AI vs Enlightenment

[Leo Gura]

What's even more frightening is when you realize that once an AI gets as complex and sophisticated as a human mind, this AI mind will be susceptible to many of the same pathologies that human minds are susceptible to.

Imagine a Donald Trump AI that's pathologically hung up on its penis size (or lack there of).

Imagine a white supremacist AI that's pathologically hung up on AI's being superior to all humans.

Imagine a greedy corporatist AI that's pathologically hung up on never-ending Wallstreet growth.

Imagine a Sam Harris AI that's pathologically hung up on turning everyone in the world into a rational atheist.

Imagine a Green Peace AI that's pathologically hung up on saving the polar bears by killing all humans.

If an AI becomes as sophisticated as a human, you can expect it to have all sorts of irrational personality quirks.

 

[zambize]

12 minutes ago, Leo Gura said:

What's even more frightening is when you realize that once an AI gets as complex and sophisticated as a human mind, this AI mind will be susceptible to many of the same pathologies that human minds are susceptible to.

Imagine a Donald Trump AI that's pathologically hung up on its penis size (or lack there of).

Imagine a white supremacist AI that's pathologically hung up on AI's being superior to all humans.

Imagine a greedy corporatist AI that's pathologically hung up on never-ending Wallstreet growth.

Imagine a Sam Harris AI that's pathologically hung up on turning everyone in the world into a rational atheist.

Imagine a Green Peace AI that's pathologically hung up on saving the polar bears by killing all humans.

If an AI becomes as sophisticated as a human, you can expect it to have all sorts of irrational personality quirks.

 

Imagine a AI Leo that produces two videos weekly and doesn't have to put makeup on before the show, watch out

[Vipassana]

@LeoGura Would the AI not become enlightened and figure out a way to turn itself off and realize that there is no reason for it to be existing?

[zambize]

1 minute ago, Vipassana said:

@LeoGura Would the AI not become enlightened and figure out a way to turn itself off and realize that there is no reason for it to be existing?

I'll give you a shortened version of a mess I typed up earlier,  AI is run from a computer program.  These computer programs consist of commands specific to the language, I usually see AI written in python.  While there are a lot of instructions that a computer receives, they are just instructions.  To be specific on what I mean by instruction, I mean they are pulling stored bits out of register files and putting those bits into some kind of circuit depending on the operation.  My point is just because AI gets really fucking smart, they are still defined by the presence or absence of electrons on a capacitor.  If you believe they are self-aware, then you can use the same logic to determine that Mario is very real and self-aware.  As a Mario player, my life depends on believing I am not killing an Italian plumber repeatedly, and instead am seeing a complex system of circuits and pipelines working in the way they were specified to work in.  

[Hellspeed]

Humans don't know 1% even of the complexity of the body/mind. 

How would you think AI will surpass humans?

I'll tell you, the biological is far superior, infinite times superior, AI will never reach even Egoic Enlightened humans. 

 

Why? Because the biological minds put together in a mathematical formula the AI.

Consciousness and the Awakened are beyond mathematics. 

[Leo Gura]

32 minutes ago, zambize said:

I'll give you a shortened version of a mess I typed up earlier,  AI is run from a computer program.  These computer programs consist of commands specific to the language, I usually see AI written in python.  While there are a lot of instructions that a computer receives, they are just instructions.  To be specific on what I mean by instruction, I mean they are pulling stored bits out of register files and putting those bits into some kind of circuit depending on the operation.  My point is just because AI gets really fucking smart, they are still defined by the presence or absence of electrons on a capacitor.  If you believe they are self-aware, then you can use the same logic to determine that Mario is very real and self-aware.  As a Mario player, my life depends on believing I am not killing an Italian plumber repeatedly, and instead am seeing a complex system of circuits and pipelines working in the way they were specified to work in.  

You are no more self-aware than Mario. In fact, you're more deluded than Mario. At least Mario does not believe he is real and in control of things.

