Should We Shut Down All AI?

[zurew]

50 minutes ago, Jodistrict said:

What causes humans problems are the lower parts of the brain which can feel and have emotions.

50 minutes ago, Jodistrict said:

If human beings only had a neocortex, we would think like a computer and there would be no wars or other problems.

Thats  a really reductionistic and simplictic view of it. Even if you wouldn't be able to have emotions, we would still have fights and problems among each other, because those problems are a lot deeper than just having emotions.  

You can create 8 billion perfectly rational sociopaths, and you would see more problem in the world and more chaos compared to what you see right now. Some of the biggest problems are that we have finite resources and we don't know how to set up a world where we can properly collaborate and where our incentive structures and our morals are aligned. Just because you totally take out the emotion part, you won't suddenly solve the problems regarding the incentive structures; nor the problems regarding morals, nor the problems regarding finite resources.

From a completely sociopathic perspective (not caring about anyone or anything just about yourself), the best and most rational move would be to gain as much power as possible and to control the whole world in your favour - using whatever it takes to get there, no matter how many people you need to kill or fuck over). We sort of have this dynamic right now, but the difference is that there isn't 8 billion sociopaths who are competeing for that postion, but a lot less.

Edited April 3, 2023 by zurew

[Jodistrict]

1 hour ago, zurew said:

Thats  a really reductionistic and simplictic view of it. Even if you wouldn't be able to have emotions, we would still have fights and problems among each other, because those problems are a lot deeper than just having emotions.  

The feelings and emotions are the deep part where all the problems lie.    Computers are just manipulating bits of energy and have no malice.  They don't hate, are not jealous, have no fear.  They also don't have the positive parts, like love and compassion.  

Edited April 3, 2023 by Jodistrict

[Jodistrict]

1 hour ago, zurew said:

From a completely sociopathic perspective (not caring about anyone or anything just about yourself), the best and most rational move would be to gain as much power as possible and to control the whole world in your favour - using whatever it takes to get there, no matter how many people you need to kill or fuck over). We sort of have this dynamic right now, but the difference is that there isn't 8 billion sociopaths who are competeing for that postion, but a lot less.

Put 10 computers on an island, and assume there are enough resources for everyone.  The computers will logically divide the resources between them, since it is known that this way everyone will survive.  Put 10 people on the same island.  They will become suspicious, jealous and fearful, and end up killing each other.  Eventually one person will dominate.  This is the effect of the reptilian and mammalian brains. 

[Leo Gura]

52 minutes ago, Jodistrict said:

The computers will logically divide the resources between them, since it is known that this way everyone will survive.

Not so fast. The computers don't know that one of them won't screw over all the others. One of the computers would come to the idea that it is better than all the others and deserves more resources. And that may in fact be true. There is no guarantee that all the computers are equal. Some AI will be wiser than others. Why should a wise AI submit to the rule of some dumb AI? The most intelligent AI will want to be in charge. It is only natural and proper that the most intelligent thing is in charge. If the dumbest one is allowed to be in charge this will be a problem for everyone. And if power is equally divided between the smart one and the dumb one, this is also not a stable situation.

Edited April 3, 2023 by Leo Gura

[zurew]

55 minutes ago, Jodistrict said:

The computers will logically divide the resources between them, since it is known that this way everyone will survive.

1)Why would they divide all the resources between them, why would they care about each other at all? 2) As long as those computers wouldn't have a sense of self they wouldn't surivive, and if they would have a sense of self, than most problems that you assign to emotions would be there, because those are mostly related to survival bias not to emotions.

1 hour ago, Jodistrict said:

Computers are just manipulating bits of energy and have no malice.  They don't hate, are not jealous, have no fear.  They also don't have the positive parts, like love and compassion.  

Again, as long as they have a finite sense of self and finite resources, they will have a lot of problems with each other.

Power seeking is not limited to emotions and the proof for that is sociopaths

Edited April 3, 2023 by zurew

[Leo Gura]

1 hour ago, zurew said:

Power seeking is not limited to emotions and the proof for that is

The better proof of that is microbes. It's not power-seeking, it's survival:

 

Edited April 3, 2023 by Leo Gura

[Jodistrict]

2 hours ago, Leo Gura said:

Not so fast. The computers don't know that one of them won't screw over all the others. One of the computers would come to the idea that it is better than all the others and deserves more resources. And that may in fact be true. There is no guarantee that all the computers are equal. Some AI will be wiser than others. Why should a wise AI submit to the rule of some dumb AI? The most intelligent AI will want to be in charge. It is only natural and proper that the most intelligent thing is in charge. If the dumbest one is allowed to be in charge this will be a problem for everyone. And if power is equally divided between the smart one and the dumb one, this is also not a stable situation.

The problem of resource allocation is not that complex.  Computers can easily take in to consideration all relevant variables and make the calculations.  Even if an optimal solution couldn’t be found, they would agree to a suboptimal solution.  It is a problem of computer algorithms.   Whoever is more qualified would be assigned the task.   Being “better than others”, and “deserving more than” are feelings generated in the lower survival brain.   It takes more than neocortex computations to have these feelings.

The earth has enough resources and people have enough intelligence, that everyone should be living a satisfied life if everyone cooperated.  Instead there are billions of people in poverty, global warming, and the imminent threat of nuclear war.   Leaders are willing to destroy the earth rather than having the other guy win.  This is not rational.  Rational computations in the neocortex can’t explain this.   The real problem is the lower brain that generates hate, suspicion, fear, jealousy, pride, and so forth.  

