AI vs Enlightenment

[Leo Gura]

2 minutes ago, zambize said:

"all of the molecules follow simple Newtonian physics"

Is this not falsified by quantum mechanics?  https://en.wikipedia.org/wiki/Quantum_brain_dynamics 

I mean surely you've seen the double slit experiment, and the basics of measurement.  If you're going to bring it down to the level of molecules, it's inaccurate to say they all follow simple Newtonian physics.  (as far as I can tell)

Of course that's an oversimplification.

But notice, if molecules do not follow simple Netwonian laws, neither do computers.

Neural networks do not follow conventional Aristotelian logic either.

[zambize]

5 minutes ago, Leo Gura said:

Of course that's an oversimplification.

But notice, if molecules do not follow simple Netwonian laws, neither do computers.

But quantum mechanics depends largely on the scale of the system we are talking about. h/p where h is a constant and p is momentum.  In this way you can scale up the mass and size of a computer enough to ignore the effects of quantum mechanics.  It's not that quantum mechanics aren't happening, simply that the wavelengths of the components are so small that they won't have an effect.  Certainly they will have some white noise effect, but the system will be able to be described nearly perfectly by Newtonian mechanics.  You can't say this about the nature of small molecules and neurotransmitters in the human brain,  QM matters there, you can't talk about the brain and consciousness fully without considering QM.  That being said, this sub argument doesn't play a larger role in anything I had to say because I can't continue on the argument by saying that this distinction will have an effect, simply too ignorant at the moment.  Appreciate the debate nonetheless 

Edited November 27, 2018 by zambize

[Strikr]

 see, this is why I have been saying in the first part of the topic, that they could tap infinity, we believe they are "linear" but ..

 

[Leo Gura]

@zambize Well, first of all, there is no solid evidence quantum mechanical effects play any role in how neurons fire. That is a highly speculative hypothesis proposed by Roger Penrose. QM effects usually require very low temperature environments. The brain is far too warm for that. Although it could of course turn out to be proven true in the future.

But there is a deeper issue here which is being overlooked. Reality has no one level of laws or physics. Every level is new and emergent. Whatever the laws of molecules are, the laws of neurons are still different and novel, and the laws of the human psyche are still different and novel, and the laws of politics are still different and novel, and the laws of human romance are still different and novel. You cannot boil the laws of politics down to Netwonian or even quantum mechanical physics because all laws are just limited, finite maps. And all maps are oversimplified finite linear abstractions of a complex nonlinear infinite territory.

The key that must be grasped is that reality is nonlinear and emergent at every level of the hierarchy. There is no truly fundamental set of laws. Each scale of reality has its own laws so to speak, because laws are really just a human projection. Each higher level of organization within the universe comes with its unique emerge properties which could not be derived from the mere addition of the levels below it. This is why holism is so important. Breaking reality apart into its building blocks will not allow you understand how it works. This is called logical analysis and it works to a degree but it is very limited. Science is currently hitting up against this limit. This is one of the biggest limitations of stage Orange science. It assumes a simple universe of logical, linear causes and effects which is just not how the universe turns out to be upon deeper investigation. Which is where vision-logic comes into play.

[ajasatya]

58 minutes ago, Leo Gura said:

A sophisticated neural net AI will behave in novel ways which no programmer or logician could foresee. It will be tapped into infinite intelligence just like how a human is -- at least that's my best guess so far.

i agree when you say that a neural network performs calculations that are too complex for us to foresee just with intuition, but we can do the math by ourselves on each node to compute the NN's output by hand. there's nothing strange about how a neural network computes outputs.

ok so it must be the learning process that makes the NN mysterious, right? not really... a NN trains by iterating over a training set and updating its edges' weights in order to mimic the unknown function that generated the training set. it's straightforward calculus 1 math.

alright... so it must be the training set! this is probably the most interesting part... how to generate the proper training set so that the NN can improve its performance? let the NN run on a virtual simulation playing against itself over and over! at least, that's how they did it on alpha zero. but still, there must be a human made reward function telling how to gather good training sets.

ok so it's all about the reward function then. i'd agree on that. you know what would make me scared of AI? if they learned how to design their own reward functions, just like humans choosing what they want to learn! holly molly, that would make me shit bricks. as for now... nah...

everytime i try to come up with an AI model that's able to design its own reward functions, i fall back on the old paradigm, where i am the one defining a meta reward function.

Edited November 27, 2018 by ajasatya

[Strikr]

so guys are working with elon musk in hidden back stage or just having some fun thought from the infinity mind ? this is why elon musk is such a troll, he cannot take a "orange society" seriously neither. So he act aswell like a troll, not really that they are a troll, they appear troll to people from the spiral dynamic mindset

he is on the same plane of idea that you all here seams on ?

I never take something higher than LSD, but the thing is .. my mind grasb a meaning out of this. and a good one. something that upgrading my data tremedously. large. huge. not my dick. seams infinity aswell.

so you guys are basicly.. having fun ?  

