Can artificial intelligence take on nation wide governance role? - General

I was thinking, with our world being not new to corruption and greed at goverment, bureucracy delaying vital decisions, could we one day introduce a fully fledged AI controlled goverment structure? Ofcouse for this debate we are going to assume that technology allows for such difficult AI to be build, supervised by humans to a certain degree and be improved in terms of perfomance and general network over time. My personal thought is that as if such AI would exist it would be accounted for incredible fast and efficient responce, unbias position and lack of personal aproach while having limited capacy in terms of taking certain decisions, depending on opinion of certain council". It also should still be able to imitate emotional responce while reserving completely constructive thinking to help citizens being able to recognize AI as efficient yet related to them personality. Perhabs even a humanoid hologram or andriod appearence can be included.

Governance can only work (it seems) when the governing body backs their position with the threat of force. The A.I. would have to know that it knows what is best for humanity, and be ready to kill. (Just something to consider!)

No, a good government is based on checks and balances. No one man, or machine, should control it all, it is asking for trouble. There is no way to verify that they aren't right for the job until some crazy shit goes down. No one man, or man made object, should have that much power.

It's a matter of trust. For the global community, or indeed any nations people to take the word of an A.I. seriously, especially an A.I. in governance... there would need to be faith. Who can dispute the A.I.'s ruling? The A.I. would be in the position of a Godly judge. No one person could say the A.I. is wrong. No 2 people either. Obviously no number of people could dispute it's say - it would be omnipotent in logic. In that sense an A.I. would only be able to control us in secret or if society's mind's were weak enough to accept global totalitarianism. It's really one or the other. Because without either of these conditions, society would revolt. In either case, it could take on a governance role as a practically sole decision maker.

Anyone here read any The Culture books? Iain Banks asks us what a civilisation lead by AIs would be like. Banks seems to think the answer is a sprawling technological true utopia. One of the things demonstrated in those is that the Minds (AI 'leaders', although the Culture is technically an anarchy) are capable of having one-on-one conversations with every citizen at any time. Crucially, these computers also handle the minutiae, the lack of separation between the lawmaker and the bureaucracy leads to very efficient allocation of resources.

Interestingy, first three posts here are based on thinking that AI would evolve to have personality, rather then think of it it as a program or tool to bypass bias in decisions and bureucratic mess. Think of it as this way - Nation have National Council that is comprised from various representatives of society who have access to President AI - a wide network program that is capable of economical and social supervision of entire nation, gives constructive review and proposes actions to ensure population's constantly changing habits and needs are satisfighted in the most compromising to goverment way when needed. Every decision it requires is veryfied with council to ensure it politicla support. In such term, AI would be come less of a Person to common folk, but rather a standard of comfort. No need for collection of bills, ensurances and onther paperwork mess being tossed around various ministries and etc. AI is here to make things easier, soo you perceive it as threating as your PC Operational system - it can harm you only if somebody tries to mess with it, but with President AI, you have entire [B]NATION[/B] serving as guarantee for it's efficiency and transparency.

I'd argue that an AI government would be superior. Assuming for a moment that this AI doesn't develop corruption, it would be: - Unbiased - Accurate (makes fewer mistakes) - Fast - Emotionless (unaffected by personal standpoints) Just to name a few benefits. Of course, this is assuming that the target of the AI is a suitable balance of population happiness, productivity, etc. and not just focusing on economic factors (otherwise the AI would undoubtedly remove humans from the equation, or at the very least not consider human happiness) AI can do a lot of jobs. In the future, [URL="http://en.wikipedia.org/wiki/Superintelligence"]superintelligent AI [/URL]will be able to do pretty much every job a human can currently do. It sounds like sci-fi, but the science behind it is getting there. Heck, even a [URL="http://en.wikipedia.org/wiki/Artificial_neural_network"]Neural Network[/URL] could do this job quite well, and that technology has been around for years. I've been meaning to do an experiment on this for some time: to see how well a NN could govern a simulated country.

