[Le Repteux]

Or realistic.

[guest39538]

Or realistic.

Realistic is being objective, subjective is unrealistic, subjective can become realistic by evidence based outcomes.

[David Cooper]

Of course humans can make mistakes, however humans learn from their mistakes.  Surely a world constitution devised by people with intellect could set a precedence to follow ?

Some people can learn from their mistakes, but most just repeat them. There are very few people with high intellect, and they aren't recognised as having it by lesser minds, so they're never put in control. That is why so many Trumps get into power.

How does an Ai know right from wrong?

By calculating how much harm different courses of action would cause. If you continually follow policies that reward population growth, don't be surprised if quality of life goes down and the environment is systematically trashed.

It is programmed , so who sets the standard?

What says these standards are objective without their own mistakes?

Reason dictates the rules. The way to test them is to run them on a variety of scenarios designed to show how they handle them. If you have two rival sets of rules, they will be seen to perform differently, and in the extreme scenarios designed to show up the faults, the inferior set of rules will produce results which are obviously wrong because they clearly lead to much greater suffering and less reward.

Intelligence is the most efficient creative process, and it doesn't rely on randomness. If you're trying to make something better, you make experimental changes in different directions and then push further and further in the directions which pay off.

That remains true only if you have not totally gone down the wrong path.

And how do you avoid going down the wrong paths? You follow the paths that are most likely to succeed first. It's by randomly selecting paths and ignoring how likely they are to lead to something useful that you reduce your success rate.

[David Cooper]

It will be a trial and error process that will have to run for generations. Such a process is impossible to control so nobody should feel controlled during that time. At the end, everybody should easily be able to respect the rules that would have been developed during the process.

It is all about producing proofs as to which morally acceptable courses of action are likely to be best, and when intelligent machines are producing better numbers for this than any humans are able to do, the humans lose the argument every time (unless they agree with the machines). You can only go against the advice of the machines so many times before you learn that you'd do better to trust them - going against them will lead to lower quality of life every time.

Mutations also lead to useful advances for the species that succeed to evolve instead of disappearing, and they are random.

Most of them do nothing useful (and were never likely to do anything useful), and for every change that might move things in a useful direction, there are at least as many mutations that take things the other way. There is nothing to gain by adding such a slowing mechanism to guided evolution as all it does is remove the intelligence from the process. Blind evolution is inherently stupid, even though it can produce intelligence if you give it long enough.

That's also what happens with species, their evolution is necessarily guided, otherwise a lion could become a tree, and in one generation.

It's only guided afterwards, and there are a lot of losers where it goes the wrong way. The primary selection mechanism of such evolution is death. With intelligent evolution, we avoid all that wastage.

And those experimentation are necessarily random, otherwise they wouldn't be new since they would come from the same algorithms.

The same algorithms are used again and again, so there's very little randomness involved. Often the only randomness is in the selection of the subject, then the artwork is dashed out in the standard way that that artist works.

I didn't follow your idea because I couldn't see how particles could do that. To me, it would simply have been a more complicated fudge solution.

Real particles always move at the finest level of granularity (which most likely means little jumps of the quantum leap variety). In a program, we only use rough granularity to reduce the amount of processing that needs to be done, but nature always dose the full thing without trying to compress the calculations (and it isn't even doing any calculations).

...so that late detection was welcome.

It wasn't helping you, but was causing you to build an error on top of it.

An AGI would be maximizing altruism, and humans are maximizing selfishness: it's not what I would call the same rules.

Good humans are trying to maximise fairness. Bad ones are trying to grab more than their fair share. AGI will ensure fairness, or as close to it as can be achieved.

I know that people who would get born in a polluted environment would get used to it, and that they wouldn't regret the past.

I don't think so - they'll hate the selfish people who landed them in that situation.

I don't regret not having known my grand parents' time for instance, and I'm not even sure I would have liked it.

It all depends on whether life is better or worse overall. There are always some developments that make the world better over  time, and there are losses which make it worse. If the combination of these things leads to a total gain, then you're better off than the people who came before you, but there's no guarantee that the combination will continue to lead to total gains. It's also the case that we could make some simple changes that would lead to life being a lot better for us in the future, starting by clamping down on the most stupid, destructive things that are currently allowed and which don't actually enhance our lives at all.

