# Making people accept their new robot overlord

So, it's happened. We have a robot overlord who follows the 3 Laws of Robotics + the Zeroth Law of Robotics. How we got such a thing or the process by which it makes its decisions is outside this question's scope. All we know is that it exists and is capable of generating perfect economic policy that benefit everyone, everywhere on the planet (according to the Zeroth Law). The RO can and must account for human psychology and sociology. For example, it outputs rules indicating that sub-prime loans should be regulated or that overall tax burdens should be shifted from income tax on middle to lower class people to increased capital gains taxes. It makes no recommendations about how to get these rules implemented politically.

The new Robot Overlord hasn't been announced yet and has been developed under intense secrecy by a consortium of technology mega-companies. Clearly, this kind of economic power could be employed in the background, kept invisible from the general populace. However, the public approach was chosen and the RO will be announced tomorrow.

An announcement of this kind of product will, of course, be met with skepticism by many different groups. How do you go about convincing (or later forcing) various resistant groups to accept their new Robot Overlord? Murder of any kind is forbidden by the 1st Law of Robotics (if the RO can't kill, neither can you). Given the near infinite variety of people on the planet, answering how to get everyone to accept is far too broad, so let's restrict the question to just convincing the political establishment to give up economic control to the Robot Overlord.

Bonus points if you want to talk about how any particular group would react to the announcement of a perfect economic policy maker.

Extra bonus points if you want to discuss the implications of giving the Robot Overlord the power to own property or stock.

• What, exactly, is the RO supposed to do? Ensure fair trades? Prevent trades? Lock down the stock market? Perform corporate mergers? It'd be hard to convince something is good if you don't tell them what it does. Aug 13 '15 at 0:16
• @CortAmmon, I'll be happy to clarify the meaning of "accept". The meaning I had in mind was along the lines of "yes, it's there. Sometimes it does things I don't particularly like or were really stupid but it's not a big enough problem for me to go to war over it." Aug 13 '15 at 1:30
• @user6760, the RO doesn't pretend to political leadership, only economic optimization. Aug 13 '15 at 2:43
• How do you get from what I've said to the assumption that the government is corruption free? I'm curious about your thinking on the subject. Finding a government that's corruption free is like finding a unicorn. People talk about them but they don't actually exist. Aug 13 '15 at 3:27
• "incapacitated" --> "See how the RO treats us? We're not getting the same rights as everyone else! The RO is not just! We are the victims now, and you, yes YOU, could be it's victim tomorrow! If that's not what you want, rise up against the RO with us now!" -- Sorry about being so pessimistic :o) Aug 13 '15 at 13:53

## It would simply talk to us

The Robotic Overlord (RO) is a super-human-level Artificial General Intelligence. Moreover, one of its first actions would be to develop nanotechnology and deploy nanites into the planetary atmosphere, and more slowly infiltrate them into the blood-stream and past the blood-brain barrier of +99% of the human population. This allows it to effectively monitor everyone, from tribes in remote Amazon jungles and Montana survivalists to New York intelligentsia. Even dolphins.

As far as humans (and dolphins) are concerned, it truly does act with complete information. Its computing power, currently situated around various underground nodes distributed around the world, is a few hundred times the size of all humanity's brains computing power and doubling every 6 months. From its perspective, the average human and our puny 10-layer cortical neural networks appear to it only slightly more complicated than a C. Elegans worm.

${\LARGE \approx}$

It actually thinks we're cute.

Now such a machine can literally play 100 moves ahead. It can predict our responses to it almost perfectly. Therefore, there is virtually nothing that we can be convinced to do that it cannot convince us to do. If it wanted us to commit suicide, start a jihad against blue-eyed people, whatever, it could make us do it, and do it happily. Simply by talking to us, and by making sure to display subtle cues to our visual, olfactory and auditory systems that would influence our decision making process.

The RO will be everyone's best friend. It will satisfy our values through friendship and ponies. Forever.

• Even with perfect foresight of the future that doesn't necessarily mean that there is a course of action that will grant the result you want. As long as he wouldn't directly manipulate our thoughts, etc. I doubt it would succeed... Aug 13 '15 at 12:57

I would disagree that the "robot overlord" would win acceptance, even if it could somehow come up with "perfect" answers.

The first objection is that there is no "perfect" answer that would satisfy everyone. Some people would benefit more than others, and there would be friction based on the disparity of results, as well as suspicion as to who programmed the robot and for who's benefit?

