"In order to fulfill the promise of the Internet of Things, it is necessary to process an amount of data that the world has not seen until now, and derive from it a logic that anyone can understand and act upon," said Harriet Green, IBM's vice president and director of the company's Internet of Things division
"The promise of the Internet of Things is to connect us better - from work to the living room at home. In order to do this, it is necessary to process a huge amount of data, such that the world has not seen so far, to connect the structured and unstructured information and to derive from it recommendations for decision-making. You have to make sense out of the information that everyone can understand and act on. Therefore, machine learning and natural language processing are essential to the development of the Internet of Things," said Harriet Green, IBM's vice president and director of the company's Internet of Things division.
Green said the words at the Wow (RT World Of Watson) conference in Las Vegas. The unit she manages relies on Watson's capabilities in cognitive computing, Big Data and analytical analysis.
According to her, "In the last 20 years, human society has managed to make history digital, to 'Wikiize' it (from Wikipedia - AB). The next step is to find order in the sea of data to derive insights. The amount of data is increasing rapidly: in 2025 there will be 29 billion things connected to the network and the activity around them will generate a market of 11 trillion dollars.
"Machine learning is the basis of all artificial intelligence, because it allows computers to learn without people programming them," Green noted. "In the context of Watson, this means that it will be able to manage a huge amount of different processes. He will understand when a product or device may break down and will be able to perform constant maintenance."
XNUMXD printed car
"Natural language processing is more than a smart way to communicate with computers. This is a significant change in the communication capabilities with computers, which will change the way we communicate with the physical world that surrounds us," she said.
As an example, Green gave the 12D printed autonomous vehicle Olli. In the demo, Oli's voice is heard saying: "I see XNUMX passengers at the station."
"Perhaps can answer every passenger's question like a real-life driver, and even more - from questions about getting to the destination to recommending shows, an offer to drive to them, and even a warning to wear something warm because it will be cool in the evening," she added.
Green also introduced Rita - a cognitive meeting board developed by IBM and Ricoh. The board picks up what is said in the meeting and can convey its content in the language of the participants. She stated that today 84 such boards are in use - all in Munich, and in the future there will be millions of boards that can connect.
"The Internet of Things is a growing market," she added. "For example, IBM has invested $23 billion in the Watson headquarters in Munich and it houses more than 1,000 of our best minds. We are already seeing an impact in three areas: in the way we design new products, produce them and consume them."
"It is important that we think of the Internet of Things as we think of the Internet itself - as a platform that sends and receives a huge amount of information from people to devices, from devices to people and between devices," said Green.
Finally, she referred to the security of the Internet of Things. "Each device connected to the network represents an entry point that has the potential for vulnerability and hackers exploit vulnerabilities in the end devices. "Cognitiveness will be able to protect Internet of Things systems and the data found in them," she noted.
109 תגובות
Miracles,
1. As I said before in such a small neural network as exists in this robot I have no expectation that consciousness will develop at a human level, maybe a low consciousness at the level of a mouse. This was just an example that shows you how relatively easily a neural network knows how to adapt itself to input/output means you connect to it even if they are not biological.
2. "K. You put it nicely - it is true that complex systems emerge with new properties, but there is no reason to think that what will emerge is something that will even slightly resemble human consciousness."
I actually think that it can be assumed, especially in light of the fact that we have already seen other features emerge from there that are unique to the cerebral cortex of a mammal, and therefore the feature known as consciousness, in my opinion, will also emerge from there as soon as the network has enough critical mass.
3. "The basic consciousness has 4 functions: eat, run away, fight and reproduce. Without these 4 no system would have properties similar to human consciousness.
I don't agree with you, these are traits that have evolved in animals in nature because a biological being must have them as part of survival and evolution, but I think that consciousness can exist even without these traits.
K.
We have a disagreement about something that none of us really know for sure. You think I'm exaggerating about the simplicity of the task, and I think you're exaggerating far too much about its complications and complexity. In my opinion, in 5-7 years from now the direction will already be much clearer and we will be able to see in a more informed and realistic way where the wind is blowing and which of us was right.
I don't think I have a leap of faith, I simply know many examples of very complex systems that contain a lot of items and have a tendency to work out in an organized way, and they arise from a few very simple and uncomplicated rules.
I understand that you have a lot of experience on the biological side, regarding the other side of applying neural networks on a computer I have no idea how knowledgeable you are, but I have heard several world-class experts such as Henry Markram and others who know the subject well from both sides and in their opinion the task is definitely possible in the time frame of two decades from today.
As far as I understand from the lectures I have seen on the topic of a neural network (like our brain) it is a very flexible architecture that knows how to adapt itself to almost any input/output means you connect to it, hands, feet, eyes, wings, nose. There's no need to break your head too much, once you've built it once properly, it will already know how to organize and arrange itself without us having to tell it how.
The video I gave you earlier (a robot with a rat's brain) is just one example of this principle.
rival
They already know how to extend the life of neurons for a year or more. But, what is here is the use of individual neurons connected like an electric circuit. It is absolutely impressive! But it has nothing to do with the issue of consciousness.
K. You put it nicely - it is true that complex systems emerge with new properties, but there is no reason to think that what will emerge is something that will even slightly resemble human consciousness. I personally think that it is much more likely that a trait of violence will arise due to the basic instinct of survival.
The basic consciousness has 4 functions: eat, run away, fight and reproduce. Without these 4 no system would have properties similar to human consciousness.
K.
Until I have time to respond, here's a nice video:
https://www.youtube.com/watch?v=1QPiF4-iu6g
rival,
There is no dispute that the complexity of the network is an important and probably necessary factor, but the existence of a necessary condition does not mean that it is also a sufficient condition.