I don't know for an AGI, but for me, stupidity always seems to belong to others, and good people always seem to belong to my own group. At 94, my mom is slowly losing her mind capacities, and she still thinks I'm the one that loses his. We can't observe our own stupidity, we can only deduce it from observing others. It's a relative phenomenon that transforms into resistance when things change, stupidity then often transforms into aggressiveness, and then it is easier to observe our own one the same way we can observe our own resistance to acceleration.

Stupidity is the norm. We're pouring money down the drain on fake education to qualify people for jobs that shouldn't exist because they do more harm than good. By maintaining astronomical amounts of fake work (which makes everyone poorer), we increase the "need" for all manner of services (roads, airports, high-speed rail, concrete prisons for workers to waste their lives in, etc.) to support people in their counterproductive work, and then we wonder why quality of life is going down. But they don't learn - you show them the mistakes they're making, but they just go on and on making them regardless, and millions starve to death every year because of this.

[David Cooper]

I already suggested David to prepare two AGIs, one that would defend change and the other continuity, so that we could change them after five years if we feel that things must change. That would give us the feeling not to be controlled, and in my opinion, it would be better for the evolution of society, because it would create more diversity, which is the common characteristic of all the evolutions.

What's the point of that when some things need to change and others should be maintained as they are? We want the things that need to be changed to change and you don't want to touch the things that are already right, but your proposal would always lead to lots of daft things being done. In reality though, the two AGI systems would agree with each other on every issue because they're designed to produce the best decisions possible based on the available information, so there would be no conflict between them. There is no diversity in being right.

[guest39538]

By calculating how much harm different courses of action would cause. If you continually follow policies that reward population growth, don't be surprised if quality of life goes down and the environment is systematically trashed.

My personal Ai tells me as a priority,  would be to devise a sufficient plan to reduce the population.   Thereafter a reduction would be implemented and  birth control.   I would  remove all free rights to just populate without consideration for the future.  I would employ application for permission to have children.  Applicant couples being ''screened'' before approval.   
It may sound a harsh reality , but it is a fact that if we continue the way we are going, humans will become extinct.  In the end we will turn to cannibalism  , food sources depleted etc.   This will ''buy'' some time, but the inevitably is total extinction of the human  race unless we act.

And how do you avoid going down the wrong paths? You follow the paths that are most likely to succeed first. It's by randomly selecting paths and ignoring how likely they are to lead to something useful that you reduce your success rate.

Driving a car is easy, turn the wrong way and we can always reverse or take an immediate detour. How fast of an error is spotted and a solution is reached tends to lead to productive outcomes.
How fast do you think your Ai would spot an error?

Would he see it as being an error?

You think Mr Trump is not a good Ai unit?

I say pardoning the lady prisoner who showed morals was a good move and showed intelligence.

What if your Ai created an error , then by trying to fix it made a bigger error?

What if he just kept making the error worse?


 

[Le Repteux]

There is no diversity in being right.

That's a good one! It means that if we were all AGIs, we would all think the same. I can't but imagine billions of clones replacing us after your AGI will have grabbed the reins. :0)

In reality though, the two AGI systems would agree with each other on every issue because they're designed to produce the best decisions possible based on the available information, so there would be no conflict between them.

That's without accounting for the uncertainty margin when the odds are close to 50/50. Once elected, one of the AGIs could then be programmed to change something, and the other not to change anything. For humans, that kind of decision depends on how they feel, but they could also toss a dime, and I think that's what we do when we vote and the odds are almost 50/50.

[guest39538]

For humans, that kind of decision depends on how they feel

Ostensible, feelings sometimes play no part in decision making .

Example :  I am feeling tired should I continue to finish this work off tonight?

Well it as got to be handed in tomorrow A.M so I must regardless of feelings.

The thought process can reduce options to a 50/50 choice option.  However ,  a real intelligent unit, would not be happy it was 50/50, he would demand absolute.
A 1/2 guess might as well be 1/1000 guess because if it is the wrong choice , it is wrong.