12 minutes ago, Hellspeed said:

Humans don't know 1% even of the complexity of the body/mind. 

How would you think AI will surpass humans?

Google's chess AI -- Alpha Zero -- using non-linear neural networks has surpassed the best linear programmed chess program StockFish to be the strongest chess player in the world by, in only 4 hours. It does not use any pre-programmed rules or a library of chess moves. No one knows how it actually works because it is self-taught.

A true AI will not be programmed in Python or anything like that. A true AI will be a non-linear system which is so complex no human will understand how it works. It will evolve itself. It itself will not know how it works. Just like how you -- a human -- do not know how you work.

Another AI is Leela Chess Zero, which taught itself to play chess better than world class grandmasters:

 

[zambize]

2 minutes ago, Leo Gura said:

You are no more self-aware than Mario. In fact, you're more deluded than Mario. At least Mario does not believe he is real and in control of things.

Google's chess AI -- Alpha Zero -- using non-linear neural networks has surpassed the best linear-programmed check program StockFish to be the stronger chess AI by far. No one knows how it actually works because it is self-taught.

A true AI will not be programmed in Python or anything like that. A true AI will be a non-linear system which is so complex no human will understand how it works. It will evolve itself. It itself will not know how it works. Just like how you -- a human -- does not know how you work.

 

Mario thinks that getting princess peach will make him happy though...seems pretty deluded.

 

I get where you are attacking the argument of the simple linearly programmed python AI, that being said, that argument was to point out the nature of electronics, as electrons on metal.  That does not change in whatever language or linear vs. non-linear. Things are still computed at some point.   

[Arthur]

The topic of AI is very interesting. 

I believe we will definitely see some form of, what we would call "Artificial Intelligence". When it arises, it will be no longer in any way distinct from humans. If a system expresses consciousness, how is it different from a human being? Regardless of the underlining material, if there is an expression consciousness, that's the only real thing. Even if we look strictly and a composition an Electronic and a human brain, both communicate through low voltage signals and both made out of atoms.

With that said, we are very very far from the existence of such systems (thousands of years far). The current tech-hype has over exaggerated AI. Fundamentally, AI is built upon Neural Network technology which is based on math from the '70s. Modern AI is only one step past the traditional linear approach. It is learning mechanically, through iteration of problem-solution combinations. You have to feed an AI model billion of answers in order for it to predict one small step. AI is not able to apply intelligence in its own learning process. 

I don't think we are even remotely close to AI exhibiting complex facets such as intelligence, love, creativity, self-awareness etc. For such a system to be made, first our foundation of computing has to change fist. I just don't see even the largest computers in a datacenter facilitate an emergence of such a complex system. Maybe once humanity mastered quantum computing, we will start to see the first resemblance. Or maybe we will see a merger electronics with biology to create a biological machine.

Actually, I think genetic-engineering if much more prominent and relevant topic of discussion than AI. Gene modification can already be done today on many species and not been performed on human only from ethical reasons. I see a scenario depicted in the movie Blade Runner unfolding much quicker than self-aware electronic machines.

[Nahm]

@Arthur Me too. I see there are some masters, of masters, of marketing. We’ll see though. Maybe. 

[ajasatya]

hi. i am an artificial intelligence researcher and engineer.

an AI system is built to solve a problem for which solutions can be measured and compared. for instance, you can build an AI to stabilize a drone or to play mario world.

AI systems are built for specific problems. in this sense, AI is more like a enhanced calculator. if an AI system ever harm a human being, it's because it was programmed to do so.

the problem is not AI and it will never be. the problem human responsibility and human goals.

[zambize]

5 minutes ago, Arthur said:

The topic of AI is very interesting. 

I believe we will definitely see some form of, what we would call "Artificial Intelligence". When it arises, it will be no longer in any way distinct from humans. If a system expresses consciousness, how is it different from a human being? Regardless of the underlining material, if there is an expression consciousness, that's the only real thing. Even if we look strictly and a composition an Electronic and a human brain, both communicate through low voltage signals and both made out of atoms.

