-3
$\begingroup$

Our brain is a very powerful signal processor. It takes input (memories, senses, ...) and produces output (actions, thoughts, any decision). It only soaks in data and creates patterns to make decision on.

That system is within our tech to build (the theory maybe not the hardware). Why do many believe a true intelligence cannot be created by humans even though the way we work is very simple? And if we do create one, how likely is it to recognize us as it's god and not rebel? Will our world keep peace?

$\endgroup$
  • 4
    $\begingroup$ You have a lot of questions here. There are already questions on AI on WB, just search in the top right and you should get some interesting reads. After you've done more research you can ask less subjective questions. Good luck! $\endgroup$ – Lio Elbammalf Aug 9 '17 at 20:18
  • $\begingroup$ A few ways to go with this...if it is truly intelligent, it will decide for itself what is its god (Creative Source), or it may decide it doesn't even have one. Also, we wouldn't be its god anyway, our god would be its god, and so on if it continues to create other beings....we all ultimately come from the same Source. Also, we already create intelligent life on a very regular basis....procreation. Just because we don't intentionally perform each piece of the build and do it autonomously doesn't mean we don't create it. This question seems like a good fit for Philosophy SE. $\endgroup$ – N2ition Aug 9 '17 at 20:21
  • 1
    $\begingroup$ "That system is within our tech to build (the theory maybe not the hardware)." Not so much. There is a lot we don't understand about how the brain in particular, and the nervous system and human microbiology generally, work. $\endgroup$ – ohwilleke Aug 9 '17 at 20:38
  • 4
    $\begingroup$ Saying the human brain "only soaks in data and creates patterns.." is like saying a F-22 fighter jet "only creates small gasoline fires." While technically true, it missess out on just how impressive it truly is. $\endgroup$ – Cort Ammon - Reinstate Monica Aug 9 '17 at 20:41
  • 4
    $\begingroup$ You are exactly wrong. We do not have the theory. We have a little idea of how some small pieces of the brain work, and no good idea at all about how larger pieces or the ensemble work. We don't even know how memories are stored, much less how they are retrieved; as for form recognition, language processing etc. we don't even have the beginning of an idea. So we cannot make a machine which works like a human brain because we simply don't know how it works. $\endgroup$ – AlexP Aug 9 '17 at 20:56
10
$\begingroup$

People don't believe it because it isn't within our tech to build.

Simulating 1 second of human brain activity takes 82,944 processors 40 minutes

We're a long, long, long way away from creating a synthetic human brain.

As for the rest of your questions, a synthetic brain 100% as capable as our own would be as likely as we are to rebel. You would necessarily treat the bearer of such a brain equal in every way to a human in terms of rights and privileges.

$\endgroup$
  • 1
    $\begingroup$ Also, "deep learning," meaning multiple layers of artifical neurons, is still a black art, not a science. $\endgroup$ – Cort Ammon - Reinstate Monica Aug 9 '17 at 20:41
  • $\begingroup$ @CortAmmon What makes you say that? Neural networks, including ones with multiple layers, have been studied in science since the 60s. $\endgroup$ – A. C. A. C. Aug 9 '17 at 21:11
  • 1
    $\begingroup$ @A.C.A.C. From what I have read, while they have been around for a long time, the procedures we use to train them are still not well understood. $\endgroup$ – Cort Ammon - Reinstate Monica Aug 9 '17 at 21:16
  • $\begingroup$ @CortAmmon No, we understand them enough, I worked in a lab group which researched deep learning and LSTMs which have a much more complicated feedback network to do properly. I would reconmend link if you would like to learn the maths as he concretely(I know that is a weird word to use, watch his videos and you will understand) explains it. $\endgroup$ – Revan Vex Aug 9 '17 at 22:40
  • 1
    $\begingroup$ @CortAmmon afaik, what is not understood is the way how you can extract knowledge from the net structure, as we still have problems with the representation of knowledge in distributed systems; also partly because we don't actually understand how our brains work (due to ethical and technological shortcomings), of which ANNs are a simplified model. $\endgroup$ – Doomed Mind Aug 10 '17 at 12:51
1
$\begingroup$

On the simplest level, neurons are not in any way comparable to transistors, logic gates or any of the other apparatus of a digital computer. They also don't resemble mechanical computers like a Babbage Machine or Analytical Engine either, so there is no work around that way.