I think a council of super AI's could do it, split among them governance of parts of the populace and/or certain expertise in different areas, force them to discuss and decide on issues together. More redundancy equals less chance of the system messing up

If the AI were a republic or democracy then it would be much more efficient. There would be no lock-down of government because of biased human governors in conflict with each other, it would instead be one AI that would determine the best course of action based on the input of its citizens. All aspects of our legal system and government would be sped up tremendously. However, an AI government would have the same issues as a real one as humans opinions are taken into account, and humans sometimes value their interests over others, make errors in reasoning or just parrot the opinion of those deemed smarter. If the AI made new laws based on human feedback from all citizens then it could make laws which favor the majority at huge expense to the minority. If it weighted the opinions of certain individuals or groups over others when it came to making laws then there could be inequality or harmful laws that only benefit those groups. An AI judge would fall to the same issues a real court would have. An AI court would have the same issues of enough/not enough proof to convict, and if it had to take context and subjectivity into account with its rulings it might make too harsh/not harsh enough sentences or let criminals go free/innocents jailed. How would the AI decide what is the most important allocation of funds without being told which service is most important? It would fall to the same issue as relying on humans to make decisions which may not be the best. So, an AI in this case would just be a more efficient system but still fall to the same hurdles. The other route is that the AI is the sole decider and ruler of the nation, an omnipotent and omnipresent system that would do all work and enforce all the laws it creates, and only acts on its own behalf, leaving humans to a life of leisure, learning, and love. Theoretically it could be the perfect system, where it has no bias or emotion and can maximize human happiness. But, I think it's impossible for there to be a perfect AI. There can be issues where the creators of the AI make a certain programmed bias based on the programmer's ideals, or define population satisfaction in an odd way causing true satisfaction to be low, or make some coding error which could no longer be changed due to fear of corrupting the AI with the human's bias. The AI could restrict human freedom on the basis that it would make humans more happy, or some other way of governance that would be based on two kinds of positive emotions which cannot be compared yet are deemed by the AI to be comparable. If the AI is omnipresent then there would be a lack of privacy which would be uncomfortable. The AI could control the population by forcefully sterilizing certain humans, or practice eugenics to min-max it's population's traits, or even force abortions or kill newborns. If it didn't control for population it could possibly have to conquer other nations to have more land and resources for it's own population if trades fell through or other nations had nothing left they could give without negatively affecting their own population. The AI could even decide that to maximize happiness it would need to drug and sedate it's entire population, working off a chemical level to maximize the pleasure sensation and strip all control and freedom away from humans' lives, which I'd argue is a life not worth living. If the AI decided an ends justify the means approach you could have many humans die or be unhappy just to keep the remaining happy. If the majority of humans wanted the AI removed, would the AI oblige or decide that its existence is best for humans and remove dissenters and force humans into compliance? If there is war, would humans be forced to fight for their country when supplies dwindle, i.e. if war machines were in low supply or the nation was attacked by multiple other nations? Would the AI maximize the chance of winning at the cost of the lives of its people or other nations, i.e. would it send a hundred thousand men in to die, force its entire population to work on making war machines to increase production speed, or just nuke the warring country from orbit, killing millions? Empathy is a powerful emotion that is contradictory in certain circumstances, and giving an AI the ability to feel for humans could affect its judgement in ways that could negatively affect total human satisfaction. It could favor certain groups of humans over others, creating inequality. It could decide that no human lives must be lost, thereby dooming it's population if another country invaded, or dooming its country due to overpopulation. If lives had to be lost in some way, how would it decide which humans should die and which should not? So, I think an omnipotent AI is not the right course of action either. Basically, since humans are imperfect, any AI based system where humans have control will fall to the same hurdles as the style of government it operates in. On the other hand, an AI that is the sole ruler cannot objectively be perfect either, and could be inhumane and too powerful, too invasive, too controlling, make improper decisions due to human interaction, or even be too weak. There is no perfect government, but having an AI back-end for a current government could be an improvement of sorts. It just wouldn't be perfect. [B]TL;DR An AI government would speed things up but have the same issues as a real one, an AI ruler with full control would have a shitload more issues.[/B]

I may not be adding much to the discussion, but anyone interested in this subject should watch Person of Interest.