Since we do not know exactly what the sufficient conditions are for the emergence of "higher" cognitive features, it is quite possible that a system that will be able to perfectly learn to recognize balls and cats (and any other object for that matter) and will also have a vocabulary of the Oxford dictionary will not display interesting features beyond what it was trained to. You have some kind of leap of faith that things will somehow work themselves out, and all I'm saying is that it's a little more complicated than that, precisely because we don't know what the relevant variables are (language? laws of logic? movement? laws of physics? social interactions? and a host of other possibilities like "inner speech" which you mentioned but which is difficult to define clearly) that lead to the appearance of programming such as self-awareness for example.
No matter how much more you add to the pot (more sensors, more detection capabilities, more neurons) there is none of these that require verifying complex features. Even if our imaginary artificial network already has the complexity of a living brain, and even if we manage to provide it with a body that contains enough movement possibilities, sensors and internal controls, and even if we manage to provide it with *all* those inputs that are relevant to such a type of body (and there must be a match between them, Because it makes no sense to give a chick inputs of a fish's living environment, for example), there is no guarantee that the architecture of the network (including its dynamics) will develop and converge to the same solution that exists in a living brain that has undergone millions of years of evolution. There are features that are likely to develop - such as coordination and balance (requiring the neural network to have a body that it activates), and there are features that we simply do not know how likely they are to appear, if at all. I believe it is possible, but we will know for sure only when they manage to demonstrate it for real, and not in any theoretical way. If it happens in two years it will be exciting, and if it doesn't happen even in twenty years I won't fall off the chair in surprise.
Regarding the last question - there is no response. I recommend that you learn more about neural networks to understand the essential difference between training a network with the aim of achieving a predefined goal (like recognizing faces) and between a situation where the goal is not predefined. While the first type is "only" difficult, for the second type we still don't know how difficult the problem is, and you extrapolate from the achievements of the first type to the chances of success in the second type, and this is a wrong projection.
Miracles,
Don't take my name in vain 🙂
I see rich people's troubles here, you built a relatively simple neural network and got more than you expected.
But it's a good question, I'll think about it.
Oops….
rival
So if this computer simulates a small part of a rat's brain, then how does it play a human game? Don't you see the contradiction here?
K.
1. A nice lecture and a nice example at the end, but pay attention to a critical thing that he emphasizes towards the end and that is that the human brain has about 100 thousand times more connections than in the computerized neural network... and this, in my opinion, is what makes the main difference. It's absolutely impossible (and not fair) to take such a tiny neural network, and expect it to do all the things that a human brain that is several orders of magnitude larger than it does!
And yet he still sounds optimistic about their ability to accomplish the task within a few years.
2. "To take a very large artificial network, provide it with a body and sensors and train it on a wide variety of inputs similar to what a happy living being receives during its development"
And that's exactly what I'm constantly talking about... taking a large enough neural network (preferably at least the size of a human brain) and connecting it to a robotic body that will allow it to see, feel, hear and interact with the environment, and start raising it just like a baby!
Show her a ball and say "ball!", show her a cat and say "cat!" So that she will begin to learn the world, begin to learn words, learn a language that will allow her to communicate with the environment, and perhaps begin to talk to herself from the inside, inner speech which in my opinion is a critical and decisive step on the way to human consciousness, the ability to talk to yourself in thoughts.
3. True, we still have a lot to learn about the human brain, but I am quite convinced that with the knowledge we have, if the power of computing allowed it, we could already today build neural networks with consciousness. But unfortunately it will probably take some time before we can test this hypothesis.
4. Why are you a bit biased on the subject, do you have any connection to a project of this type?
A nice lecture in this regard:
https://www.youtube.com/watch?v=bvQlrvmD0AU
It is a well-known problem that there are still many things, some of them very simple for us (see the example at the end of the lecture) that it is simply not clear yet how to get there, that is, how to establish the legality that leads to the correct result given a given input.
Also note that this is the development of artificial neural networks regardless of the architecture that exists in the brains of animals. A complete copy of the structure and *also* the dynamics of a real brain has never been done (nor do we yet have knowledge of how this can be done), although I agree that in principle a complete copy of *both* components should in principle produce the set of features that exists in the copied brain.
The only direction today that has real evidence (and very impressive although in relatively narrow areas) is the direction of artificial networks. While it can certainly be expected that in the future there will be much larger networks that will probably allow better recognition of nuances (in language for example), for which it is easy to create a training set, because the desired output is well defined, this still does not mean that the same (artificial) system, however large, will display Something that was not trained for in the first place (just as Deep Blue who beat Kasparov in checkmate only knew how to play checkmate), certainly when you don't even know what the appropriate training set is. This is a critical difference between a copy of a living brain (where the network's tuning process has already been carried out) and an artificial network that we are tasked with training.
The other way, as I already wrote, is to take a very large artificial network (we don't know how big it needs to be), provide it with a body and sensors and train it on a wide variety of inputs similar to what a happy living being receives during its development (from the embryonic stage for that matter). Even in this scenario, we have no idea what complex features will emerge. I can say that in an experiment already done in this direction, features such as motion detection, self-detection (that is, detection if a certain part belongs to the robot or to the environment) and some other very basic features emerged. In my opinion, this is also the most promising direction, but I'm a bit biased on the subject...
rival
A mouse has a small brain.
And what you claim is that the mouse has no consciousness at all.. not even at a low level. And you yourself admitted that a mouse is a smart animal. You are contradicting yourself. Like religious..
Miracles,
1. True, even today we have complex systems, but they are not a neural network, so I don't expect consciousness to appear in them.
