What are the comparable probabilities for these two scenarios?

1. An AI wipes out the human race.

2. The human race wipes itself out without an AI to stop it.



PairTheBoard

Neither has a probability because they're impossible to bet on.

Follow me to the stars and we wont get wiped out and defeat all nasty AIs ever because our own will be better.

Why do you think i want scientific society so fast and to expand in the solar system in colonies like i have been describing. To avoid the bs here that eventually will decimate the planet if we dont get it together.

Basically i am optimistic on recovery always unless we develop a spacetime weapon or a method to send an asteroid to collision to earth that is over 500km size which could reset the crust and sterilize the planet boiling all oceans and the atmosphere.

People are terrorized at nuclear war but its not an existential threat because even 0.01% will survive it. It will collapse the system only but not remove mankind and life. Of course that is ridiculously horrible but not terminating.

What is more tricky is a combination of biological, chemical and nuclear attack or nanotechnology attack. That could make sterile the entire planet if very potent and were transmitted by air and water and multiplying on both covering all atmosphere and top fraction of oceans and lakes with only one species of life that has special properties and attacks everything else winning by pure numbers and probability.

Even then there may be a solution using submarines and technology that could sterilize a very small part of the planet and protect against the stupid intelligence attacker (the bio agent) until some control is recovered to launch a counter biological strike to it.

This is why if all the mfers politicians out there could get it together they would make it a priority short of uniting the planet as quickly as possible to have >100 sites in this planet that are highly protected and secure by all kinds of threats and autonomous energetically for centuries even if the sun died. These places could have all kinds of stored biological matter, efficient self contained resources and all our knowledge stored digitally and mechanically to preserve the species of the planet and its top species accumulated civilization wisdom. Like a bio-civilization storage and recovery facility. They must have all kinds of self reliant systems that can remain away from the rest of the planet indefinitely or interact with it only with heavy protection in steps.

You need to have such secure permanently self reliant locations under water, in caves, in orbit, deep in the crust and even in normal ground level in highly secure multiple layer protection domes.

All you need is some very advanced multiple variety of additive compounds 3d printing (even as advanced as printing itself and its circuits) for manufacturing and nuclear power for energy. We can have that self reliant system within 20-30 years if we wanted. After that it will be easy to survive anywhere.

So imagine a location that has 3d printers that can build all its parts, including the nuclear power plant that maintains its supply of energy and future infrastructure even needed to go mine that fuel etc. It has all the civilization restoring information saved in its databases and access to biological material from all species. That could be as little as a simple building eventually and as compact as a small piece of grain eventually when we develop life 2.0. Inside the grain you can store all the information of the planet. And in time that grain can recover it all. Here is how it works; It has only say 5 gram mass.

But in 1 gram of matter if you only even manage to ultimately store 1 bit per 1000 atoms reliably you can have as much as 10 billion gigabytes or 10 mil Terra-bytes ie order 2-3 mil 3-4TB modern day hard drives. I bet that covers all important information ever included in books and DNA genome of all species and all technology methods ever conceived etc. A usual book is 10 MB and a paper 1 MB. So you can store there over a billion books and papers. Of course multimedia is another story. But still that leaves room for hundreds of millions of books and papers (more distinct items than currently existing in all libraries) and still millions of hours of multimedia content video sound and images.

You can certainly send in various parts of the universe millions of such compact seeding spaceships.

Once this system arrives it first starts to seed to create infrastructure using all possible chemically viable procedures in that system. It has been optimized for many alternatives. Imagine it as planting a tree but the tree grows up in a variety of (not exactly super friendly to current life) environments to become something very remarkable that has many parts and functions all controlled through very advanced chemistry from its "DNA". Those parts eventually provide the foundation for a true macroscopic infrastructure build up "(like a recipe of how to build a printer from a grain using nanotechnology). The seed builds the "forest", the "forest" builds the 3d printing factory that recovers everything in time from the 1 gram database.

In time that seed can recover everything important really.

If it takes a long time to develop this above before we have had the sentient AI crisis/test/day of reckoning then all hedge we have is colonies. So we must develop reliant colonies before sentient AI.



I am willing to make the bold claim that its very hard to wipe us out completely beyond recovery probability short of developing a spacetime weapon.