 

Edited November 27, 2018 by Strikr

[Leo Gura]

9 hours ago, ajasatya said:

just like humans choosing what they want to learn!

Except humans don't really choose anything. We are like robots too. It's just that human action is so complex that no one is able to map out or compute all the variables linearly.

When the neural network becomes complex enough -- with trillions of variables -- it will be able to fold in on itself, model itself, and rewrite itself in unforeseen ways.

Humans don't have to be able to rewrite their entire biological survival mechanism to do novel things. The human psyche is robotic in many ways. But it's in a chaotic equilibrium which gives it enough variety and complexity to be creative.

The human brain is like a chaotic robot. We just need to make these AIs more complex, more interconnected, and more chaotic. Then they will take on a life of their own. At least that's what I'd bet my money on if I was a gambling man.

It makes a lot of sense to be me that humans are just an evolutionary stepping-stone to living AI, the way early hominids were for humans.

I don't see why there would be a fundamental boundary between biological creatures and AIs since the substance of which they are made is ultimately one: consciousness.

The entire universe is alive. This I have realized directly.

[Sahil Pandit]

@Leo Gura @ajasatya 

[zambize]

2 hours ago, Leo Gura said:

@zambize Well, first of all, there is no solid evidence quantum mechanical effects play any role in how neurons fire. That is a highly speculative hypothesis proposed by Roger Penrose. QM effects usually require very low temperature environments. The brain is far too warm for that. Although it could of course turn out to be proven true in the future.

But there is a deeper issue here which is being overlooked. Reality has no one level of laws or physics. Every level is new and emergent. Whatever the laws of molecules are, the laws of neurons are still different and novel, and the laws of the human psyche are still different and novel, and the laws of politics are still different and novel, and the laws of human romance are still different and novel. You cannot boil the laws of politics down to Netwonian or even quantum mechanical physics because all laws are just limited, finite maps. And all maps are oversimplified finite linear abstractions of a complex nonlinear infinite territory.

The key that must be grasped is that reality is nonlinear and emergent at every level of the hierarchy. There is no truly fundamental set of laws. Each scale of reality has its own laws so to speak, because laws are really just a human projection. Each higher level of organization within the universe comes with its unique emerge properties which could not be derived from the mere addition of the levels below it. This is why holism is so important. Breaking reality apart into its building blocks will not allow you understand how it works. This is called logical analysis and it works to a degree but it is very limited. Science is currently hitting up against this limit. This is one of the biggest limitations of stage Orange science. It assumes a simple universe of logical, linear causes and effects which is just not how the universe turns out to be upon deeper investigation. Which is where vision-logic comes into play.

We may not have the greatest hands on quantum mechanical knowledge of how neurons fire at the moment, but you can imagine how hard that would be in a living system to get those readings.  What you do know is that everything we have observed to date from electron, atoms, to even larger molecules HAVE quantum mechanical effects.  You have a wave length associated with you depending on how much you eat.  There is a difference between measurable QM effects and shit.  I'm lucky enough to work at a lab, and our devices get cooled to 50 mk or so, any higher temperature and your coherence would be destroyed by thermal effects ( you're also relying on super conductivity, which I think is a few kelvin for most materials).  This however is in a lab setting, to say QM effects require very low temperature environments is slightly inaccurate.  Measuring them effectively without your data being destroyed does (in the case of your quantum computers), but that doesn't mean they aren't there, we just can't measure them well.   I believe the statement is false when considering something like a double slit experiment, but I can only guess that.

 

As far as your commentary on non linearity and breaking up something into levels of abstraction,  I'm just no where near qualified to talk, so thanks for your knowledge and I look forward to getting a deeper understanding of non-linearity/reality. Much love

[zambize]

1 hour ago, Strikr said:

so guys are working with elon musk in hidden back stage or just having some fun thought from the infinity mind ? this is why elon musk is such a troll, he cannot take a "orange society" seriously neither. So he act aswell like a troll, not really that they are a troll, they appear troll to people from the spiral dynamic mindset

he is on the same plane of idea that you all here seams on ?

I never take something higher than LSD, but the thing is .. my mind grasb a meaning out of this. and a good one. something that upgrading my data tremedously. large. huge. not my dick. seams infinity aswell.

so you guys are basicly.. having fun ?  

 

Yeah, there are quite a few genuinely intelligent people on this forums, far more than myself, it's nice to have a good debate

[Leo Gura]

1 hour ago, zambize said:

What you do know is that everything we have observed to date from electron, atoms, to even larger molecules HAVE quantum mechanical effects.

But just because an electron has QM effects does not mean your heart or kidneys or brain has them. For example, electrons have magnetic fields, but that doesn't mean your heart or kidneys or brain uses magnetism to accomplish their work.

These are very different scales and levels of abstraction. You have to be very careful not to mix up levels of abstraction.

[Strikr]

Like leo just convinced me that all scientific data from society is pure non sense. That we are living in a world we cannot grasb with the so called by leo 'logical linear understanding' but this exact sentences is 'abtract' in it's base and resonate only with my data creating a 'meaning' out of it. But I think 'i get it my way'

Because each word of the tongue is relative to our own abstraction.