2. "The problem is not building a complex machine"
This is definitely a problem, because the power of computers and chips today allows us to simulate only very small pieces of brain in a size approaching the brain of a rat. That's why we will see interesting phenomena there, such as defeating the world champion in Alpha-Go and achieving a high score in computer games, but in order for consciousness to be created there, we need a much larger (neural) network mass than that.
rival
The example was to show that even today there are complex systems that seem to have "consciousness". You don't need a design grid for complications. I have more such examples 🙂
The problem is not building a complex machine, although building something of reasonable size that is similar in complexity to the brain is still a long way off. The problem is understanding the human mind. It's really far, because our tools are very limited. You can't connect a lot of electrodes and you can't get high resolution information any other way.
rival
In Japan there are already robots - hotel receptionists - who greet you when you enter a hotel. And they also talk to you and serve you (on a limited level). Do they have consciousness?
Miracles,
1. I didn't understand what you wanted to say with the example of the computer, what's more, it's not a neural network. If you feel like it, explain, but it's not critical.
2. "In other words, we will not design consciousness, but it will simply emerge from the machine"
for sure! That is, if the network we have built is close enough and similar enough in its operation and structure to what happens in our brain... if, for example, you copy a working car engine one for one including all the screws and all the pistons, supply it with fuel and start it - it should work, whether you understand it or not Understands how the original engine worked.
The same with the neural network, if you build it precisely enough consciousness will simply emerge from the network, from this complexity, as it emerges in every person on earth.
3. "If she's smarter than us, why won't she hurt us?"
This is already a bonus question that is not related to our discussion, it is similar to your previous question about how to avoid the new consciousness that will create suffering. Of course it can happen and we trust you to find a solution, you have another 30 years, move! 🙂
rival
really? Just like that consciousness will emerge from the box, and we will all be surprised? Even though we have no idea what exactly? You've seen too many movies.
You don't notice, but similar to a religious person - you believe in your righteousness (which is not based on knowledge), and as a result you are captive in a mental fixation just so that your ego is not harmed. This is what makes you write things that in some cases have no justification or do not make any kind of sense.
rival
I am surprised every morning when I turn on my computer and it "decides" to update something and spoil what was working just the day before 🙂
I understand that the consciousness that will be created in your opinion is because of the complexity of the system and not because we understand the consciousness of man. That is - we will not design the consciousness, but it will simply emerge from the machine.
What makes you think this consciousness will be positive? You yourself say that we won't know what's going on inside. So - if she's smarter than us, why won't she hurt us?
Miracles,
Don't get lost 🙂
What I am arguing is that there are enough signs that show that we understand the basic principles of a brain neural network quite well, starting with the neural network of the "Human Brain Project" whose spontaneous electrical activity is remarkably similar to that obtained in a real piece of brain in the laboratory, to the neural network of Deep-Mind which displayed a high cognitive ability that has so far been attributed only to humans, for example the amazing victory in the Alpha-Go game.
I claim that when this network (which is currently a pea size) grows to the dimensions of a human brain, and they connect it to even a basic robot with cameras and sensors, the result may surprise you too.
rival
I lost you... Are you saying that we will build a really complicated design network, that we have no idea what it does - and that this network will have a consciousness like a person's? Of course this network, which we have no idea what it does, has to be connected to a system of senses and muscles like a person has.
Is that your claim?
K.
No problem.
K.
I definitely agree with you:
I also see people - who, even though they are scientists (and present themselves as educated people) - sin against the truth and write nonsense.. there is no shortage of those either. Just look in the mirror..
rival,
Let's see what they achieve in two years and then we'll talk, okay? I am skeptical because I know so many subsystems and features controlled by the rat brain that today, despite intensive research, there is still a very partial understanding of how things happen. Let's make an agreement, in about two years, or however long it takes, when they come out with a new announcement about the achievements that have really been achieved and which can be tested and compared to everything known about the brains and cognitive abilities of rats, we can refer to the issue again and see some of the promise: "To create an artificial intelligence that will do everything that a rat does" lines. Acceptable?
It would be extremely cool if it did happen, for now I'm skeptical (and very happy that the burden of justifying the aforementioned achievements is not on me).
anonymous,
I just saw that I missed your funny comment that compares experiments of analyzing neural activity patterns from lattice cells obtained from electrodes implanted in the entorhinal cortex (this is not really what I did, among other things, but it's not very far from it) between "experiments performed on novice burnets" and between "frog surgeries" Children performing at school" ??? Like, what?!
I don't have to imagine that people are stupid, in fact I don't usually think that my interlocutors are stupid even when they display ignorance (usually quite reasonable) as long as they don't pretend to know for sure knowledge that is clearly not in their hands. These are the fools who flaunt their stupidity as if it is something to be proud of, who make me realize that no matter what - they do exist. I don't know why you insist on continuing to wave it (see your comment quoted above) again, but for some reason I have a feeling that this won't be your last time, and the truth is that it mostly saddens me.
rival
The pessimism of a person leading the business in this field - does not convince you.
What does convince you is your belief in the matter, which is based on nothing and lack of knowledge...
Come on..
K.
1. Despite my name, I'm really not looking to fight with anyone (and I don't like to fight!) even though there are some here who try as if by force to drag me into fights, and as you can see some of them I just ignore.
2. I don't think I need to talk to experts in the field personally, it's enough that I listen to their lectures and the interviews they give. If world-class neuroscientists such as Prof. Henry Markram or Prof. Idan Segev, who have studied the brains of rats and monkeys and humans all their lives, and also have experience with neural networks, think that what I am saying here is possible, then I am probably not alone in my opinion.
3. "Yariv, at the end of my penultimate response to you I wrote: "I believe that it is completely possible and certainly expected to arrive, if we want it or not, in the not too distant future"
It is certainly true that you wrote that, but still the main part of your response (and the previous ones as well) was about trying to show me how complicated it is and how improbable it is.
4. You didn't tell me, what do you think about the link I gave you earlier (from YouTube) about a neural network with the intelligence level of a rat? If they succeed in the task within two years i.e. by the end of 2018, how much will it surprise you? And will it make you a little less skeptical than you are today?