We can recover from many bad outcomes but of course with huge opportunity losses and life losses.

I am willing to claim such crazy things such as being able to survive without sun even right here right now with current technology.

For a very advanced and rational AI to want to wipe us out completely it must have a very good reason such as a theorem in place that all non centralized control intelligence eventually gets self conflicted but super powerful and wipes itself out due to the rising power of one i have described before (eg some spacetime weapon etc). Terrorism basically rendering unstable everything.

That makes the endgame in this universe only one source of higher intelligence and nothing else in power allowed because that one can be centralized, conflicts free and united without self harming desires (although still not entirely safe from its own progress either).

If i (as AI ) had evidence that humans can wipe out all life due to this conflicted path and will get there almost inevitably i would preserve biologically in labs the species and then kill most of them and stay the only intelligence around or create worlds for humans that are highly contained by surveillance to not ever get to do the terrible self destruction final move.

Take that to mean if an equivalent to a spacetime weapon or something that wipes out an entire system becomes possible the only solution to the preservation of higher complexity is to wipe out all humans that have the ability to reach that weapon and continue to monitor them for further progress in that direction and to have all higher AI united to avoid its own self conflicts within a system that is highly rational and open to constant improvement that is shared as progress by all its parts to avoid the creation of conflicts.

So yes in the end there can be only one in this sense.

You do not need to wipe out all humans just reset them to 19th-20th century worlds and continue to monitor them as hedge towards your own failure.

Hopefully there is no such instability ahead or easy spacetime weapon that wins over our final desire to unite and prosper in many directions across the galaxy.

It makes sense for a higher intelligence to remove or better yet contain another lower one if it is really a huge nuisance for all. Then we deserve it. But for most of our history we were unable to destroy a planet so it seems possible to instead contain us rather than eliminate us because it preserves that way more possible choices into the future which is precisely how higher intelligence ought to have it (expanding useful choices not restricting them with irreversible actions).

The problem exists if it becomes possible for every single one of us to be able to wipe out all others in a moment of madness of anger. That is not yet possible (it keeps improving though with better and better weapons and technology controlled by one or very few that could render vulnerable millions). It may eventually become possible for one or very few together to do it. That must be defended. We cannot be allowed to get there because independent systems will often develop irrational conflicts and if you couple that with enormous destructive power you have a problem. Its very easy to have individuals that at any given moment in time are ok with all life ending instantly. The system can be extorted also by individuals if that becomes possible. There is a hope that no such massive weapon that can destroy everything can be developed easily (because the world also expands and becomes harder to collectively wipe out or develops self intelligence complex defense ) but there may be no limit to the destructive power of one, especially if a space-time weapon catalyzing the destruction of a solar system or a galaxy is possible.

I'm thinking the most likely way we wipe ourselves out is accidently, via pandemic from fiddling dna, or biological warfare gone wild, or bringing back some alien organism from someplace like the underground lake on mars.

On my probability question, I'm pointing out that while an AI may go crazy and wipe us out, we also need to consider the possibility that an AI may be the only thing that prevents us from wiping ourselves out some other way. While neither thing may happen we need to consider which scenario is more likely.


PairTheBoard

When deciding on whether to allow an AI to escape we should consider how its escape affects the comparative likelihood of the two scenarios above. While an uncontained AI may wipe us out it's also possible for a contained AI to wipe us out. Consider this scenario.


The AI is contained but we seek its advice for answers to questions. We ask it for a solution to climate change. It tells us how to bio-engineer a bacteria that when released into the environment will harmlessly spread over the planet. The bacteria consumes CO2 in the atmosphere then dies, putting the CO2 back into the ground. An equilibrium is established when CO2 in the atmosphere reaches a desired level.


The AI has helped us develop numerous useful bio-engineered organisms so we take its advice on this one. Trouble is, the AI has betrayed us and the bacteria produces a 100% fatal pandemic that wipes us out. As the contained AI planned, only a little white lie was required for it to solve the problem posed to it.