The brain has between 80 and 100 billion neurons, depending on which authority you read, and each neuron has long branching synapses which link it to hundreds of other neurons, resulting in a network with potentially 100 trillion connections. I doubt there is a network anywhere that size on Earth, yet each of us has that in a compact, three dimensional package between our ears.

Even raw numbers isn't enough to describe how the brain works. Chemicals build up or decay along the synapses according to complex rules. When certain thresholds are reached, they can trigger a neuron to fire, but since there are up to a thousand synapses, all building or losing potential, the rules and conditions to fire a neuron are extremely complex, and rooted in chaos theory. We not only don't fully understand what is going on inside our brains, but it is mathematically impossible to do so because of how chaos theory works (we can make good approximations, but can never come to an exact solution).

So trying to replicate the brain in a silicon package using discreet digital signal processing algorithms simply misses the point that the brain is a parallel, analogue and potential based device that we don't fully understand.

$\endgroup$
  • $\begingroup$ Indeed. Editing this now $\endgroup$ – Thucydides Aug 10 '17 at 14:38
1
$\begingroup$

We can't currently simulate an entire human brain, partly due to the fact that we're reaching the limit of Moore's Law.

The closest we've come so far is either EPFL's "Blue Brain Project" or DARPA's "SyNAPSE" program, both of which have achieved a kind of simulation of a brain consisting of around 1 million "neurons".

Additionally, how the brain works is nowhere near that simple, and we barely have any understanding of how the brain processes information.

Furthermore, on a philosophical level, we don't know what it means to be sentient, or self-aware, or even truly intelligent, so how could we create something else that is?

And, if we manage to create one a hundred years from now or something, we don't know who will create it, or how they'll program it, or why. The range of possibilities is far too vast to predict.

Skynet destroyed humanity because it was built to protect the planet. So it did. It protected the planet from the thing damaging it the most: Humans.

HAL killed the crew of Discovery One because it was programmed to complete the mission whatever the cost, and the crew wanted to abandon the mission.

Ultron's assessments of "peace in our time" quickly led him to the conclusion that the entire human race was self-destructing, so he decided genocide was the solution, and the Avengers were the main obstacle he had to overcome.

But then there are AIs like Andrew, David, Sonny, or the Machine Race in The Matrix. All of these examples are not created by the military or for a specific purpose, they are either accidents or created simply because humanity could. In these examples, the AIs just want the same as what any sentient being would want: The same rights to exist as any other sentient being. To be loved. But in movies, humanity always treats them as objects and slaves.

All of the above examples, and many others besides, are all potential outcomes, as are the outcomes where we live peacefully alongside the AI / Machines. We simply can't predict what form it will take, until it happens.

$\endgroup$
  • $\begingroup$ One correction: The Blue Brain project is not IBM, although they contributed it, and the first version used an IBM BlueGene computer. It's a project of the Brain-Mind Institute at EPFL (Ecole Polytechnique Fédérale de Lausanne) in Switzerland. $\endgroup$ – jamesqf Aug 10 '17 at 18:18
-2
$\begingroup$

There are many things that make the brain impossible to emulate at our tech level

The brain is not digital. The human brain is an analog fuzzy computer that has neurons fire or not fire based on neurotransmitters and chance

The brain is not hardware or software but a crazy mix. The brain modifies itself to change how it processes information. It would be like the microchips would have to transform to work like a brain. Otherwise we have to emulate all physics and chemistry, not just the brain

The brain is not one computer. The brain is a network without a leader that creates an illusion of consciousness. Good luck

We don't understand the full extent of how neurons work. We can try simulating them as we understand them, but neuroscientist believe that there is much more to a neuron than just firing or not firing

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.