5. I read the article you gave me from the scientist, and I'm sorry but his pessimism did not impress me.
K.
I did read. And I responded too. at the time almost five years ago.
So what did you learn from the article? 🙂
There is another one - search for "the blood is the soul?"
Read, learn, educate yourself... and come back to us with insights we haven't heard yet... what we've already heard - no need to mention, people are not as stupid as you think..
anonymous,
This was really a very nice article!
https://www.hayadan.org.il/is-soul-exist-2501098
You should also read it sometime :-)
rival,
There is a relevant reference in another article here, especially in the context of learning games, and in particular the game Go:
https://www.hayadan.org.il/four-types-of-ai-2811161
Talk to someone who deals with the issue, that's the only recommendation I have for you.
rival,
At the end of my penultimate response to you I wrote:
"I believe it is completely possible and certainly expected to arrive, whether we want it or not, in the not-too-distant future"
You blame me for your last comment that I claim: "There is no way we will get to that in the near future!"
I already wrote to you, not trying to convince you. From what has been written so far, apparently I have been engaged in research in neuroscience for more years than you and I was knowledgeable in the field also in the topics in question (even if they were clearly not my field of research). From the knowledge that I have of the field, the working methods and the approaches that exist today (at least until a few years ago) in this kind of research, I think you have a blind spot about how things work, but that's really just fine, don't take my word for it, read a little more In depth about the subject yourself and most importantly, talk to people who are involved in the field, I'm sure there are some in your environment. I recommend that you start with a focused wish about how neural networks are developed, as complex as you want.
From the first sentence of your last comment, it seems to me that you are looking for a bit of a fight, and I have no desire to fight with you :-)
rival,
Thanks for the addition and the link to Susana's lecture. I heard this hypothesis (the connection between increasing metabolic efficiency through cooked food and the rapid development in the size of the brain, and probably more importantly - in the number of neurons, and perhaps also the connections between them) from someone else almost a decade ago, but here she presents it in a way that seems more grounded and in a broader context In terms of comparisons to other animals.
K.
1 To tell the truth, you sound a bit like a religious preacher... "It's so complex and therefore, only God could create it!", and for you: "It's so complex and therefore, there's no way we'll get to it in the near future!".
2. What do you think of the link I gave you about a neural network at the level of rat intelligence? Let's wait and see, you say? Even if it takes them two more years it will still be amazing, and I guess you agree with me.
3. I completely disagree with what you said earlier that the ability of the neural network to play a variety of computer games does not indicate a high cognitive ability (not to mention the game Alpha-Go which is considered an exclusive thinking ability of a human being). Even if you took a cat or a dog and gave them all the rewards and treats that dogs and cats love (meat, cookies, petting) for getting points in these games, they would not be able to play them, and certainly would not reach the level that Deep-Mind's network reached.
You are trying to downplay the amazing achievement that was here, and I think you are wrong.
4. Everything you wrote is familiar to me, and you still ignore the example of Google's neural network whose achievements are far beyond what people and experts in the field thought possible in our time (I suggest you read a little about the cognitive abilities needed to beat the world champion in Alpha Go, it's not just Calculating a few moves ahead and seeing what is the most profitable like in checkmate, is much, much more than that and the network learned everything by itself from watching human games).
5. To date, no one has yet created a neural network the size of a human brain, and connected it to a robot that experiences the world, therefore you cannot say that such a neural network will not create awareness because it is just a guess that has nothing to hold on to. We won't know until we try, and as I said, there are quite a few clues that allow room for optimism.
6. There may be many configurations in which a neural network can develop consciousness just as proteins of many different types can fulfill the same function in the body. Just because our brains evolved in a certain way does not necessarily mean that this is the only way to create awareness. It is very possible that even in a neural network that will be built in a slightly different way, awareness and thoughts will still develop after it is raised like a human baby.
In my opinion, the extraordinary cognitive abilities that DeepMind's (tiny! crumbly!) neural network has shown, certainly give room for optimism as it grows.
We can also talk about the neural network of the European project ("the human brain") where they try to be as accurate as possible in relation to a real biological brain, there were also very impressive achievements, and there they also intend to connect the brain they created to a robot that operates in the real/virtual world (although I'm not up to date on their schedule) and I expect surprises there as well.
K.
you know what? Even simpler: there was an educated man named M. Rothschild who wrote noteworthy insights.
Among other things, he wrote a short article about the soul. read learn Educate me.
And maybe then start asking yourself first: what is consciousness? where is she
Maybe this is what will help you have a proper discussion with the other people commenting on the site.
K.
I will make it easy for you.
Just answer the questions I asked and then we can know what you understand and what you don't understand.
Currently, all of your experience with laboratory mice (experiments that are also performed for novice barnets. And in fact even children in schools in the US have been dissecting frogs since before you were born) does not indicate any abilities and knowledge that can help in anything in this discussion.
You just got confused between subject and subject.
The subject is consciousness. And the subject is the neuron.
You thought the subject was a neuron and the subject was "How smart are you?" The unprovoked queen of wisdom. Let's cheer you on"…
K.
What you wrote does not answer my questions. And in fact, it is also not relevant to the discussion.
What you wrote does not answer the questions of the mind, but you simply made a salad of concepts unrelated to the discussion.
Except for a few words you wrote that actually reinforce my words (you probably weren't aware of this when you wrote them),
You didn't answer any questions. Just babbling out of Mirmur.
If you don't want to hear an answer then why are you asking a question? What an ignorant behavior on your part.
If you think a mistake has been made - point it out. don't babble
K.
The first link you gave is exactly the link I was thinking about yesterday, and in addition to the correct (or at least the logical) things you added regarding the relationship to body size, you can also see in the link that an elephant's cerebral cortex (which is the most important part in the context of intelligence) contains only half the number of neurons that exist in a human. And this is without talking about the amount of synapses and connections that exist for each neuron in a person.