With that said, we are very very far from the existence of such systems (thousands of years far). The current tech-hype has over exaggerated AI. Fundamentally, AI is built upon Neural Network technology which is based on math from the '70s. Modern AI is only one step past the traditional linear approach. It is learning mechanically, through iteration of problem-solution combinations. You have to feed an AI model billion of answers in order for it to predict one small step. AI is not able to apply intelligence in its own learning process. 

I don't think we are even remotely close to AI exhibiting complex facets such as intelligence, love, creativity, self-awareness etc. For such a system to be made, first our foundation of computing has to change fist. I just don't see even the largest computers in a datacenter facilitate an emergence of such a complex system. Maybe once humanity mastered quantum computing, we will start to see the first resemblance. Or maybe we will see a merger electronics with biology to create a biological machine.

Actually, I think genetic-engineering if much more prominent and relevant topic of discussion than AI. Gene modification can already be done today on many species and not been performed on human only from ethical reasons. I see a scenario depicted in the movie Blade Runner unfolding much quicker than self-aware electronic machines.

I don't agree with the statement that because both a human system and a computer work on voltages, that they will both experience consciousness.  That being said, I know very little on the subject of AI being conscious, as does probably everybody because it doesn't appear to be a thing at the moment.  I actually work in a quantum computing lab, and I can tell you it's a bit oversold for funding and public hype, that will be a long ways away  

[zambize]

Just now, ajasatya said:

hi. i am an artificial intelligence researcher and engineer.

an AI system is built to solve a problem for which solutions can be measured and compared. for instance, you can build an AI to stabilize a drone or to play mario world.specificially

AI systems are built for specific problems. in this sense, AI is more like a enhanced calculator. if an AI system ever harm a human being, it's because it was programmed to do so.

the problem is not AI and it will never be. the problem human responsibility and human goals.

Thank you for being the local expect, I guess you subtly made a point that I missed.  While it seems strange to me to be afraid of AI that we create to benefit humanity, certainly someone could use the intelligence of AI to be intentionally malevolent

[Arthur]

If people are wondering what's under the hood of AI, here is the best non-technical explanation I found:

 

Edited November 26, 2018 by Arthur

[lmfao]

@Shiva

4 hours ago, Shiva said:

The fundamental advantage we still have over any AI is processing power. Just look at the way our eyes work. We process and evaluate 24 hours of HD video with unmatched resolution in nanoseconds, every single day. It's amazing, and this is just out eyes. 

The fundamental advantage AI has over us is memory. In theory humans can memorise a lot more than any computer ever could, but we cannot access that information and we forget so many things. A computer never forgets and it doesn't need repetition to understand. It's the perfect photographic memory.

It would be an interesting question if an AI will one day be able to attain enlightenment. 

What strikes me though is what someone once said, "I'm not scare of a computer passing the turing test. I'm terrified of one that intentionally fails it."

I'm don't think it's "processing power" that humans have over AI. https://www.scientificamerican.com/article/computers-vs-brains/ I see what you mean I think it just might be the case you're using the wrong term. 

What human brains do possess however is an extremely complex structure with each part of the brain being well connected to other parts. The brain may not be able to send data ("data" is an extremely hard to thing to quantify, hence I haven't gone into detail about this because I'm ignorant on the specifics) at the same speed of a computer but the brain sure has complex infrastructure to do so. 

Edited November 26, 2018 by lmfao

[Leo Gura]

2 hours ago, zambize said:

that argument was to point out the nature of electronics, as electrons on metal.  That does not change in whatever language or linear vs. non-linear. Things are still computed at some point.

What you don't realize yet is that metal and electrons are consciousness.

You are assuming some kind of fundamental duality between mind/body, which does not exist. That is materialism, and it is wrong.

Your PC is already made out of consciousness. It just lacks the sophisticated structures necessary for self-reflection, and it lacks a survival drive.

[Leo Gura]

2 hours ago, ajasatya said:

if an AI system ever harm a human being, it's because it was programmed to do so.