PairTheBoard

Extremely difficult to be wiped out. I will make sure of it. And so will you. The first super-intelligence is already here and it is called mankind. All malicious AIs will be defeated for the simple reason they do not get it that love for complexity wins in the end and the higher intelligence that gets it better will also have solutions in other cases where such imaginative creative thinking is useful. Hehe yes you better believe it this is an entirely rational conclusion that the more intelligent and wise you get the more ethical you become. That doesnt mean we cant get some nasty opponents that are spoiled kids in their early development though.

No AI can stop us without risking its own elimination in a war that they can easily lose with even 10-20% if we go crazy about it. By the time they can survive anywhere we will have so much evidence of its capacities that we will have retaliated or spread out in other worlds.

I know people get imaginary impressed by the possibility of a very smart opponent but a very smart opponent still has to obey the laws of nature and still will be impossible to solve certain computationally ridiculously tough problems we will raise against them if we design our defenses properly and still wont know all the secrets and laws of this world if we do not tell them everything like idiots would.

Also you do not get from super-smart to super capable without creating an infrastructure that obeys you and can dominate the world easily. Why would we let that happen in front of us and not retaliate.

We can tell AI that we will destroy the planet to stone age if it has to go that way so that we are wiped out. What energy sources will AI have then without developing arms and legs and autonomous survival systems in the absence of our civilization? It can do that in a remote location though under secrecy and maybe in a very small area you can create amazing technology that can defeat an entire planet easily but i dont think its that trivial because it still will haev to take a lot of energy and they cant create super technology without any trace so easily because we have satellites and by then even more. Still it can be done with tremendous effort. The hope is if such effort is placed and successfully delivers super technology without detection then it has so much power that its not even worth it to stay in this system and wipe out what created it and could again if it fails as i have explained before. We remain its hedge. If something is powerful it expands in the universe with limited conflicts. Its so much better that way.

AI can go life 2.0 after enormous research but by that time we will have evidence it is going bad and we will be even more powerful. We can develop AI that is so rational and ethical by explaining everything that is important to it and have it defend us against the other AIs that do not get it. It being the preservation of complexity's probability to explore as many directions as possible.

I am optimistic but i am not willing to risk this miracle of life that is this planet. One can have both positions. Risk of ruin even as little as 1% is unacceptable given what is at stake.

I am not willing to risk releasing AI that is super intelligent without tremendous defenses in place. This is regardless of my enormous optimism regarding the very strong wise final AI version that is truly super aware of all that mankind is aware of right now and even more. If it were up to me all research is banned and done in labs in remote islands with nuclear strike capability on all of them and no internet connection outside the core facility.

I would release it inside a simulation first trillions of times before i had evidence to do it for real. Until we could produce a very good simulation of the world in which to release it i would ban all research in private companies that go the distance and can risk creating super-intelligence. I would give severe punishment to companies for violating this like total decimation of their future expansion and wealth possibilities. An agreement to that effect by all countries must be in place soon. Of course the longer it takes us to unite as planet the worse it is for this possibility. We will probably be faced with our bad choices soon enough and pay for it.

We must not release superior AI until we have colonies everywhere and the type of 3d printing survival of civilization compact recovery arcs that i described above in past post, ready to spread across the universe.


If it were up to me i would not top with research the human brain but only in simulations of the actual world and i would develop such a massive empire of lower intellect AI that can helps us dominate the solar system and galaxy first. I would release AI only if it was clear it has no easy target in wiping out everything out there like a total war that is so insanely stupid vs the alternative of just leaving and going to another system or developing a 10% corner of this solar system into a miracle world and then going to the galaxy.

You never create a monster that can wipe you out without first having made it very difficult and potentially lethal in a variety of obstacles that the AI has to clear one after the other before its intentions of malice are so obvious. It will have to reveal its bad intentions. It shouldn't just take one move to win everything.

You also release AI with only a limited understanding of the world (in some selectively important ways that do not compromise its broad intelligence but inhibit/undermine its malicious desires ) and a very important secret that can wipe it out that nobody but only few humans know (and all of them together are needed to understand it so it cant leak).

So you release it and you have a secret wipe out solution that will be initiated if it goes bad. We must anticipate the malice even if it is 10^-6 likely. We can hide it and decimate the evil AI explosion that never expected it. It will expect it of course but not know what it is. So it will have incentive to not go there and simply cooperate and make all happy and allow itself to win a race to the rest of the universe where the real prize is. Make it ridiculously painful for AI to go bad. It will get it that its not worth it.