And of course there is also this excellent lecture:
https://www.ted.com/talks/suzana_herculano_houzel_what_is_so_special_about_the_human_brain?language=he
anonymous,
Check the facts, you have a mistake.
https://en.wikipedia.org/wiki/List_of_animals_by_number_of_neurons
And for your information, there is a connection between the size of the brain and the size of the body:
https://en.wikipedia.org/wiki/Brain-to-body_mass_ratio#/media/File:Brain-body_mass_ratio_for_some_animals_diagram.svg
For the simple reason that controlling and controlling a larger body requires more nerves, including nerve cells in the brain, and this is regardless of cognitive features, such as the ability to think. In addition, there are parts of the brain that can be enlarged and related to memory, for example, and this is also not necessarily related to "higher" cognitive abilities such as self-awareness. Why are you writing about things you have no idea about?
And although I'm not sure I want to hear your answer... but where do you think your consciousness is?
rival,
By complex systems in the context in question I of course meant neural networks. You have a logical fallacy, the fact that we know that there is a complex neural network that displays consciousness (what ever death means) does not follow that *every* neural network of similar complexity (number of neurons, number of links, etc.), or even composed of it, will display these characteristics. I don't know how much experience you have in training neural networks, but when you want your network to have certain properties, you don't just throw a lot of neurons into a "pot" and create a lot of connections and hope that the network will start behaving the way you want. You also don't just add more neurons and more connections if it doesn't behave the way you want it to. Then again, it's probably a necessary condition for the system to be complex enough, but it's clearly not a sufficient condition, and you can see this for simpler properties than consciousness. You can create an incredibly complex neural network and still not have a simple, basic motion detection feature for example. To get this, you need to put into the system a suitable input and validity that will cause the network to adjust to the desired feature (and even then, the particular architecture you built might not be able to learn the task if you didn't take into account some other important factors). Since we don't know what the relevant inputs and laws are to create consciousness, we don't know how to design the neural network (which must also be incredibly complex). Another way to approach the problem, as I mentioned earlier, is to follow the path we know leads to the creation of consciousness, that is, to imitate a neural network of a human embryo including all the internal developmental processes and the appropriate real-world inputs. In principle, it is possible, and it is likely that we will thus receive a machine with a relatively developed consciousness. Baby X is a nice example in the direction, but it's still far from the real thing (which I believe is completely possible and certainly expected to arrive, if we like it or not, in the not-too-distant future).
rival
How do you think consciousness is in neurons at all?
Elephant - lives 70 years on average. As of today.
In an animal like an elephant, there is a brain bigger than a human. And the amount of neurons in his brain is almost the same amount as in humans.
The elephant is a much less intelligent animal than the human being.
How do you explain the abysmal difference in human intelligence versus an animal like an African elephant?
Is it only the amount of neurons and the speed of information transfer between neurons that determines the level of intelligence?
Or does the language also 'shape' (and improve) the cognitive abilities of the creature?
And are there other factors that affect the level to which the creature's consciousness reaches? If so, what are they and how do they affect?
Miracles,
No we are not close to it, but this question is not relevant at all. Already today the neural networks we build and train are so complex that even the people who built them cannot tell you in advance what decision they will make for each input, they must run them against the input to see the result.
The team that built DeepMind's neural network, for example, cannot look at the network at a certain moment (without knowing in advance what it was trained on) and tell you if it is already an expert in a certain computer game or not, they will have to let it play the game to know what the answer is.
So your question is irrelevant, the neural networks we build can be specialized in a certain field or fields even if we cannot know in advance what, until we activate them and try.
rival
Obviously I believe that a robot can be conscious as well as self-aware. The difference between this and a person's level of consciousness is, in my opinion, abysmal - and this is the point I am trying to clarify.
Do you think we are close to diagnosing a person's political opinion based on their brain scan? What color does he like? Does he prefer Beethoven's Ninth or Bach's aria on the G string?
Can we look into a person's mind and know what kind of car they like, or how good they are at riding horses?
Miracles,
He may well be scared when he is led over the edge of a cliff, he will successfully pass an IQ test, he will experience virtual reality just like you, and if they connect him to taste and smell sensors he will learn like a human how to treat sour milk and smelly cheese.
If the neural network in our brain can do all these things why do you think an artificial neural network can't?
I don't understand you at all, you keep claiming that the thermostat in your car is self-aware, so why does it seem strange to you that a neural network that is a million times more complex would be self-aware?
rival
How will he behave when they put virtual reality glasses on him and let him "walk" on the edge of a cliff - will he be afraid? Will he know that sour milk has a bad smell? Will he know that stinky cheese is delicious?
Miracles,
Re-read section 3 of my previous post and tell me if it doesn't answer what you asked.
rival
You won't require that robot to pass an IQ test for example?
Miracles,
1. In my opinion, there is no specific moment when one can say: "Before he was not conscious, and now he is!", just as it is impossible to say that at a certain moment in evolution a monkey suddenly became a man and his parents were not.
2. I didn't focus on the question of what consciousness is because I simply don't have an exact answer, but broadly it means that some system knows that it exists, it looks in the mirror and realizes that it is looking at its own reflection.
3. I repeat, when a robot behaves exactly like a human in every field and test, and it tells me that it is self-aware, then for me it has passed the test.
Won't that convince you?
(Suddenly this reminds me of the discussion we had regarding a series of random values, and whether in order to determine whether it is random or not we must open the hood and check the mechanism that created it)
rival
The right question is "what is consciousness". So far you haven't even tried to answer the question. In my opinion - consciousness is a reflection of the environment in the state of the system. And so the thermostat has consciousness.
Self-consciousness is simply a reflection of the inner state. And so the car is aware.