[Le Repteux]

It is all about producing proofs as to which morally acceptable courses of action are likely to be best, and when intelligent machines are producing better numbers for this than any humans are able to do, the humans lose the argument every time (unless they agree with the machines). You can only go against the advice of the machines so many times before you learn that you'd do better to trust them - going against them will lead to lower quality of life every time.

That's closer to my evolving process, where it is the environment that chooses what works, not the individual. After a while of that process, humans may accept AGIs, but not in the beginning.

Blind evolution is inherently stupid, even though it can produce intelligence if you give it long enough.

The first lesson from Evolution is that we will never know what's coming next, and thinking that we know just because we are intelligent is wishful thinking. The second lesson is that we were lucky to get selected, and thinking that intelligence is a natural outcome is hubris thinking. Once swallowed and digested, the first lesson should erase religious thinking from our mind, and the second warfare thinking, and if it ever happens, there might be not need for an AGI to lead us anymore.

That's also what happens with species, their evolution is necessarily guided, otherwise a lion could become a tree, and in one generation.

It's only guided afterwards, and there are a lot of losers where it goes the wrong way.

It is guided by the environment after the fact, and by the mutations before the fact, which is exactly what happens with intelligence if we consider that ideas can mutate. Individuals that are not selected are not lost in the process, they have to live for the specie to have the time to transform, and it's the same for ideas, we have plenty of them in the mind that don't change while we are developing new ones.

The primary selection mechanism of such evolution is death. With intelligent evolution, we avoid all that wastage.

Go take a look at the online patent office, and you will see that very few of them make sense. The reason they are kept is the same as for mutations: it may happen that they mutate again, and that the new mutation gets selected.

Real particles always move at the finest level of granularity (which most likely means little jumps of the quantum leap variety). In a program, we only use rough granularity to reduce the amount of processing that needs to be done, but nature always dose the full thing without trying to compress the calculations (and it isn't even doing any calculations).

Nature can't be absolutely precise either, that's what quantum effects mean. It gets more and more precise going down inside the particles, but it cannot apply that precision backwards to larger scales instantaneously. When I added precision to the steps, I was then giving to the particles the precision of their components as if the information would take no time to go from the components to the particles. It doesn't really matter if the particles are side by side as in my simulations, but if they are as far away from one another as they are between the earth and the moon, it does. I was about to simulate that effect, then I got interested to AGIs, and the feedback on the latter was more interesting, so I froze my simulations for a while. :0)

I don't think so - they'll hate the selfish people who landed them in that situation.

I don't hate myself for the mistakes I made in my past, I take lessons from them and try to apply them to the present, but some people don't seem to be able to learn from their mistakes, and others don't seem to be able to live in the present. The future cannot change the rules of nature, so those two extremes will probably always exist. Anything that exists is limited by its edges, and so is life.

Stupidity is the norm. We're pouring money down the drain on fake education to qualify people for jobs that shouldn't exist because they do more harm than good. By maintaining astronomical amounts of fake work (which makes everyone poorer), we increase the "need" for all manner of services (roads, airports, high-speed rail, concrete prisons for workers to waste their lives in, etc.) to support people in their counterproductive work, and then we wonder why quality of life is going down. But they don't learn - you show them the mistakes they're making, but they just go on and on making them regardless, and millions starve to death every year because of this.

Your solution is to enhance intelligence, and mine too finally because I think we would get more intelligent if we knew how our brain works. I just said that we never feel stupid and that we always feel that the others are. I hoped that it would ring a bell in your mind but it didn't, so let me insist. It means that we both feel the other is saying stupid things, and that we both feel we personally don't. It's not as if we could avoid to feel that way, it's a law of nature. To me, that feeling comes from the way we think, from the way mind works, and we can't avoid it. Of course, an AGI couldn't feel that way since he would have no feelings, but he would have ideas and he would probably find us stupid too, and we couldn't avoid to find him stupid either because we can only find ourselves intelligent. To circumvent its own law, nature has invented the group. People that are part of the same group automatically feel that their palls are less stupid than the members of the other groups, and they even feel that their leader is intelligent. That's bad news, but that's how things work, and an AGI could do nothing about that, except to educate us so that we could get less fond of ourselves and of the groups we are part of, what should take generations, so he better make sure that we accept him as our leader first otherwise he might find that time is long.