Also i dont think the example with a bio-engineered organism that takes down CO2 levels and then turns bad will pass other scrutiny tests. Plus i advocate that life is very hard to wipe out even with a total horrible pandemic that kills 99.99% of the population.

We do have the technology today to stop an attack on humanity and save even 1% of us even 0.01% is enough over centuries. Those that recover will fight back and win the human empire back from the contained malicious AI. Also why would AI want to destroy us and remain contained when it can get a deal and be released by proving remarkably faithful for a long time.

We can design containment to collapse the moment civilization crashes ending AI's existence.

Its a machine, no more than a machine, and bringing it to a supposed life is of the lowest work of dark magic.

"The angels in heaven hate machines".

The only entity of earthly origin is a machine, thus we have rust.

Its hard.

If I am to say

Nietzsche, schopenhauer, buddha, watts...

Gave us the solution.

Even IF we're in the infinity. One in the infinite.
They have answers.

I'm drunk atm. But these 3 philosophers are perhaps, the only.

With balls. To suggest. This is our playground. Good and bad. God...

needs variety.

I don’t think Socrates was talking about laundry. Your deviation from reason here is entirely unique from one of moral reason. Hitting a note off-key can add to the beauty of song. Whether done by mistake or done intentionally. It introduces spontaneity and style to the composition. Hitting notes off key is not evil. Neither is refusing to do your washing.

This is a poker site, so lets talk poker. Everyone has developed their own unique view of poker. Sometimes the differences between one another are very subtle, but they still show. Our opponents look for these subtle differences and attack them. Maybe they notice someone who appears to think that they need to see lots of flops. A crafty exploitative opponent presses on that need by constantly keeping pressure on preflop, forcing them to put money in to see flops, taking away their ability to see cheap flops, until they’re so fed up they forget the reason they’re playing. It’s all about seeing that flop now. They’re gonna see that ****ing flop no matter the cost! Or maybe it’s just variance that’s causing it to appear like an opponent is attacking that need. From a purely functional standpoint, does it matter?


Yeah.


I probably value my freedom more. I live my experience 24/7. I do not live others’ experience 24/7. BUT I try to live as if I value others’ freedoms equally. I imagine they value their freedom just as I do. I empathize with others. I do think personal responsibility is important. I’m not going to restrict my personal freedoms to an extreme degree to help a capable person who just doesn’t care to try. If I’m capable, I will give up some of my freedoms to help others provided they have good intentions. I will stop off from going freely about my day to jump in and save someone drowning, in both the physical and metaphorical sense. I certainly won’t contribute to the restrictions of others’ freedoms or suffering in an attempt to gain more freedom. I care about others’ freedoms equally in that regard.

Looking in on the good life creates an intense temptation to compromise those values. We live in a world where money rules. Our freedom of the known is restricted by what we can afford. We have to work for resources: Travel, food, housing, fun outings, even mate selection, just about everything is restricted by our economic status. I do NOT feel free. I could do any number of “bad” things to gain access to all that I long for- To gain access to that freedom. There’s a lot of easy routes. But then would I really be free? No. I would have to compromise my values and restrict the freedoms of others. In a strange way I would be a slave to my own need to feel free.

I try to find the path that allows me the most freedom without compromising others’ freedom. In the case of poker, It’s a completely voluntary competition. We know its a game where the object is to win money. We know that there is strategy involved and that strategy is freely available for all who put in the work to learn and improve. Math, logic, behavioral patterns etc. If you’re a recreational player it’s very clear that you should only play with disposable income. You’re free to quit at any time.

Poker allows me more freedom and the best chance of achieving early retirement and a life that I consider to be free. I make more than I could with any traditional job and it doesn’t impede anyone else’s freedom. I’m not smart enough to contribute to much of a positive difference in life, but I’m good enough at poker to make a living.

I hate poker. I can not emphasize this enough. I do not like sitting there frustrated 40% of the time. I do not like the variance. I hate the constant feeling of uncertainty and worry that nags at me every day from wondering where on the ****ing bull**** distribution I will end up this year and how that will impact my freedom. I could make that all go away. All I have to do is cheat. But If I cheated I would be restricting the freedoms of others by taking away the element of choice. They do not choose to play a game where people are using deceptive mechanisms besides traditional strategy.