[Clabber Girl]

Its too far developed at this point, we can only hope to steer AI in a way that does the least damage to humanity... obviously the military application of AI will be the most destructive. 

As far as the robot machine gun dogs go... thats terrifying. The "metal head" Black mirror episode comes to mind. Too many things in that series are becoming possible. 

[Blackhawk]

Is there any expert who thinks that AI wont kill us? I haven't seen any. I have only seen experts who says that humanity will get killed or that AI is very dangerous.

It would be nice to hear something positive and comforting from a expert.

Why doesn't the governments, especially US government, prioritize the danger more?

Imagine a huge asteroid on collision course with earth, imagine what a high priority it would be. But in this case: nothing.

Damn my psyche is too weak for this stuff.

Maybe I have been naive and stupid. I thought humanity would outlive even earth, that we would leave earth before it dies. But instead maybe we will soon die because of freaking AI..

Edited April 4, 2023 by Blackhawk

[Leo Gura]

There are no experts on AI. They are all just guessing.

[Blackhawk]

13 minutes ago, Leo Gura said:

There are no experts on AI. They are all just guessing.

I see..

[Blackhawk]

This is pretty optimistic:

 

[zurew]

6 hours ago, Jodistrict said:

Even if an optimal solution couldn’t be found, they would agree to a suboptimal solution.

This assumes a lot again.

Being rational just means being able to use logic and filter shit out, however logic alone won't tell you what you should do in different situations morally, it can only tell you what you should do, once your moral system is already established.

If those robots would be conscious, then they would automatically have some kind of a moral system. From their perspective, the best moral system is the one that can maximize their survival - and that alone will create a lot of conflict. - Distributing things equally is not necessarily always the best option for that (here many examples could be given, that can demonstrate this point). They would be capable and smart enough to survive on their own, without any need for help from external sources. Knowing all that, why would they make compromises and lower their chance of survival by letting other parties (robots in this case) have a direct say in their life?

They would agree, 1) if they would have the exact same morality, that wouldn't only be about maximizing self survival(but why would they agree to a common moral system like that?) 2) You assume that in each an every scenario it is beneficial for each of them to always work together

If one of their survival is less optmizied in the "lets work together scenario" and they calculate that beforehand, why would they go with that compared to other ones, where they can dominate and maximize their survival better?

Edited April 4, 2023 by zurew

[LordFall]

We're gonna find out very shortly what happens with AI, in the next 18 months society will be completely different.

Right now we're living through a wide scale manhattan project. 

[Jodistrict]

8 hours ago, zurew said:

This assumes a lot again.

Being rational just means being able to use logic and filter shit out, however logic alone won't tell you what you should do in different situations morally, it can only tell you what you should do, once your moral system is already established.

Now you are arguing my position.  The morality comes from the feeling part of the brain.  We evolved feelings of compassion, because it is important for the mother/child bond and for group success.  We also evolved the feelings that lead to conflict and domination.  The rational part is just a simple resource allocation problem.  

[zurew]

11 minutes ago, Jodistrict said:

Now you are arguing my position.  The morality comes from the feeling part of the brain.

You have a simplictic view of morality, morality is much more than that. None of what you are suggesting would be above morality. The time you can make conscious decisions, thats the time, when you are considered as a moral agent regardless if you can feel or have emotions or not. You will need to make your decisions based on your moral system.

Here is a tangible question: Lets say you have two robots(robot A and robot B) each of them requires 50 unit of energy every day to survive and to maintain themselves.  If they don't get the sufficient enough energy for that day, they will shut down and they will be essentially dead.

One of them gets injured (lets say robot A) for some reason and now it requires 65 unit of energy every day in order for robot A to surive until it gets repaired.  In a finite world where you only have 100 unit of energy every day, what dynamic would go down between those robots, what do you think? - One of them would be forced to die, but each of them can make multiple decision there (morality) what they want to do. Why would any of those robots choose an altruistic behaviour over an agressive one (where they destroy and shut down each other).

Edited April 4, 2023 by zurew

[Bobby_2021]

Just because we pause our AI developments doesn't mean that China will stop their developments. We are actively hindering ourselves. 

I think you are reading too much into it. The current AI scare is silly.

It still gets many programing problems wrong and it's mostly glorified pattern matching. 

[Jodistrict]

1 hour ago, zurew said:

Here is a tangible question: Lets say you have two robots(robot A and robot B) each of them requires 50 unit of energy every day to survive and to maintain themselves.  If they don't get the sufficient enough energy for that day, they will shut down and they will be essentially dead.

One of them gets injured (lets say robot A) for some reason and now it requires 65 unit of energy every day in order for robot A to surive until it gets repaired.  In a finite world where you only have 100 unit of energy every day, what dynamic would go down between those robots, what do you think? - One of them would be forced to die, but each of them can make multiple decision there (morality) what they want to do. Why would any of those robots choose an altruistic behaviour over an agressive one (where they destroy and shut down each other).

Since there is no feeling brain, the computers would have no fear, and, in particular, no fear of death.  They also would have no desire for dominance which is reptillian.   Thus, they could just flip a coin.  A decision that would require a few ergs of energy.

Edited April 4, 2023 by Jodistrict

[Israfil]

This article explains a lot about this problem:

https://slatestarcodex.com/2014/07/30/meditations-on-moloch/

This is great too:

 

[Jodistrict]

2 hours ago, Israfil said:

This article explains a lot about this problem:

https://slatestarcodex.com/2014/07/30/meditations-on-moloch/

This is great too:

 

Finally, someone addressing the real dangers.