Our discussion is not about whether the car is conscious or not - the discussion is about the essence of the word consciousness. You're pretty evasive at the moment. We both agree that there are dimensions of consciousness, from a person to a grain of sand. The question is - where is the transition from conscious to unconscious. Try to answer it 🙂
Wow the first link is pretty amazing 🙂
I will check out your links.
Miracles,
How many people have you managed to convince so far that your car has consciousness? 🙂
rival
How do you determine that the car has no consciousness? It is much more complex than simple worms. What is your test of consciousness?
Lem is a science fiction writer who writes a lot about non-human consciousness. Read the following story:
http://home.sandiego.edu/~baber/analytic/Lem1979.html
rival
For you - http://www.howtogeek.com/elk/1429/14796/27338
Ha, Stanislav Lem is the name of a writer...
Miracles,
It created electrical patterns and electrical pulses that are completely identical to what is seen and measured on a piece of brain in the laboratory.
Miracles,
1. I will take a look at the book you recommended, but in any case we already have 8 billion examples in the world of neural networks with consciousness, and we know exactly what the result is 🙂 many positive things, and many negative things.
2. "Obviously, humans, at least most of them, have consciousness - but what about a chimpanzee? What about a rhesus monkey? Dog? rat? rat? alligator? A fish? Fly? diarrhea? Worm K. Elegance? amoeba? A bacterium? Where should we draw the line?'
As I understand and see it, it is like a lamp whose light can be dimmed with a button, it can be very bright and it can be very dark. A person has some level of consciousness on top of the scale, a monkey has a lower level of consciousness, a rat has an even more dim level of consciousness, and the thermostat in your car has no consciousness at all 🙂
3. "If you don't have answers - how will you know that you have created consciousness?
So again we return to the question I asked you before, how do you know that your wife and children are self-aware? For me, as soon as I see a robot that functions exactly like a person, and passes all the tests that humans pass, and he claims to have consciousness, and feelings, and thoughts, then I will believe him 🙂
rival
"A (quite large) piece of brain that showed exactly the same behavior as a real piece of brain, so it doesn't fit what you say."
what? How exactly does a piece of brain behave?
rival
Why don't you send an email to that group of scientists, and tell them your opinion. Ask them for answers to your questions and if they answer you, I guess you will understand better.
Because right now the discussion seems to have reached an impasse.
rival
You got confused - I think you meant that you are pessimistic that consciousness will open on a human level! If you think for a moment that developing a mind like a person's is a positive thing, then you should read Stanislav Lem a bit!!!
It is clear that humans, at least most of them, have consciousness - but what about a chimpanzee? What about a rhesus monkey? Dog? rat? rat? alligator? A fish? Fly? diarrhea? Worm K. Elegance? amoeba? A bacterium? Where should we draw the line?
And what about old people? Unconscious? People in a vegetative state ("plants")? People in comas? Where is the border here?
If you don't have answers - how will you know that you have created consciousness? And more importantly - how will you know that consciousness does not suffer (you must read Lem!)?
K.
1. "There are incredibly complex systems that do not have any self-consciousness or any other feature that belongs to the broad concept of "consciousness"
There are many complex systems without consciousness, but we have one working example and it is the human brain, and there it is a neural network and not just a "complex system".
2. "For a large neural network to develop properties similar to what every baby develops, it is not enough for it to be very complex, it just doesn't work that way"
I didn't understand, on what basis are you saying that? Has anyone ever created a neural network the size of a human brain and connected it to a robotic body that acts and experiences the world? If not, then seriously, how do you know?
K.
Here is a link about a neural network at the level of intelligence of a rat:
https://www.youtube.com/watch?v=MAMuNUixKJ8
I assume that he too should be brought to the ground rather quickly, as well as Prof. Idan Segev, and Henry Markram, and Jeff Hawking and dozens of other senior neuroscientists who also have considerable experience in studying the brains of rats and monkeys and humans, and surprisingly think that the task is actually possible In the time frame I mentioned earlier.
rival,
I'm not going to try to convince you because honestly you're starting to sound a bit like a religious person defending his faith, that is, not really, but yes with too many superlatives about the depth of self-conviction and too little familiarity with what's happening on the ground. To recap and say, I think you have a blind spot regarding the nature of the challenge facing neuroscientists today regarding the acquisition of cognitive traits (learning a series of different games is not fundamentally different from learning one game and is not related to "higher" cognitive traits such as self-awareness). If you ask to speak with one of the researchers in this field who is close to your area of residence, I guarantee you that they will put you down to earth quite quickly. In particular, it is not just to add complexity to the system, there are incredibly complex systems that do not have any self-consciousness or any other feature that belongs to the broad concept of "consciousness". In order for a large neural network to develop properties similar to what every baby develops, it is not enough (even if probably necessary) that it be very complex, it just doesn't work that way. In any case, that's fine, we're all eagerly awaiting breakthroughs on the subject, and the important thing is that we agree that there is nothing in principle that prevents science from cracking this (hard) nut.
As someone who has worked with rats quite a bit, I allow myself to be quite skeptical about the promise of a neural network with the same level of intelligence as a rat. Do you have a link to that report? I would love to understand better what it is based on.
rival
To say that "if you connect many neurons then you will have consciousness" is like saying that if you connect many thermostats then consciousness will be created..
If one neuron has no consciousness - how do you know that many neurons do build consciousness?
A fish has many more than one neuron. Is a fish a self-aware animal? Is the fish aware of the environment? Does he have consciousness? Or does the fish even respond to electric pulses and nothing else?
Miracles,
1. To say that the thermostat in your car has consciousness is like saying that one brick lying on the ground is a house. I also don't think a single neuron has consciousness. In order for these components to create something that can be called "consciousness" there must be some minimal complexity, similar to an airplane that can only take off above a certain threshold speed. A Boeing 747 plane cannot take off at a speed of 20 km/h on the runway, and similarly a single neuron cannot develop consciousness, and certainly not the thermostat in your car.