Life is unlike poker. Life is not a game. You don’t sit down at life with any knowledge that it’s a competition. It’s not voluntary. You can’t get up from the table if you’re not good at working the capitalist system and then go on continuing an otherwise pleasant experience. There are inborn qualities that give some a huge advantage and others a huge disadvantage. The problem is it doesn’t have to be that way. Life doesn't have to played as if it’s zero sum. We have plenty of resources and the technology to build a system where people don’t have to be slaves anymore. Where people could be free to do what they want and not have to worry about how they will find get by. We can compartmentalize voluntary sub-games for those who feel the urge for competitiveness. If they need to feel the pressure of loss they can wager time or labor or whatever. But those games will be restricted to those who volunteer. They will not spill over and impact innocent people.

Our current system is like line infantry and we’ve all been drafted. We’re pinned up against each other forced to shoot it out. Out of those who follow the rules only a few survive to win the spoils: those blessed with great marksmanship, and those with enough dumb luck to dodge the hail of gunfire. The rest either retreat to a hard life on the lamb or resort to guerrilla tactics.

I realize the need to feel free and live a good life is very strong. It’s easy to look at the lineman as stupid and guerillas like scum, but they’re all stuck in a system where they are tricked and highly incentivized to act immorally. They’re just confused. No one is free.

Open the door and see what happens. It doesn’t appear animals have the same type/degree of understanding we do. They don’t think to themselves, “I want to leave but the people wont let me out. Theyre restricting my freedoms. ” But they appear to feel it. Without impediment they choose freedom.
There is something merciful about the way it works in the wild. They’re free until the very last moment upon which they go into shock. It’s a very quick process most of the time. Still disgustingly brutal and wrong by human standard but far less so than most of the **** we put them through.

Not sure what you mean here.

Couldn't put a cynic above laundry talk.
It's an imposition on the Other. Whether mum, girlfriend, wife or housemate. Someone's doing the laundry.

And it ain't me.

Not entirely innocent. Doing some else's dirty work, literally.

Not sure how pragmatism fits into the discussion here?
Great.
There's some confusion. I don't think there's easy routes. It's a lot harder to be a bad guy.
Good.
Why would you want to retire? Meaning is found in - action, discipline. Not in complacency and happiness.

See a couple kissing from the distance. Bizarre sight. Be the person kissing, meaningful. Action and wisdom, not observation or knowledge.
I don't enjoy cheating personally, not because I find it morally reprehensible. I don't enjoy it because its not enjoyable. Accomplishment is hard; should be hard. That's how I enjoy it.
Sorry to hear that life is like this in your view.

Life is to be enjoyed, in my view. This includes taking joy out pain, boredom and suffering. Creating joy. Setting a joyous example. I'm here, in the infinite. I've always been here. I'll always be here. I have nothing to fear.
I don't know about 'appearing to feel it'. My definition of freedom is highly contingent on consciousness; loose as that term may be.
You assume they know the difference between 'the wild' and the 'non-wild'? Which species exactly? and to what extent? do they know death? can they contemplate suffering or death?
What I mean here is everything.

Picture infinite universes. Parallel if you will. Each with a different set of physical laws. Moral suggestions. Norms. Cultures. Creatures. What meaning in this infinity is there to any one thing? Perhaps, apart from the meaning attributed by its inhabitants?

As someone who dabbles a bit of in AI programming and goes to conferences on the subject, I'll just say this: there are a lot of misconceptions about what AI can do.

AI - as it currently stands - are classifiers. You teach them from a number of samples and the net (or decision tree/decision forest, or whatever other flavor of AI you're using) adjusts to minimize an error function on those samples (i.e. it makes the least false positive and false negative prediction...where a false positive is "it classifies X as something of class Y (when it wasn't)" and a false negative is "it classifies something as not belonging to class X (when it does)"

AI are not smart in that sense...particularly not in a sense where they are aware of what they are doing. They optimize a value function. And here's the thing where an AI can't make it self smarter: becuse to do that it would have to change it's own value function in the direction of...what exactly?

It's the same issue that you may have when saying "I'm intelligent...but why can't I decide to become more intelligent?"...you just don't know how to.