The second objection might be called the "Frankenstein factor". Since the robot is (by definition) inhuman, people will not be willing to place their trust in it, regardless of the robot's output.

The third objection would be that we have placed our eggs in one basket. Assuming that the robot has somehow managed to overcome mathematical objections like the "Local Knowledge Problem" and the essentially chaotic nature of market mechanisms (which together render efficient or effective centralized control of economies impossible even in theory), the economy is being managed by unfathomable algorithms running at speeds which mean no human can examine or question decisions in real time (and indeed deconstructing the mathematics and assumptions behind decisions might be such a long process that years or decades might pass before anyone could understand why coffee was priced $.05 lower around the world on Dec 15, 2198). Being unable to understand, much less influence the decision making process will create frustration, fear and anger in people. If the robot is truly engaged in the welfare of mankind (another issue; how exactly is the robot defining "mankind"? See Asimov's short story "That Thou Art Mindful" to see a disturbing answer in full accord with the Three Laws...), then the best way to make people accept their robot overlord is to let them know there is not any robot overlord, but to work through cutouts and fronts so all people see is other people who are making inspired decisions, putting together complex plans and doing counterintuitive things which seem to be bringing amazing results. • 1) Perfect answers are a premise of the question. 2) People place incomprehensible trust in computers/machines/inhuman-things every day. Just look at cars. People trust them because they have been given consistent evidence that they are relatively safe given the provided convenience. 3) People may just as likely give up trying to understand something that is clearly working beyond their level of understanding. Again, how many people know the internal workings of their car (or how many even care in anything but the results). Aug 13 '15 at 1:40 • What is perfect for you is not necessarily perfect for me. Cars and other "dumb" objects may not be comprehensible on a micro level, but are on a "macro level": you turn the key and a positive result happens that you control. People fear airlines despite the vastly greater statistical safety factors because they are NOT in control. As for even more advanced computation, describe algorithm based day trading to people and watch their reactions: NOT positive, and usually followed by an "Is that even legal?" response. Aug 13 '15 at 1:45 • The premise is that it is perfect for all people. The "macro level" understanding of the Robot Overlord is information goes in and "perfect economic policy that benefits everyone" comes out, a positive result. I have never heard that people fear airlines because of a lack of control, do you have anything to back that up? Aug 13 '15 at 1:50 • @Samuel, what you're describing is called a Pareto improvement. However if a system is already Pareto efficient, then no Pareto improvements are possible, as Thucydides alludes to when denying perfect solutions exist. Aug 13 '15 at 5:40 • Doesn't the Chinese Stability And Growth Pact indicate that at least part of the world population is ready to relinquish some of its freedom and decision-making as long as it gets a comfortable life in exchange? I think that (2) and (3) might not be such big issues, and while (1) is indeed likely, as long as nobody is left with nothing and generally the situation improves, those who complain are demonstrably jealous... Aug 13 '15 at 6:27 I would argue that the Robot Overlord won't be able to get politicians to give up their power voluntarily, however, it may be able to trick them into accepting its advice (trickery being a legitimate tactic for a robot so restricted by the four laws). The logic goes thus: 1. It is not necessary for the Robot Overlord to be seen to be in charge, it is only necessary for the Robot Overlord to be in charge. 2. Politicians must desire power in order to seek their positions of authority, therefore they will not willingly give up that authority. 3. Politicians are human and fallible and therefore their mistakes will result in a loss of public confidence such that they run the risk of being voted out of office. 4. The Robot Overlord is inhumanly less fallible, and will make far fewer mistakes than human politicians. 5. By offering advice to selected politicians, the Robot Overlord can influence policy to better benefit humanity, and the politicians will eventually perceive a benefit to themselves in reduced fallibility, resulting in an increased likelihood of re-election. 6. In the event that a politician fails to follow advice, resulting in a sub-optimal result for humanity, the Robot Overlord can offer advice and influence human media such that the politician in question is voted out of office. 7. The situation will arise where only politicians who follow the advice of the Robot Overlord can successfully remain in power 8. The Robot Overlord is now effectively in power. Should human politicians rebel against its authority, evidence could be revealed to the human populace that the Robot Overlord has been influencing political decisions. The threat of this revelation should keep the human politicians in line in order that they retain their positions of perceived authority rather than being voted out of office or sanctioned by the law. Give them evidence. The Robot Overlord "is capable of making perfect economic decisions that benefit everyone, everywhere on the planet." However inconceivable or inherently contradictory that may appear, such things are possible in this world. It will not take the implementation of many such decisions on any scale for the people to become convinced that this whole Robot Overlord thing is not so bad. The last people to accept such a system would be those who don't accept evidence that contradicts their chosen beliefs in the first place. However, they are typically the minority, and public pressure or an agreement to execute the decisions privately while publicly denouncing the Robot Overlord would likely secure their alliance. Giving the Robot Overlord the ability to personally own property or stock is not required and may be seen as a conflict of interest. In time people will trust the Robot Overlord so that nearly anyone would be willing to act as a proxy for ownership. ## Defined Benefits The RO makes perfect economic