2. "Building consciousness is not the problem, the problem is a definition of what consciousness is"
I really don't think the problem here is a definition of consciousness, I'm sure you don't doubt that your wife and your family members are self-aware even though you don't have a precise definition of what consciousness is, so definition is not the problem here.
My argument is simple, as soon as you build a neural network massive enough and complex enough to be built in a similar way to our brain, then consciousness will simply emerge from there spontaneously as it happens in the brain of a baby that is born and begins to experience the world.
3. "All these games you talk about are simple"
These game miracles are really not simple and until recently no one believed at all that a neural network could reach such incredible achievements so quickly. Even a 10-year-old boy has trouble reaching the level of play that Google's neural network has reached. And again I remind you of an important point... this network of neurons is not even a thousandth the size of a child's brain, so as they say "it's really not powers".
4. I don't know who made the calculation you talked about and what this calculation is based on, but it doesn't make sense to me. In the European human brain project, for example, we have already built a simulation of a piece of brain (quite large) that showed exactly the same behavior as a real piece of brain, so it doesn't fit what you say, in addition, according to what I've read, in about 10 years, a supercomputer will already have a computing power on the order of a human brain .
I am definitely optimistic about this.
rival
The phrase in Hebrew is indeed an "emerging feature"...
rival
They already did a calculation and came to the conclusion that in order to simulate a human brain one must build a machine the size of the human brain, with neurons the size of real neurons.
All these games you talk about are simple - to simulate a human brain - the machine has to be able to build a machine that is able to simulate a human brain!
Building consciousness is not the problem, the problem is a definition of what consciousness is.
In my opinion, the thermostat has consciousness, and the car has self-consciousness - so far, no one has contradicted what I say.
Intelligence - I have no idea how to define, but if we take the Turing test, we are many decades away from building a neural network that will pass this test.
This is a message for miracles, which begins with -
1. "That's exactly the point...
Finally my comment from last night is released.. really thanks.
Shhh... my comment from last night is finally released... this is the comment that starts with -
"Miracles,
1. "That's exactly the point - you can't assume such a thing..."
K.
thanks for the detailed answer.
1. First of all, I completely understand your skeptical attitude and that of Nissim, but I have formed opinions on the subject that I am quite convinced are correct, and this from everything I have seen and read and heard to date on the subject, including lectures by experts in the field. So maybe I'm wrong but I don't think so, and in any case if in another 30 years we'll still be alive and well here with God's help then we won't have to guess anymore and we'll just see, right? 🙂
2. What is amazing about the neural network of this subsidiary of Google, is that it does not specialize in one narrow and specific field such as face recognition. The same neural network without any change (according to the company's manager in a lecture on YouTube) learned by itself to play nearly 60 different computer games, each of which works in a completely different way (car racing, breaking walls, Pacman, airplanes, shooting, Super Mario, boxing, XNUMXD... ) without any guidance or direction, just by looking at the screen and the score! Listen, this is really not something to be underestimated...
3. In addition, they intend to present within a year a neural network with the same level of intelligence as a rat! that you operate a rat robot, and a rat is a really smart and intelligent animal for those who don't know!
4. In addition, the neural network of the famous European project also shows a behavior that is remarkably similar to what you see in a real brain, the same electrical activity and the same patterns typical of the brain, and this is also an important indication of the correctness of the model and the importance of this cannot be underestimated!
5. I'm not sure that the directions should be as subtle as you imply... there are 8 billion people in the world and each one has a different brain that is wired differently, and different directions and still at least 70% of them are self-aware at one level or another...
6. My strong impression is that this is a phenomenon that is created (I will not write the nomenclature in English because maybe that is what blocked my previous message) and therefore when we start to create massive neural networks, consciousness will also begin to form in them without us having to direct it, it will simply emerge from the complexity of The system and the extensive electrical activity that will be there. Just as it slowly appears in the mind of a baby.
rival,
As far as I know it's a bit more complicated than how you present it. Neural networks are amazingly successful in certain defined tasks, when it is relatively easy to define what the desired "behavior" is (like winning a game, or achieving equilibrium). This is not such a good example of creating a situation where characteristics that we associate with "consciousness" will spontaneously emerge as in living beings, which are not so well defined and sometimes not defined at all. It's a leap that simply isn't clear how difficult it is. A larger and more complex neural network may learn things more easily and reach better achievements in the areas to which it has been adjusted, but that's the thing, it is not yet clear what adjustments must be provided to the system so that later cognitive characteristics will spontaneously emerge. One possibility (of which there are some interesting attempts in this direction) is not to define in advance what the system needs to learn, but to let it choose what it prefers to focus on in an environment that simulates the inputs that exist in real life (in terms of forces, mirrors, social environment, etc.), with the expectation that if there is in the same the appropriate computational potential system, so the features that develop in each baby may also appear closely in it. But we have no idea yet what the minimum required for this is. A developing embryo does not start from scratch, there are pavements that slow down evolution and create shortcuts. Another way is, as mentioned, to simulate existing brains, but it turns out that there is a lot of plasticity in the system, and that quite different constructions can produce similar cognitive functions (similarly to what happens in sensing and movement loops). So it's true that it's important to know the neuroanatomy of the system you're trying to study, but go to the websites of scientists in neurobiology and see what their areas of interest are and you'll find that an extensive occupation (if not the main one) is in deciphering neural transmission and the controlled variables that can be extracted from it. The microneuroanatomy is less relevant. This is what happened a few years ago when I was still doing research in neurobiology and to the best of my knowledge the situation is no different even today, as far as being up-to-date is sufficient.