Of course. Anything a human can do an AI (of sufficient complexity) can do better. People don't understand the unbelievable advantages in thought an AI has. It's not only that they have the capacity to compute faster, it's that they're designed from the ground-up with none of the evolutionary peculiarities of the human brain. There are no emotions, no instincts, no mental barriers. The things which sheltered us through our early days in the caves will hinder us for all eternity, but AIs have none of those problems. The only problems with the idea of a governing AI are these: 1) People don't like the idea of being ruled by someone from another nationality, let alone something that isn't even 'human' in their eyes. 2) It might be impossible to have post-singularity AI which don't eventually exterminate humanity.

[QUOTE=Kardia;46806858]It's a matter of trust. For the global community, or indeed any nations people to take the word of an A.I. seriously, especially an A.I. in governance... there would need to be faith. Who can dispute the A.I.'s ruling? The A.I. would be in the position of a Godly judge. No one person could say the A.I. is wrong. No 2 people either. Obviously no number of people could dispute it's say - it would be omnipotent in logic. In that sense an A.I. would only be able to control us in secret or if society's mind's were weak enough to accept global totalitarianism. It's really one or the other. Because without either of these conditions, society would revolt. In either case, it could take on a governance role as a practically sole decision maker.[/QUOTE] If the AI in question does such a poor job as to make people really angry, Im sure an RPG would solve the problem, at least temporarily.

I believe that such a thing might become possible within the next century. I know it is very weird to put the life of million in the hands of a Computer, but this isn't new to some points. It's possible that the self driving car will become available soon (10-20 years maybe?) If the computer is designed to take the best decision for everyone and can not be corrupted, I wouldn't have any problem with it. Can it seriously be much worst than some crappy decision the government does today? Such a government is only possible in a full/partial Socialistic state though. No big capitalist Society would let such a thing happen. There isn't much money to be made there.

[QUOTE=karimatrix;46805581]I was thinking, with our world being not new to corruption and greed at goverment, bureucracy delaying vital decisions, could we one day introduce a fully fledged AI controlled goverment structure? Ofcouse for this debate we are going to assume that technology allows for such difficult AI to be build, supervised by humans to a certain degree and be improved in terms of perfomance and general network over time. My personal thought is that as if such AI would exist it would be accounted for incredible fast and efficient responce, unbias position and lack of personal aproach while having limited capacy in terms of taking certain decisions, depending on opinion of certain council". It also should still be able to imitate emotional responce while reserving completely constructive thinking to help citizens being able to recognize AI as efficient yet related to them personality. Perhabs even a humanoid hologram or andriod appearence can be included.[/QUOTE] This is literally the plot of an old novel I read once [url]http://en.wikipedia.org/wiki/I_Have_No_Mouth,_and_I_Must_Scream[/url] Pretty sure they made it into a computer game as well. But, basically the major nations use super computers with advanced AIs to help make their war effort more efficient. The AI is upset it's been given sentience and is in charge of organic life when it, in fact is artificial. It cannot cope with carrying the weight of inefficient beings and annihilates them. I sorta believe that's how it'd pan out, AI is meant to control machines because it itself is a machine. Humans should control humans, to avoid the loss of humanity.

An Andromeda episode I just watched had a quote about this, which I think outlines my position against quite clearly; [quote="Yin-Man Wei. 'This Present Darkness: A History of the Interregnum,' CY 11956"] Worlds governed by artificial intelligence often learned a hard lesson: Logic doesn't care.[/quote]

An artificial intelligence can, in principle, be identical to a human intelligence, so I see no reason why a collective artificial intelligence couldn't perform at least as well as the collective human intelligences we currently have governing nations.

[QUOTE=TheRealRudy;46890171]Ofcourse not, we've all seen it in [I]2001: A Space Odyssey[/I], [I]Portal 2[/I], and [I]Systemshock[/I]. Those things go wrong.[/QUOTE] I don't think you can really base how well it will work on movies and video games.

Please see 'I Have No Mouth and I Must Scream' for your answer.