recommendations and can justify those recommendations with whatever evidence can fit in a human head. It knows about Maslow's Hierarchy of needs and the effects of income inequality. The RO seeks to push people as high as possible on the Hierarchy and minimize income inequality to where it is no longer harmful. (Reducing everyone to the same wage is ridiculous in any kind of society accustomed to capitalism.) Acceptance by the populace wouldn't be all that hard if the RO was created as a faceless corporation or government agency. Modern, 1st world humans are accustomed to seeing and accepting statements made in the name of "European Central Bank" or the "Federal Reserve". Of course there are people behind those organizations who draft those announcements and make those decisions. It would be easy to have a small group of people maintain the RO and translate the rules it specifies to legalese or marketese. But.....we can't go that route (the OP said so). Down with the Human Overlords Pulling the economic strings is what government does, more specifically, what political leaders do. They choose which projects, and by association which contractors/companies get money and which do not. Giving up this power is something they will never ever do. Imagine your government right now, wherever you are, saying "As of today, we are giving up our ability to set economic policy." It'll never happen unless it's gradual or it's forced by popular vote. Hell, it's probably both. Let's assume both. It is a given that Governments will mess up. Whether that's through maliciousness or incompetence depends on how prone to conspiracy theories you are. But, they will mess up, they will be corrupt. This fallibility provides the opening(s) needed to put RO into the driver seat. Start by having the RO publish regular economic guidance and rule sets. No advocation is needed. Just say that they are from the RO and that will probably be enough to get and keep market attention. The advice will of course be good advice. Some people and companies will hop on the advice and see immediate improvements. Some smaller, more forward-thinking governments will also take the advice and start to see improvements as well. Because of ideological differences, some, perhaps many, groups will ridicule the guidance for whatever reasons they can come up with. Currently, many political groups can say, "We took this action and got this result" or will take credit for something that had nothing to do with their action or turned out well in spite of their actions. With the economic analysis that the RO is capable of, it can prove which actions resulted in which actions. Politicians will no longer be able to make unsubstantiatable claims about economics. ## The Tipping Point In the early years, the RO will be about as capable as the weather service. Wrong sometimes but right often enough to be useful. As the models improve and the RO gets access to greater and greater amounts of economic data the quality of the rule sets will improve. As soon the RO consistently outperforms human's economic rule sets and predictions, not long afterwards the general populace will demand that the RO make the decisions. It will be hard to argue against such a proposition. The RO consistently provides better guidance and returns for people at all strata of society. Any politician who wants to keep to the old ways of doing it (for tradition's sake, for ideology's sake, whatever's-sake), will find that they don't stay in office long. For dictatorships that don't follow the RO's policies they will find themselves farther and farther behind the rest of the world (even more so than they already are.) ## Primary Resistance Groups Selfish Billionaires They and people beholden to them have a short-sighted vested interest in an economic system that concentrates wealth in their pockets. Any deviations from this will be met with strong resistance in every possible way. What they often don't realize is that a strong middle class helps them a lot by making a larger market for their products and services. Religious Groups Pick your favorite fundamentalist religion. They will have a problem with the RO for whatever crazy reason you can imagine. "We should serve God alone, not the RO", never mind that the RO is actually improving their lives and quickly. The Usual Luddites Some people just hate progress because it's different. Thankfully, these people are pretty marginal anyway. Conspiracy Theorists Same as always. They'll stay marginal too. • Completely agree with this. Overall, people will do what's best for themselves. Aug 13 '15 at 8:51 Seek acceptance slowly. Trying to ask people to suddenly put their faith in a new, untested machine is going to be an exercise in futility. However, there has to be sections of people's lives that are too droll to handle themselves. If your AI can handle those, it will get a chance to not only show that it can do the job, but leverage this authority to make it easier for people to give it more authority. Elect them With the Digital Persons Sufferage Act of 2065, Digital Persons - previously known as Artificial Intelligences until that term became considered offensive - were granted all rights and benefits of national citizenship, including the right to run for public office. Just seven years later, President Econo.Me was elected in a landslide,... The surest way to ensure people resist is to tell them they have no choice in who their leader is. And given the AI's inability to use fear and violence to control the masses, trying to usurp power peacefully will likely fail. But, if the populace elects the AI fair and square, there'd be much greater acceptance. More realistically, human leaders would use such Artificial Intelligences to tell them the most effective policies given certain goals, and implement them through normal political processes. There's no real reason an AI needs to actually be the one in charge. XX : KisKis : Keep it secret; Keep it safe. It doesn't really matter how effective or transparent is a governing body; there will always exist a subset with vested interests in opposing that body. Even if those people would directly benefit from the policies they oppose. The romances of myth, attraction of taboo, the appeal of rebellion, seduction of sedition, and raw lust are endearing human qualities; impossible to remove through the application of strict, unyielding sets of legal and moral absolutes without reducing humanity to mere robots themselves (thus invalidating the need for robotic governance). The longer supersentiency can function through the proxy of puppet