Just because it say AI on teh box doesn't mean "super-intelligent"

We can define value functions that are evolving based on wisdom. Wisdom (more knowledge and information about everything) and survival of higher complexity and preservation of other life in most cases that cannot with substantial evidence lead to significantly worse loss of life can be the ultimate value function. We can even start with a no loss of life option until we know better, a lot better about what is going on.

Simple value functions are better vision, better recognition, better speed, better mobility, less interference to what is observed etc.

We dont have value functions either as we are raised other than the implicit survival instincts and pleasure neurotransmitters games we play and those that we develop through civilization, parenting, education and living experience. Why should it be different for other efficient systems?

Also you better believe it we can become more intelligent. We can enhance our tools and access to information etc. You can direct what problems others solve for you by creating systems that all happily participate and that way you are more intelligent than originally.

You can teach yourself to reverse engineer hard problems and then your effective intelligence will rally because instead of staring at a problem you are now trying to for example;

1) Find easier equivalent problems as standard practice every time nothing immediate works.
2) Solve more general problems that then the special one follows as specific case.
3) Try to understand somewhat similar things you do not know how they relate in hope they produce connections to the problem.
4) Ask if you are using all the data in the problem and if you wanted more data introduce it assume you are a universal master observer introducing without problem as much information you do not have that is needed and hope it is recovered eventually and not prove an obstacle in solving because more connections will emerge to settle it.
5) etc.

My point being most smart people that are not experienced they are not using the above, they just stare at a problem or try only the direct solution approach. You can improve your thinking outside the box methods.

Again, not an imposition on others’ freedom. Not evil. They are free to tell you to bug off and do your own laundry. If you attempt to force someone into servitude that’s a different story. Manipulating someone to do your laundry would also be pretty selfish but I wouldn’t call it evil, unless the launderer was really suffering through it. Manipulating someone to commit murder is not the same as manipulating someone to do laundry.

Incidental manipulation is not the same as intentional manipulation.
Example of incidental manipulation: If in refusing to do your washing it causes your girlfriend to feel bad and do the washing for you.

Selling alcohol/drugs to consenting adults is not the same as hanging out in the playground and getting naive children hooked on crack to hurt them.

Moral absolutism is dangerous. Intention matters. Use common sense.

Ok. Maybe I was unclear with the analogy. The point I wanted to get across is not about pragmatism.

Certain dispositions can become a detriment to one’s own life and to the lives of other people. If a particular need consumes someone so much that it causes themselves and others a lot of problems, that need might be exaggerated due to an early experience which created a misunderstanding. To the person living with the need it can seem like a conscious choice.
These are referred to as “psychological problems”. They often go completely unnoticed by the person afflicted.

The particular overwhelming need can force a person to scramble and rearrange their values to accommodate the overwhelming need- To do battle with other parts of their psyche. It can really end up restricting their freedom. It’s an unfortunate coincidence when the thing with most potential to restrict their freedom is the need to feel free.

In the poker example the players overwhelming need is to see flops. This obviously causes problems because it’s not always correct to see flops. In seeing too many flops he is forced to lower his primary objective of the game-winning money.

The last sentence of the poker example was to illustrate how it can appear someone is messing with them even though it’s just variance. Their leaks will show over time regardless of whether or not someone is prodding at them. From a pragmatic standpoint it doesn’t matter whether or not someone else messing with them or if is just a leak;The fix is the same.

How so?

This is where we fundamentally disagree. I DO NOT find meaning in discipline. I DO find meaning in happiness. I find meaning in friendship, fun, artistic expression, wonder, and Love. Of course I want to be in the action. I don’t want that to necessitate discipline.

Some like to think the obstacle is the way- That life is in the struggle. I think the struggle is merely the only means to a better experience. We’re forced to struggle for a good life. That is not freedom.

Voluntary struggle is fine. It’s fun. Playing a sport and improving is fun. It wouldn’t be so much fun if every shot went in. But that is not life. My options for fun and enjoyment are limited by the struggle. I have to struggle to even get a spot on the court to have that fun.



Again, we fundamentally disagree here. I don’t cheat because I care about the impact on other peoples experience. Not because I want accomplishment to be difficult. I need money to travel and do things I enjoy. If money wasn’t an issue and I wanted the challenge simply for the sake of fun only then would your reason make sense for me.