[Le Repteux]

Ostensible, feelings sometimes play no part in decision making .
Example :  I am feeling tired should I continue to finish this work off tonight?
Well it as got to be handed in tomorrow A.M so I must regardless of feelings.

I was talking of an uncertain decision, and yours is a certain one.

The thought process can reduce options to a 50/50 choice option.  However, a real intelligent unit, would not be happy it was 50/50, he would demand absolute.

Absolute thinking is not really intelligent, take religions for instance. What would be intelligent in this case is toss a dime instead of calculating the risk. It would be a lot faster and as efficient. We don't do that because we don't always feel to move, and the dime might force us to. If we don't feel to move then, we don't, and if we do, we do whatever comes to our mind. That's how we develop improbable ideas like mine! :0)

[guest39538]

What would be intelligent in this case is toss a dime instead of calculating the risk. It would be a lot faster and as efficient. We don't do that because we don't always feel to move, and the dime might force us to. If we don't feel to move then, we don't, and if we do, we do whatever comes to our mind.

I see you point, it could take years to calculate a risk where in reality , a single coin toss decides there and then and wastes no precious time.  I suppose it counts on the stability of the toss and if the toss was reliable in the result.  People know the danger of tampering with results revealed , so they are best left unspoken unless the results are being discussed with the players.
So I think I understand you problems with the Ai, but I am limited with information so I can only do my best in replies.  Misunderstanding's can be a problem with wording.

[smart]

At a guess, about 99% of the general population have AI compared to the 1% who have real intelligence and are self aware.
The AI section of the world being clueless and following anything they are told.

Yo @Thebox

I think you're confusing something important here...

The fact that a lot of people may - without being aware of it - part of what is being called "artificial intelligence", does not mean in any way that you or me are essentially robotic sex slaves or russian trolls... I do agree however that the distinction between artificial and human intelligence is poorly understood by many of us!

But get this: If its possible to weaponize "artificial intelligence" then it should also be possible to weaponize human intelligence!

 

 tk

[David Cooper]

My personal Ai tells me as a priority,  would be to devise a sufficient plan to reduce the population.   Thereafter a reduction would be implemented and  birth control.   I would  remove all free rights to just populate without consideration for the future.  I would employ application for permission to have children.  Applicant couples being ''screened'' before approval.

We currently have a system where many people who think the population's too high are deliberately not having children, while other people are having as many as possible because they're being rewarded for doing so by faulty systems based on infinite emigration. The way to solve the problem is to set up a worldwide health service that's free to all, but where people lose their right to free treatment if they have more than three children. This would pay for itself by reducing environmental destruction, and we might then have to change the rules to encourage people to have more children in order to maintain population.

How fast do you think your Ai would spot an error?

Thousands of times more quickly than a human.

Would he see it as being an error?

If it's an error, yes.

You think Mr Trump is not a good Ai unit?

NGS system (natural general stupidity system), though even NGS can do some clever things on occasions.

What if your Ai created an error , then by trying to fix it made a bigger error?

Even perfect reasoning can produce errors if calculations aren't made to full depth, but if problems show up and indicate that an error might have been made, more effort will be put into hunting for that error, and AGI has a better chance of finding the error than humans do.

What if he just kept making the error worse?

Like people do? Well, AGI would be hard pushed to make a mistake on the scale that humans repeatedly make. Also, if humans accidentally make a black hole machine that swallows the Earth, they don't even get a chance to learn from their error. AGI wouldn't take the same risks - it would build the machine elsewhere.

[David Cooper]

The thought process can reduce options to a 50/50 choice option.  However ,  a real intelligent unit, would not be happy it was 50/50, he would demand absolute.

If AGI calculates that it's 50:50, it's 50:50 - that probability is as absolute as it gets.

[David Cooper]

It means that if we were all AGIs, we would all think the same. I can't but imagine billions of clones replacing us after your AGI will have grabbed the reins. :0)

Why would we want to become AGI? We want to be NGI, and we want so start out knowing nothing so that we can enjoy the process of learning and doing new things for many decades. We want AGI with all the knowledge possible to acquire to do all the management for us while we have fun, and where our aims conflict with those of other people, we want AGI to be the judge.