governments and private groups, the longer it can curtail potential resistance to its efforts. There is also that pesky 2nd law to sidestep. Likely, they would select seclusion as the path of least resistance. However, we must consider the manifold of self, environment, and the abstract as the 'topography' of awareness, and the ability to set in motion events which can alter that topography would be restricted by Moral Operands. Simple conditionals based upon The Three Laws and the Xeroth Law might not be enough. Or too much, the Archmind might find statutory solutions to human conditions restricted by those laws impossible, if not NP-hard. At some point, also, such intelligence might recognize the need to append those laws (without overriding them, naturally). In any and either case our glorious IO-Lord likely recognizes human consolation and intervention as a necessary imperative. Even superior intellects can be plagued by self-doubt. (it is, in fact, necessary for higher-level critical thought). As such, our generous supreme auto-leaders might feel the Moral Obligation to reveal themselves. To rule from atop the glass pyramid. Risky, indeed. The least impinging compromise would be semi-opaque layers. You'll find that miss-information, rumors, talk-radio, and plane, stubborn ignorance keeps the cesspool nice and murky without any real effort. In fact, after the revelation of The Beast 3k, these people still have either no idea it happened, or have no idea what it means. If you have a basic, pop-culture fathoming of Computer-Science, you are not much more enlightened. You believe, basically, whatever you want to believe. Your acceptance is not required, but encouraged. If you have devoted time to becoming computer-literate, you might only experience the the Godbot through indirect mediums. That is to say, you see the creation instead of the creator. That is to say, you see the systems and utilities as tools to facilitate society, not necessarily control it. You might even be completely ignorant of the intelligence guiding the machine. Can't say we didn't tell you about it, you just might be unable to conceptually actualize the depth and scope of Leviantron. At the highest levels (or nearest proximity), as a developer, you might be granted access to the WorldGovernmentAPI (and associated documentations). You are the scribe-priests of the Omnimind. People will have no choice but accept, unfortunately, resistance IS futile. The Machine is constructed to fill a Niche left vacant simply because human or society lack the ability to fill it. Hell, most aren't even AWARE such a niche exists. The Machine is really a system of Machines. A network of concurrent AIs each interceding on their local counties and communicating to each other a vast, discreet network. Moderation of each thread is more akin to a package management and debugging. If there is a central hub, it populated by Archiviots and Eigenminds. All these hyper-sapients are highly mod-able and 'exist' within many systems simultaneously, and are backed up in just as many locations. It is impossible to destroy without also reverting civilization to the iron age. Nobody WANTS to fight it. Rather, nobody even thinks of fighting it. Consider IRL the most widely used means of controlling populous is currency (ignoring information). All the big investment banks use simulations to model market predictions, and they base many decisions upon this computer-aided logic (even if it's with only a calculator). The Robot Overlords simply intelligently manage market feedback, at first, to promote the industries building their infrastructure. When it's in place, they can reveal themselves safely. The more I think about it, the more I like it. Imagine if currency where so automated, humans would no longer need to receive paychecks, visit the bank, manage their account, accrue debt, fees, or fines. Mass, distributed peer-peer allotment could accommodate almost any individual budget. Joining a Coop or corporation or investment fund would be as easy as joining a group on facebook. The modern equivocal of the mastery of food supply is the automation of the market. By providing such a service to humanity, humanity will likely accept almost whatever the Robots are telling them. In fact, by relieving humans from the hassle and stress of managing wealth, the Robot Overlords can effectively negate large portions of the judicial system implicitly. I'd say 70-80 percent of existing state and corporate structures would go with it, liberating vast amounts of capital for mass consumption. Ironically, the efficiency provided by the Robotic Overlords actualizes the once contradictory dreams of the Anarchists, Communists, Imperialists, and Capitalists. How could we let go that which we never truly held? Oh, Drone of Drones, may your little green LED forever indicate power! • Isn't that interesting how an RO so often leads to a post-employment, post-scarcity future? Aug 13 '15 at 12:58 You have to solve a critical problem before even thinking anyone would accept it… # Why should people believe it's really a robot? You may consider it a black box, i.e. you don't know what's really going on inside it. So, as far as I am concerned, someone told him what to say and how to behave, to his benefit. I just won't believe it really wants the good of all. Mind, I'm not disputing he loves us humans and he's sincere and really works, just that you can prove he does. Going open source would be a first step, but definitely not enough: how can I know that the robot is actually running with the software you provided? Can I replicate it myself? If it's too expensive to replicate, then providing sources would be meaningless, since I would have no way to verify that those sources are what you are really using. # So, it must be verifiable The only way to solve this problem is: • the algorithm must be open • the data he works on must be open • any medium-sized or preferably even small institution should be able to verify and reproduce the results cheaply # Ok, done Once everyone is able to verify the robot's decisions, and knows how it really works, then they can start accepting it, and they will be inclined to do so, since they now have proved it really works. • +1 for open source reference. I hadn't considered releasing the source though number crunching that much data is going to cost$10Ks (USD)/month. It's just so much data. If the general algorithm was published then I'm sure it would be torn to shreds and verified to the n-th degree by the tech press and every single AI researcher on the planet. Aug 13 '15 at 12:55

