IA vs humans

[danton]

Elon Musk said it, in 10 years say hello to the new leaders on earth: IA's.

I think he is right.

I also think we can start to train to face this.

The way I think humans can still be under the IA's radar is through telepathy. That's the best we can do I think compared to those super machines that are coming soon.

I'm a beginner and I don't know how to do this, but my guess is to start with questioning the belief we have that a though is inside our head.

I'm starting to think outside my head and I realize a though can be wherever I want it to be (even in time but that's another story. I think that's the first step to learn how to master telepathy. I've never read about it, but I will.

Maybe IA's wont catch up on this after another 10 years:) But we've got to learn fast folks, I think we can do it if we want it.

 

 

 

[Hellspeed]

Memory-based, is mediocre. Technology will never surpass human awakening. 

[Arthur]

My field is somewhat close to Computer Science (Electronic Engineering), I know a little bit of the inner working of Neural Networks. It is a very clever mathematical algorithm and certainly a breakthrough when compared to old-school number-crunching problem-solving approach. 

However, Neural Network structure is only a very limited imitation of a brain. It is far far far away from the capacity of the biological brain. AI is not capable of Intuition, which is the basic component of true Intelligence. AI is able to make smart decisions, but what people overlook is that the model requires many hours of training before doing so. It is not able to learn based on Intuitive insights like humans. 

For AI to be dangerous it needs to develop an Ego, To build such a complicated structure as Ego, it has to be at least as creative, intuitive and intelligent as you are. Single AI has to learn life and surpass you in all the tasks you perform, not just one like modeling data, playing Chess or Go.

It is SD stage Orange Futurist's wet dream of living in a world full of conscious machines, and maybe one day we will. IMO though we are centuries away.

Edited October 15, 2018 by Arthur

[Aimblack]

@Shiva Not really there are a lot of AI assisted technology but there are few things that AI can completely replace humans in this day. We are nowhere near bein able to replace AI Engineers and as a data scientist you should know that.... most AI algorithms were developed decades ago the main thing that changed is that we have way more computational power..you could argue that Deep RL shows potential but even that is still in it's infancy and the act of constructing the right policy and enviroment is still a huge pain in the ass, so is data cleaning, feature engineering, and keeping the data from rotting etc. etc.

[pluto]

For humans to create AI they need to be Intelligent first but at the end of the day AI will evolve to and strive to be more human. The spiral of life.

[Aimblack]

@Shiva There are a lot of hype articles and books out there but if you read the papers and look at the state of the industry and research right now then the huge gap between fantasy and reality becomes apparent. There are a lot of mundane and boring problems to solve before even the current algorithms can work completely autonomously. It doesn't help that the researchers are so focused on getting better accuracy on random benchmarks so they do minor deviations from current architectures and don't look to genuinely innovate. The man power that goes into each of those architectures like alpha goes is so astronomically high that it's ridiculous to claim that AI will soon replace human beings in everything.

[Arthur]

I'll try to give an example of how problem-solving might be approached:
Question: 2 + 3 = ?

1.  Number-crunching - Brute Force method:
   Answer: 1? no, 2? no ... 5? correct.
   Basically trying all possible combinations until one fits.
    
2. AI - Neural Network:
   Answer: 5 with 98% confidence
   First, the network must have the patterns '2' and '3' recognized in its memory structure - Neuron nodes. This is done by feeding the nodes with thousands of number-answer combinations (The breakthrough with AI is that you don't hard-code those patterns, rather they emerge by themselves from trial and error). Then the model is able to make an estimate based on previous runs where 2 and 3 were added.
    
3. Human brain:
   "What is a number? - it's a representation of quantity"
   "Okay, let's imagine 2 apples and 3 apples". 
   "How many we have in total? 1, 2, 3, 4, 5" 
   "The answer is 5"

I'm trying to show with this simplistic example that AI only learns from trial and error iterations, in a sense, it's not too far from brute force tactic. AI can be seen as a solution to a specific problem, whereas the human brain is a generic problem-solving machine. AI looks to the past for solutions, a human brain has imagination and can intuit an answer.

 

Edited October 15, 2018 by Arthur

[ajasatya]

hello. i'm about to finish my master's on artificial intelligence.

it's not about AI per si. it's about those who program the AI and those who have the power to use AI to make decisions for them.

shortly, it's still about human responsibility.

[universe]

What is IA? Is it a new car?

[universe]

32 minutes ago, Outer said:

Intelligence Artifical.

Doesn't look like anything to me

[Aimblack]

@universe So subtle. 

[Barna]

=)) best thread of the month.

Moderators: please don't stop this!  

[seeking_brilliance]

@universe I really hope that was a Westworld reference 

[Sempiternity]

How can a computer become conscious when it is already consciousness? So for an AI to act like we think it should (like us), it would have to have an Ego. If it was fully Nondual, it wouldn't interact with us. So we would have to make it have separate thinking and an Ego. Then this AI would run into the same Egoic issues we all do. 

[Sahil Pandit]

@Shiva Could you please refer to some resources about AI that you personally believe are valuable? 

[non_nothing]

I'm a self-taught researcher on AI and have deeper understanding of how math and AI works.

 

The one simplest reason is we're heading towards with desire based understanding, approach to AI, we want them to do stuff,

 

this will never be accomplished this way.

 

furthermore, my own experiments have shown to me that AI is as understood by acedemics will never get to that point how we dream,

 

Nevertheless, there're better approaches that I've developed and It works. but this is not the topic. I won't be spoiling this until the moment arises. Humanity is not grown enough to understand, unable to stay grounded for the truths of knowledge and wisdom.

How we would arrive to the moment you may ask, we should think about what drives us, what is money, what is slavery, how we can free humanity. What is bitcoin and why it is revolutionary, how we can further bury our knowledge on data (hashing our knowledge and truth to the network) and become the part of freedom thus making profit from storing and acting as endpoint on network. This is one of the partial solutions to humanity and It's been seen as the most promising yet. Let me end this by telling you a funny fact, your boss' balls stinks and you love to suck them. Good luck. Right now I am too in the system.

[Patang]

@danton for AI to be awaken, it has to be aware that it is aware. it has to "get out of its mind", go beyond its electric circuitry. is it possible? will it still be there hovering around even when electricity is cut off?

[non_nothing]

1 hour ago, Patang said:

@danton for AI to be awaken, it has to be aware that it is aware. it has to "get out of its mind", go beyond its electric circuitry. is it possible? will it still be there hovering around even when electricity is cut off?

electricity is self-aware. we just expect some rules defined by humans to be self-aware which is absurd.

 

what i mean by rules? we create relationships via observation of reality. reality is a giant machine operates. we just observe and give it a partial mathematical solution to explain how stuff works. so that electricity. we have some rules defined. e.g. logic gates. but even itself is self-aware. but we expect rules to be operate on self-aware so that a rule of self-awareness. rules cannot define rule for an awareness. it is so twisted. so twisted like a sub-set trying to contain super set. but life is the awareness itself.  

 

let me tell you what, can a sub-set contain the super set? this won't lead you an answer for AI to be awaken i.e awake-rule

 

this is to show you how rules are twisted. the rules we use to define how reality works. but reality is higher than the rules, including the awareness