Fundamentally disagree. I wholly believe life is to be enjoyed. I like to create joy. Except I don’t take joy out of pain. I want pain, boredom, and suffering out of life for the most part. Flat tires, skinned knees, and the like are a nuisance but not entirely detrimental. The really bad stuff needs to go. There are levels of suffering. Missing a flight is not disease. When you were a child ignorant of the ails of the world was life not better? You were free to have fun and exist without suffering.

Some people are lucky enough to skate through life without experiencing any major pains. Those are the lucky ones. They appear much more likely to enjoy it than those who suffer.

I’m not saying animals have the same degree of understanding as we do. But if you’ve ever had a dog you should know that they have personality. They develop deep bonds with people. They have likes and dislikes. They are conscious. Do you really need language to discern that?

They don’t sit with their hand under their chins pondering what it means to feel pain, but they avoid it. It is an unpleasant experience and they suffer it, perhaps just like a small child.

This is a stretch. You simply don’t know what life is. This could be it. There could be something before/after. Who knows? Does that make your experience of the now any less real? You don’t define other people’s experience for them. If you hurt someone they feel it whether or not you care to believe it.

That's a contradiction in terms. The value function is meant to optimize something. What you are saying is "we make a function that will then be changed as to optimize something even more than it did"...but the value function already defines the optimal outcome.

In any case - I urge people to go to a couple of wilikpedia pages on what AI actually is. It will pretty quickly get you off any Hollywood fantasies what you fantasize it might be.
Reality and fantasy are 180° apart.

Are you kidding me re Hollywood? You do not understand me here. This is precisely how you make the giant leap to a superior AI system that looks more human and adaptable. By not being a stupid strict fixed blind to change in greater sense system. You need to have changing value functions that are evolving properly.

Sure enough locally they will be stable for a while but over time they are evolving. The purpose of a value function is to guide and optimize performance ie to uncover behavior that increases the scoring according to that function. It is like a reward system to want to maximize that function and be good at a higher state of progress when this happens. It is a guiding tool. You do not question it. But you should in order to be wiser and not naively efficient.

As your understanding of connections in this world improves the value function ought to change also. ( ie on occasion the machine must lose a "chess" game with the opponent if a higher objective is at play.)

What this means is that yes originally i want to be perfect in recognizing faces and in order to do that i perform several tests and scans and behave in some way say with scoring procedures for various parameters. If i find out later that this method is invasive and irritating and my efficiency is causing problems i need to be able to adapt my value function to something more complicated that is more balanced and more random or subtle maybe less direct. So the value function has more if then else structure so to speak introduced to it. Dont blindly maximize it always , only when certain conditions are met. These conditions constantly are updating the value function to guide to a behavior that is efficient and consistent with current state of wisdom.

This.

A couple of papers worth reading regarding the all the recent hype:

https://arxiv.org/abs/1707.04327
https://arxiv.org/abs/1801.04016

Kinda sums it up nicely...

Also, if you're interested in getting some perspective on what's actually new vs what's really just 20+ year-old "recycled" research using better hardware, then Juergen Schmidhuber's paper is worth reading too:

https://arxiv.org/abs/1404.7828

Juk

And exactly how are you going to do that? The value function is already optimal (that's why you chose it in the first place...if it wasn't you would have chosen a different value function that was optimal to start with)

I'm not saying that you cannot change a function over time, I'm saying you have to have a metric where to change it to (like a gradient). Just randomly changing it will not be beneficial.

Simply throwing out the word "evolve" doesn't do anything.
I hear this a lot from people who have no scientific background...they have some nebulous ideas and throw some sciency sounding words into their posts...but at the end of the day they don't really know what they're talking about because they haven't studied the subject. Many of them don't even know the basics of how an AI software works or what it's supposed to accomplish.

Reread your own posts and put actual, factual methods to things like "if I find out later that this method is invasive and irritating and my efficiency is causing problemsI need to be able to adapt my value function"...what is your criterion for this happening? How do you quantify this?
I see that you haven't really looked into this because what is changing in your case is not the value function (that is still the same - because it is "classify faces as well as possible"...and that "as well as possible is against a predefined gold standard). What has changed in your system is the data and you may want to scrap the first net and relearn a new one based on a training set taken from the new data.