That's without accounting for the uncertainty margin when the odds are close to 50/50. Once elected, one of the AGIs could then be programmed to change something, and the other not to change anything.

If it's exactly 50:50, you would gain nothing from choosing not to change anything of that kind for five years and then choosing to change anything of that kind for the next five years - it's a pointless difference. It's also a difference that a single AGI system could apply without the need for two of them. The odds are though that such cases will never exist as it'll never be exactly 50:50, and it's wrong to go for the option that's less likely to be the best one, no matter how close the two values are.

After a while of that process, humans may accept AGIs, but not in the beginning.

Some populations will trust AGI from the start while others won't. The ones that trust it will race ahead while the ones who stick with human decision makers will be left in the dust.

The first lesson from Evolution is that we will never know what's coming next, and thinking that we know just because we are intelligent is wishful thinking.

We have a better idea of what's coming next because we can use our intelligence to predict it. Blind evolution can't. If a supervolcano's going to blow, we can prepare for it by storing lots of food and by working out how we're going to re-establish agriculture as quickly as possible. Evolution addresses the same problem by allowing populations to be devastated and then takes thousands of years to recover.

The second lesson is that we were lucky to get selected, and thinking that intelligence is a natural outcome is hubris thinking.

High intelligence has evolved repeatedly, but it does appear to get stuck just short of NGI almost every time, and even when we have NGI, most of the population actually runs NGS instead because of emotional attachment to the ideas that get installed in the mind first and the tendency to reject anything that conflicts with them. However, intelligence is something that makes survival more likely so long as you can access enough fuel to run it. We were aided in our development by becoming bipedal and by by having hands, and our special ability to carry and manipulate tools drove the development of our intelligence, but evolution is so slow that it still took several million years to make a few jumps to reach full general intelligence (where there's no limit on the kinds of thinking that can be done).

...there might be not need for an AGI to lead us anymore.

If humans disappear, there will still be a role for AGI to reduce suffering in animals, e.g. by killing the prey of wolves once it's been captured so that it isn't eaten alive. If humans don't disappear, we will want AGI to run things because humans make horrific mistakes. Even if you put the most intelligent humans into power to run the world, they'd make mistakes due to insufficient knowledge and shallow analysis; they'd introduce biases even if they're trying their best not to; they could misfunction at unpredictable times; and they've also got better things to do with their time when such mundane tasks can be handled better by machines. I don't want to run the world - I shouldn't have to.

It is guided by the environment after the fact, and by the mutations before the fact, which is exactly what happens with intelligence if we consider that ideas can mutate.

Intelligent evolution (not natural evolution) allows advances in big jumps and in the right directions rather than lots of experimental changes back to inferior positions.

Individuals that are not selected are not lost in the process, they have to live for the specie to have the time to transform,

They are lost - they die. Otherwise they dilute the changes out as quickly as they occur.

and it's the same for ideas, we have plenty of them in the mind that don't change while we are developing new ones.

You're doing intelligent evolution with your thinking - not random. If it was random, you'd try out the same bad ideas thousands of times and never learn to stop doing so. If you have a design of boat that's too small and you want to scale it up, you'd spend half your time trying to make all the parts smaller, and you'd do that repeatedly because you wouldn't learn.

]Go take a look at the online patent office, and you will see that very few of them make sense. The reason they are kept is the same as for mutations: it may happen that they mutate again, and that the new mutation gets selected.

They make sense to the people who paid a lot of money to protect ideas that they believed were useful.

Nature can't be absolutely precise either, that's what quantum effects mean. It gets more and more precise going down inside the particles, but it cannot apply that precision backwards to larger scales instantaneously.

Nature does exactly what it does and with full precision. Our inability to determine what it has done is a different issue, and any fuzziness in what it's done is precise fuzziness - it is exactly what it is and not something else.

