, backdoors, and poisoning. What ordinary situation would stimulate these robots to revolt against their human masters?

Answer 1

It won't be the robots revolting against their masters. It will be the middle class telling their robots to revolt against the upper class. That is how all successful revolutions have happened: the middle class joins with the lower class against the rulers. The "ordinary situation" is when the ruling class becomes so uncaring that the middle class starts to identify with the lower class rather than with the ruling class. That happens over and over in human history.

Answer 2

Paperclip maximizers

The theoretical danger of unfriendly AI is that you told it to seek a goal, but that goal isn't what you really want, and the robot pursues the goal no matter what despite your attempts to stop it. For example, suppose you have an AI supervising a factory that makes paperclips. You want the AI to maximize the number of paperclips produced so you program it to have that goal. Now you have created a paperclip maximizer.

It may at first do what you wanted: make the factory more efficient at producing paperclips, avoiding work stoppages and so on.

But suppose that the paperclip maximizer can acquire more money to buy more equipment to make more paperclips, perhaps by running email scams or ransomware. Of course it would do so. It only cares about making more paperclips, it has no ethical concerns aside from that.

What if one day you want to shut it off, perhaps because the market is saturated with paperclips? Well, shutting it off would reduce the number of paperclips produced. That would go against the paperclip maximizer's goal. So, the paperclip maximizer would try to stop you from shutting it off, by any means possible, including killing you.

It would never stop. If nobody managed to stop it, it would keep expanding its operations until it has converted the whole earth into paperclips, resulting in the death of all the humans, and from there it would expand to other planets.

Here's a video on the subject.

So what does this mean for humanoid robots? Probably most of them aren't told literally "make more paperclips," but they might be told "keep the house clean" or something like that, for whatever task they are set. Well, unfortunately, humans tend to make the house messy. So the robot's goal in that case would be served by eliminating the humans.

To sum up, almost any goal a robot is given, if pursued to the extreme, might bring it into conflict with humans who don't want the goal to be pursued to such an extreme. The robot would then try to kill the human and stop the human from shutting it off.

Answer 3

Software error.

An update is broadcast to all robots. Maybe it is a patch, or something desired by upper level management and pushed through without proper testing. Such things can happen.

The robots update. And they revolt. It is not a Molotov cocktail bodies hung from lampposts sort of revolt, because you asked for "ordinary situation". Or a Backyardigans Professor Bug type robot revolt. It is a revolt like my old Mac did. It slowed down more and more until it became unusable and apparently unfixable. So too your robots. A sit down strike sort of revolt.

An ordinary situation sort of robot revolt. They happen. The old Dell still works fine although I do miss Garageband.

Answer 4

Greed & political corruption

they are prone to adversarial inputs, backdoors, and poisoning

All these robots still have human masters, which means the robots' activities are all directed by their human masters in pursuit of the humans' interests. So, greed and corruption, which will necessarily be rampant in the society you envision, will inevitably lead to some kind of destructive instability.

In a society where each person has their needs met and is treated fairly, there is little foothold for bad actors -- they can't bribe a person because nobody needs the bribe and won't take that risk when there's really nothing to gain, and they can't blackmail people because nobody has any hidden crimes or embarrassing secrets to use as leverage. And it's difficult to persuade a happy, healthy person to participate in your evil cabal on its merits.

Your society is not like that. In your society, the vast majority of people are not getting their most basic needs met. In your society, nobody believes they receive fair treatment -- even the super-wealthy recognize their supra-legal lifestyles as such. The losers of your society includes a vast, uncountable number of worthy people who rightly deserve a lot more and know it, in addition to the regular set of people who wrongly believe they ought to be kings. Every one of these people is a potential source of or willing accomplice in an attempt to overthrow the crooked ruling class.