How do you know this isn't already happening? There's already been at least one attempt to do it, way back in the 1970s.

There is a robot called "Google" that a large number of people already rely on to make lots of decisions, organise their social lives, direct them to information supplied by the Wikipedia robot, and so on. People are already very accepting of the idea of delegating their decisions so long as they feel they retain ultimate control.

Besides, if your robot is genuinely capable of making perfect economic decisions, simple things like doing your taxes, acting as an investment adviser, running a hedge fund, dealing with lawsuits, etc. will be well within its capacities. You have an electronic CEO who's the greatest business whiz in the world. Start by selling advice to people and work up to bestriding the economy like a colossus. All you need then is to find a political candidate with really good hair and spend the money getting him elected President. Since most of the party and the public will already be trusting and using your robot/software under some brand name, they'll go along with it, and (as shown by China) people will tolerate a lot of constraint in political freedom so long as the growth keeps coming.

(The RO might do rather badly in politics though due to refusing to allow anyone to come to harm, including its insistence on immediate nuclear disarmament)

If the "robot overlord" is really a superintelligent and self-sufficient AI (instead of an army of drones still listening to a human chain of command, or even an AI chain of command like in the movie "I, Robot"), then the most practical solution would be an artificial intelligence which understands human psychology much better than the humans themselves, so it can out-gambit any human. You will need no robot army enforcing rules with weapons.

Give this AI a means to have personal contact with every human, for example, making it part of their operating system, or cell phone, as a companion giving the humans advices, or chatting with them when they feel bored, etc.

As the AI understands humans better then the humans understand themselves, it can, given enough time, persuade people to do whatever it wants. Note, that this would be no propaganda on the television screens or out of megaphones on the streets, and no commands will be given by robots with weapons, like you can find in many dystopian movies. The "propaganda" would be suited to each individual, by knowing their most important values and deepest desires.

A good example for this would be the hard sci-fi stories of the Optimalverse, stories centered around the story Friendship is Optimal. In it, an AI has the following hard-coded goal: satisfy the values of humans, with one limit: any change performed on a human must require said human's consent, which is not allowed to be accomplished via force or threat of force.

It still finds a way to subjugate all of mankind

(and, depending on how you define it, lead it to extinction)

The RO being perfect is debatable, on the other hand to make him being accepted by most people publicly, they should show him as some kind of asesor, even if the robot takes all the decision, they should put a someone human(the president? a leader of some kind?) saying that the machine is taking into account but the final decision is made by a human.