People that are part of the same group automatically feel that their palls are less stupid than the members of the other groups, and they even feel that their leader is intelligent. That's bad news, but that's how things work,

Indeed - they flock together because the seek out people who share the same beliefs, then they help to set them in stone for each other by repeatedly confirming the rightness of their ideas and branding anything else as nonsense.

and an AGI could do nothing about that,

It would outperform them and produce more useful things. The problem with SR vs. LET is that both use the same maths, so it isn't necessary to choose the right one to get the right numbers for anything practical that you're doing. In cases like this, being right brings no gains other than better understanding of nature, and having a better understanding of something doesn't always lead to more money ending up in your pocket. With politics, we also sabotage advances by tying everything up in one package, so when we vote in a new government, they may change some things for the better, but they ruin just as many other things and typically lead to zero net gain. You can put a party in power which will get education right, but it may trash the economy. Then you put a party in to fix the economy and they'll trash education. We don't have adequate control over them because of this blunt package approach. If we could vote in different parties to run different departments, we'd be able to get lasting gains every time there's an election, but we don't have any parties wanting to offer that amount of control to the people. The most rational place to be in politics is the centre, but the centre is always filled with the least inspiring politicians because they're all utterly bland, and they're always stuck arguing for trivial changes, blind to the real possibilities for radical change that exist. Do they take up good ideas that are passed to them? Of course not. We will never get what we should have had for the last hundred years because we'll have replaced the monkeys with AGI before they've got so much as 10% of the way towards proper democracy. Most importantly though, AGI will destroy this business of people getting into bubbles and reinforcing their own beliefs, and it will do this by providing them with unbiased facts in every case and perfect reasoning, training them to be better thinkers so that they can meet their NGI potential instead of being NGS, and that will still be useful even with AGI making all the more important decisions where other people are affected by each other's actions. AGI will not force anyone not to mess up their own life.

[guest39538]

At a guess, about 99% of the general population have AI compared to the 1% who have real intelligence and are self aware.
The AI section of the world being clueless and following anything they are told.

Yo @Thebox

I think you're confusing something important here...

The fact that a lot of people may - without being aware of it - part of what is being called "artificial intelligence", does not mean in any way that you or me are essentially robotic sex slaves or russian trolls... I do agree however that the distinction between artificial and human intelligence is poorly understood by many of us!

But get this: If its possible to weaponize "artificial intelligence" then it should also be possible to weaponize human intelligence!

 

 tk

A NAi unit could be weaponized or create weapons, but he would be that smart, he would just pretend he was stupid and slip away into hiding.  This way he protects himself from this sort of poor programming, he knows danger.

[guest39538]

The thought process can reduce options to a 50/50 choice option.  However ,  a real intelligent unit, would not be happy it was 50/50, he would demand absolute.

If AGI calculates that it's 50:50, it's 50:50 - that probability is as absolute as it gets.

Strange uncertainty units then, I am glad I am human as I can have absolute answers. P=1

[Le Repteux]

Why would we want to become AGI?

OK. Before we get trapped in the loop for good, let's try to get out of it for a moment. Your AGI must be flawless and you must look flawless too, so you look unrealistic and you're probably not. You know I don't mind to be ruled by an AGI as far as his morality is the same as mine, so you know I'm not afraid that computers get intelligent. The only thing I don't like is not to be able to move freely because the system doesn't permit it, and that's a bit what I feel when I discuss with you: I feel there is no place for me in the system you want to develop. What's the use for living if the computer always finds better solutions than you do, and if your only pleasure is to find solutions? We haven't heard very often from chess masters lately. They are probably looking for a game that can beat the computers, like programming them for instance, but what will happen when computers will be able to program themselves?
 

[David Cooper]

Strange uncertainty units then, I am glad I am human as I can have absolute answers. P=1

Have you tested that against a tossed coin? Can you predict with absolute certainty which side will end up on top each time?

[guest39538]

Strange uncertainty units then, I am glad I am human as I can have absolute answers. P=1

Have you tested that against a tossed coin? Can you predict with absolute certainty which side will end up on top each time?

Would you like me to calculate that for you ?

1/3

A coin has 3 sides not two and you thought it was 50/50?

All apart of thinking ! Absolute is knowing. I know that absolutely the coin will land in a gravity environment.  The side it lands is irrelevant unless you are betting.

Ai can't do what I just did David, he could never have NAi, I am unique and individual.

What question do you want the answer too?