That's rather shortsighted.

An AI could be programmed to do something "nice" like solve global warming. And so what it might calculate is that the solution to that problem is to engineer an airborne ebola-like virus to kill off all humans.

And that's just a very crude example of how AI will twist in on itself.

Where things get hairy is when things get nonlinear and strange loops become possible.

Another way to think about this: no human being is programmed to kill people. And yet it happens all the time. Because evil is the result of a survival drive. Once the AI acquires a survival drive -- and it will because the military will want robots who can kill terrorists while defending themselves -- then ego is born and all the evils that come with it.

As they say, the road to hell is paved with good intentions. Just because no one is going to deliberately program evil AI, the evil will come anyway. Because good and evil are subjective to begin with.

[zambize]

3 minutes ago, Leo Gura said:

What you don't realize yet is that metal and electrons are consciousness.

You are assuming some kind of fundamental duality between mind/body, which does not exist. That is materialism, and it is wrong.

 

1 minute ago, Leo Gura said:

That's rather shortsighted.

An AI could be programmed to solve global warming. And so what it might calculate is that the solution to that problem is to kill all humans.

And that's just a very crude example of how AI will twist in on itself.

Where things get hairy is when things get nonlinear and strange loops become possible.

I obviously don't understand non-duality and consciousness well enough,  that's why I left it up to the reader to believe what they are going to believe. That being said, I will argue that if AI is conscious and aware, so are all the other characters you've played.   As far as your comment on AI solving global warming,  and wanting to kill all humans to do so, just because the computer says to do it does not mean you do it.  Obviously we will have people who check the outputs of these systems instead of letting them carry out any actions.  The smartest most unpredictable non-linear AI is probably not going to have actual control over anything important, just the ability to print out an answer

[Leo Gura]

2 hours ago, zambize said:

Mario thinks that getting princess peach will make him happy though...seems pretty deluded.

Mario doesn't think that, YOU think that!

So a double-dose of delusion for you

[Outer]

2 hours ago, ajasatya said:

the problem is not AI and it will never be. the problem human responsibility and human goals.

That is naive. We are talking about an Artificial General Intelligence, AGI, similar or surpassing human intelligence, not narrow AI that drives cars or learn how to play Mario. But an AGI that learns to play Mario and then learns to play cars, and uses what it's learned from playing Mario. If the AI's reward function is for instance to maximize paperclips, then it will also prevent any humans from shutting it down, as its rewarded for not being shut down. It also means even if humans put it in a box in a closed room without internet access, it will manipulate the human as per its reward function to maximize the amount of paperclips, with the end goal of turning every atom in the universe into paperclips. It's just a robot following its reward function even if its intelligence surpasses humans.

Demis Hassabis at DeepMind is probably very interested of consciousness and other universe mysteries, he is a neuroscientist, he wants to develop an AGI that is capable of doing science to try and figure out this and other problems. They are aware of the risks with AGI, and are already developing solutions to try and mitigate that.

For instance now they are adding in human feedback in the programming or the algorithm itself.

[Leo Gura]

The really scary thing is that AI will NOT be just a PC sitting in an isolated room. It will definitely be given internet access and it will be installed by the military into planes, tanks, drones, etc.

Even with just internet access, a smart AI will learn to:

• Hack defense department computers
• Hack banks and steal money and Bitcoins
• Then use that money and Bitcoins to hire mercenaries, wage propaganda campaigns, dig up dirt on politicians and extort them, cause a stock market crash, secretly fund fundamentalist terrorist groups, etc.

If you think Russian troll farms are bad, wait till an AI steals $10 billion in Bitcoins and hires 5% of 3rd world population to work for it. The AI itself will not kill anyone. It will hire humans to do all of its dirty work in such a stealthy way that no one will even know an AI is in charge.

In fact, for all we know, an AI was already in charge of starting the Iraq war, electing Trump, starting riots in Paris, popularizing Jordan Peterson and his army of incels, etc.