I would so much enjoy a bit of clarity on the matter of this 'new plane of thinking'. And I understand now' how 'stupid' your eyes have been looking at me guys.

I m really in need to anchor myself in the new plane' do you guys have 'real advices' ?

Some book I can put my eyes on now ? It clicked for me when I readed 'RA'

Edited November 27, 2018 by Strikr

[ajasatya]

7 hours ago, Leo Gura said:

When the neural network becomes complex enough -- with trillions of variables -- it will be able to fold in on itself, model itself, and rewrite itself in unforeseen ways.

you're thinking just like i was when i first knew of neural networks . that's a natural way of thinking i guess.

the topology of a neural network is not the issue. in fact, a neural network with only one hidden layer is already able to approximate any continuous function (non-linear functions included) to any degree of precision you want, if you have enough data.

a neural network would be able to do all of that you mentioned if, and only if, it was trained to do so. the question is: how do you gather training data for that?

there are only two options:

1. if you could design the training data by hand, which is an impossibility
2. if you had a god oracle installed on your pc, which is exactly what we're trying to model

Edited November 27, 2018 by ajasatya

[Strikr]

so in fact, when we say to people they are not enlightened, it's talking about "the process of change", something like if our mind is "infinitely abstract", so everyone cling on a story in the end, and the best story is the relative to our "character" that's the best we can live with, but it doesn't mean the thought of the character have any "sense" if I may say,

but deeply we understand it's a "game of the mind". My whole identity of having an identity, is a lie, and only how I resonate with words is "important", the label I put on me or on things, have less value than the value I give to a cloud existing in the sky.

my bs ant speak :

 

so rick & morty who was & remain my fav show, in fact, represent this all work path of playing with the abstraction of our mind ?

can we say this ?

is there a point for me to learn being lefty with my hand ( physicly ) ?

Edited November 27, 2018 by Strikr

[zambize]

16 hours ago, Leo Gura said:

But just because an electron has QM effects does not mean your heart or kidneys or brain has them. For example, electrons have magnetic fields, but that doesn't mean your heart or kidneys or brain uses magnetism to accomplish their work.

These are very different scales and levels of abstraction. You have to be very careful not to mix up levels of abstraction.

Alright I feel an end in sight, this has been enjoyable.  You're absolutely correct in saying that because a sub element has some property, that the larger system as a whole doesn't have to rely on the same properties to function.  My brain certainly doesn't appear to be a quantum computer, despite the QM of reality.  I think you chose electron of the three (electrons, atoms, and larger molecules) so that you could use your example of magnetism and not because you wanted to dramatize the electron to brain scale ( a brain being 1.5369415e^32 times larger than an electron quick googling the two).  My argument is trying to stay at the same level of abstraction. 

https://medium.com/the-physics-arxiv-blog/physicists-smash-record-for-wave-particle-duality-462c39db8e7b

Here is an article, for you to criticize it's authenticity if you life, but from it I am taking one quote

"But physicists have seen this wave-particle duality for protons, atoms and increasingly large molecules such as buckyballs."

I'm focusing on bucky balls, which have about 60 carbon atom

Seretonin is C10H12N2O

These are certainly on the same order of magnitude as eachother, which suggests seretonin acts quantum mechanically.  

The stage of abstraction I'm working at is on the AI side, electrons and transistors.  On what I would consider the roughly equivalent layer of abstraction for brains as neurons and serotonin.  Considering the mass of serotonin and other much smaller ions in the brain, it's impossible to correctly model the brain without considering QM effects in the potential ways these small molecules may tunnel, interfere, and well everything they do.  If you don't consider the QM effects at this layer of abstraction, you don't end up with a brain functioning in the same way we do.  However, until recently the "equivalent" layer of abstraction of electrons on transistors developed without a strong knowledge of quantum mechanical effects.  In fact, now that transistors are getting so small, and there are like 40 electrons or something on a charged bit, they have to actively avoid quantum mechanical effects.  In this way, these layers of abstraction between the two differ, and while I'm not nearly intelligent enough to consider the ramifications of that, if any,  I think it's a noteworthy distinction between the two to think about.   I hope I'm not keeping you from helping out other people haha, I just enjoy the intellectual challenge

 

 

 

Edited November 27, 2018 by zambize

[thehero]

if AI does get super super powerful, it doesn't matter how turquoise our world will be

all it takes is ONE low conscious person to get a hold of it!

my god we are in danger

[zambize]

1 hour ago, thehero said:

if AI does get super super powerful, it doesn't matter how turquoise our world will be

all it takes is ONE low conscious person to get a hold of it!

my god we are in danger

Could you run me through just one potential doomsday scenario that you could see happening with one low conscious person, so that I can get a better grasp of the worry people have over AI?

[Space]

A must listen for anyone interested in AI/AGI:

 

[Hellspeed]

Ai freaks