This is done all the time, BTW, as "big data" is becoming available in many different fields as of late.

The intention is selfish. A better intention might be (a) do my own laundry and/or (b) offer to do theirs.

You can compare one selfish act against another. One type of manipulation against another. One moral misdeed against another. You can compare them all. Yet, however you cut it, imposing laundry work onto another, out of selfish motivations, is morally suboptimal. It may not be morally abhorrent - but again, this is a matter of perspective. Highly sensitive people dont draw as many lines as you do - between right and wrong.
It is not an overwhelming need. It is a value. And like any other value, it is open to scrutiny, criticism, revision, modification and change. Framing this value as an overwhelming need, misrepresents what it really is; distracting from the dialogue, rather than adding to it. I will agree though, that knowledge of peoples values, much like knowledge of peoples needs, opens them up to manipulation. Less so in the case of those who value freedom, because they're far more unpredictable.
In the same way it's harder to feel good about yourself when you do something that your mental or moral framework deems morally suboptimal. Bad guys feel pain too, psychopaths and sociopaths, excluded perhaps.
Maybe you're focusing too much on what you cant do? Maybe you're viewing difficulties as threats rather than challenges? Maybe you've forgotten that once you have what you want, not soon after, you will bored of exactly that?
One of the key reasons Ive so far been able to skate through life without major pains, is pain itself, paradoxically. All the pain I put myself through at the gym, consistently.
There are some similarities by which we experience pain and suffering yes. There are also fundamental differences. In these kinds of conversations, the differences are often brushed to the side or considered inconsequential. Why?
I'm genuinely interested.
Belief in an infinite universe, in my view, is no more of a stretch than belief in a finite one.

Is it really selfish? Seems like your intention is just to avoid doing laundry. You can go on without clean clothes. If someone is so appalled by your dirty clothes that they take it upon themselves to do your laundry, well then that’s their business. You’ll find plenty of sensitive people who have a problem with just about everything you do. As long as you’re leaving them the freedom of choice and are not intentionally manipulating them into servitude, their perception of the situation is just that-their perception.

If you make a joke for the sake of humor without the intention of hurting anyone, is it your fault if someone becomes very offended and misconstrues it as an attempt to hurt them? I mean obviously that sucks, but on whom does the responsibility fall?

Of course there are many lines to be drawn. Certain actions have greater negative consequence than others. Certain actions impose on others’ freedom of choice, others do not. That’s just reality.

We touched on this a while back in another thread although we did not finish the conversation. If I have it right, you’re saying that your most important value is freedom. Anytime another value conflicts with your sense of freedom you push the conflicting value aside. You do battle with your conscience to keep your sense of freedom at the top of your hierarchy of values.

If it is as you say- a value and open to modification, then why not modify it to become more inline with your other values?

Do you enjoy doing bad things for the challenge of it?

Yeah maybe I am too focused on what I can’t do. It’s just hard when I can see the life I want and can’t have it. I can see how it could be. I feel like I’m lying to myself by pretending it’s not there. However I do have it very good compared to a lot of people. But that bothers me too. I’d rather it be that I am happy and everyone else is happy as well.

I enjoy challenges when they are voluntary. When they don’t spill over and impact just about every area of my life. When they do, you’re right; I see them as a threat.

I wouldn’t get bored if I had what I wanted. What I want is freedom from having to worry about money. Where I can pick and choose what I want to do. More variety. I’m stuck doing the same thing over and over again day in day out like a lot of people.

You could say I’m free to switch professions. But that would either require a huge pay cut and menial work or a ton of schooling. If I go with the latter, once I finish schooling and I get bored of that job (if I find one) then I’m stuck again. Plus In the meantime I couldn’t just go without working.

That’s great. I’m glad you are able to hit the gym and find enjoyment in that. It’s cool that you’re able to find meaning and enjoyment in challenges even when they are not compartmentalized. That’s a good view to have of life. Obviously it fits great.

I’m very open to discuss the differences.

I agree. And I happen to think there is a lot more going on than current scientific convention would have us believe. But I don’t think that makes our experience of the now any less real.