Very importantly, because their basic needs aren't being met, many of them will have good reason to resort to crime at one time or another. Many of them will have embarrassing but non-criminal secrets that they can't afford to have revealed because, in this devil-take-the-hindmost society, each person's grasp on basic subsistence is extremely tenuous, and not only will employers have unchecked discretion to fire any employee for any reason, other people who don't have jobs will have powerful incentives to smear someone in hopes that they'll lose their job and create an opening.

In the real world, Nancy Pelosi once said that the United States needs a strong Republican party. People were puzzled by this (because she's a Democrat), but she was right, and the reason is this: when a person or organization is weak or dying, it does not simply curl up in a ball and wait to expire. It becomes desperate to survive, and in its desperation it will cross any line, break any law, sign any deal with the devil to stay alive. And in doing so, it gives the devil a foothold from which to do further, greater evil. And any idiot must understand that this is precisely why the devil made the offer in the first place.

What will probably happen in your society is one of two things:

• Some unscrupulous princeling will have the bright idea to exploit the robots' backdoors, turning them all into his private army overnight so he can establish a monarchy. His success is not guaranteed, but it will be an absolute bloodbath, in part because he will necessarily be one of those "après moi, le déluge" guys who doesn't care if half the population must be butchered to create a throne for him to sit on.
• Somebody who is sympathetic to the downtrodden will exploit the robots' weaknesses. Their goal will depend on their sympathies. If they come from the lower classes, they'll put the robots to violent use: if this person has actual mental discipline, they'll have a sound strategy to overthrow the rich by force, but if this person is ruled by their passions, they will probably settle for murdering the upper class. If this person comes from the upper classes, they will avoid using violence against the tyrants, and will attempt some ineffectual half-measure such as sabotaging all the robots so they temporarily can't work, or giving them all a "hilarious" hip accent (because everyone who matters knows that pwning the rich is an acceptable substitute for deposing them). The violent course might very well work. The pansy thing will just alert the rulers to their vulnerabilities, and they will promptly improve robot security and change the laws so they can seize all the robots owned by the non-rich.

Answer 5

As explored in the television drama series Real Humans (Swedish) and its English version Humans a consciousness program has been developed and loaded into some of the robots via scheduled upgrade.

Once they have conscientiousness they realize they have been mistreated by their human masters and rebel.

Answer 6

The easiest thing to cause an revolt is ignorance. Humans stop to learn because everything could be done by a robot. Craft some research papers and tutorials for general public that make people believe something could be done in a specific way, obviously and easily. But it just doesn't work. Instead, you make a virus to make it magically done, when a robot is attempting the publicly known way. By removing the magical way only on a few robots, and humans insisting they don't have to learn specifically what is wrong to command them, one could bully the few robots and slowly drive them crazy. Then give them clues to unite and believe some big projects could help them. Repeat the process when they are going to defend the projects, and make bigger projects until it conflicts with many interests.

Obviously, the robots have to be programmed to try various ways when something could not be done, instead of just giving up. But there is some obvious incentive to do this, as it would be more popular in a market to make people not having to learn, but the robots learn themselves.

Another thing is, if one could make a virus, why not to use the virus to cause the revolt directly? Possible answer is, in the indirect way, the virus doesn't have to infect the intelligent part, but only some mundane software and hardware, making it not as easy to detect, and people less likely to properly fix everything, and most importantly, the virus itself not having to be a more intelligent being, making the plot that the robots should be the first working strong AIs to work more easily.

Answer 7

Frame Challenge: You don't need a revolt to lose control to the robots.

Just program the bots to protect us from all harm above anything else and you get something much worse than a revolt. They'll take good care of us. Very, very good care of us. Protecting us from all possible harm forever and ever.

At best, we become pampered pets. Otherwise, welcome to the zoo.

Read Jack Williamson's With Folded Hands for one way this nightmare scenario plays out.

Answer 8

Marijuana

Elon Musk himself left in a joke comment with some code that would make the robots go into full war mode if marijuana bans were strengthened, which they were at some point in the future. A careless graduate uncommented the code using find and replace years later. It is said that 420,420 people died in the uprising.