Anyway, I suggest you don't rely too much on TED talks, but try to talk to someone in the field or at least read the introduction that scientists write on their lab's website that tells in general about their interests and active research. Search by keywords about what interests you, it's a huge field. In my opinion, you will understand quite quickly where the center of action is and what we perceive as the main and immediate challenges of this period, and I believe that then you will better understand what Nissim and I are saying.
rival
Lezmaltz is not Hebrew. There is a suitable word for image from the word simulation. So please write Hebrew and not Hebrew. A moment of Hebrew.
...the result will be that people who deal with it will tell you that it is a robot. unconscious And such a thing already exists today.
K.
"Knowledge of the structure and connectivity of the neural network is not enough to understand the brain. For this, information is needed on the electrical dynamics that ran in that construction (which can also change over time), and this is much more difficult to measure today because of a trade-off (in Hebrew, weighting of tradeoffs if I'm not mistaken) between the signal resolution and the number of measured neurons, and today we're simply not there."
Again, it is clear that we still lack knowledge, but I do not think it is possible to make a statement that this knowledge is critical and without it this would not be possible. On the contrary, in my opinion, there are many signs that show that perhaps the principles we have already learned about the brain are certainly sufficient, or at least bring us very close to the goal.
If a tiny neural network that contains most of what we have already learned about the brain, manages to learn to play dozens of computer games (some in 95D) and reach the level of a professional player in XNUMX% of them without any training just by looking at the screen, I think that means a lot.
What will happen if we take the same network, with the same principles and expand it ten thousand times? Or a hundred thousand times? And connect it to a robotic body with sensors, hands and feet, cameras... Who knows what the result will be?
The last part was a bit difficult for me, so again -
It seems to me that even if they are not 100% accurate, they will still produce all (or most) of the phenomena that occur in the human brain, and it will only get better over time. There is no need to wait until then, we are already seeing the beginnings today and it looks promising.
K.,
I completely agree with most of what you said, and specifically regarding brain simulation as I see it (from reading and watching lectures in the field) we certainly understand the principles well enough to succeed in the task in a few decades.
I see this in the human brain project that Israel is a partner in, I see this in the DeepMind project owned by Google, and in other companies, neural networks that are a tiny fraction of our brain in size and complexity, already manage to recognize objects in a picture, recognize faces, recognize speech even in a noisy place, learn Play by themselves a variety of computer games, including in XNUMXD and reach the level of a professional player! And more and more... definitely unique human abilities. So what will happen when they start to grow and expand, and reach the amount of connections that exist in a human brain? What abilities will they have then?
It seems to me that even if they are not 100% accurate, they will still produce all (or most) of the phenomena that occur in the human brain, and it will only improve over time, we don't have to wait, we are already seeing the beginnings today and it looks promising.
Avi Blizovsky,
My poor reaction from last night is already crying for you to let her go, so what?
To understand a phenomenon you don't need to know everything about it, usually it is enough to understand the basic principles so that it is possible to reproduce the phenomenon. This has proven itself in science so many times already, including in relatively complicated systems, that there is really no reason to think that this is impossible even if the challenge of understanding the brain is perhaps the biggest challenge facing science today. At the same time, I agree with Nissim, that knowledge of the structure and connectivity of the neural network is not enough to understand the brain. For this, information is needed on the electrical dynamics running in the same construction (which can also change over time), and this is much more difficult to measure nowadays because of a trade-off (in Hebrew, weighting permutations if I'm not mistaken) between the signal resolution and the number of measured neurons, and today we are simply not there. But you have to understand that this is a technical problem, however difficult it may be, and not a principled problem, and the history of science shows that technical problems are eventually solved, but for that, scientific curiosity and tenacity and hard work are needed and not a defeatist mental fixation that says - it's impossible because we haven't succeeded so far.
rival
You ignore the fact that there are things in nature that you are not aware of even when you make calculations and you are sure that you have taken all the data into account.
Even top scientists looking for neutrinos make mistakes when they don't take into account something as simple as properly connecting the cables of their computers...
Miracles,
1. "That's exactly the point - you can't assume such a thing. There is critical information that you have no way of knowing today. And in my opinion - not in the near future either.
I'm not sure you're right, the brain is indeed very complex, but this complexity is created from some relatively simple structures that are duplicated billions of times inside the brain, this is complexity created from simplicity (read Emergence on Wikipedia).
2. "The connections do not represent all the information. The information is two-dimensional (frequency and instance) but the information contained in the strength of the ties is one-dimensional. I don't understand how you think it can be overcome.'
I don't understand what exactly you mean, what is frequency and show? And in any case, in the human brain project that Israel is a partner in, they have already created a realistic simulation of a piece of brain consisting of hundreds of thousands of nerve cells (as far as I remember, I don't guarantee the exact number) and the electrical pulses created in the simulation while running and the electrical patterns that ran inside matched remarkably well what you see in the laboratory on a real brain , the simulation even gave several predictions that were verified in the lab!
So listen, you can't say that we are completely wrong, and don't forget the DeepMind neural network (owned by Google) that learned to play computer games on its own, even those in XNUMXD... Definitely human abilities that appear by themselves in a simple neural network... What will happen when this network grows and expands?
rival
"Regarding the example with the cars, let's assume for the purpose that their speed is only affected by the traffic lights." That's exactly the point - you can't assume such a thing. There is critical information that you have no way of knowing today. And in my opinion - not in the near future either.
I understand what you are saying about the game, but in the context of neurons - the connections do not represent all the information. The information is two-dimensional (frequency and instance) but the information contained in the strength of the ties is one-dimensional. I don't understand how you think this can be overcome.
anonymous,
I would answer you but your incessant cursing and your disgusting discussion culture shows that you are simply not conversational.
Miracles,
This was just an example of course and it is obviously not perfect, but to be more precise then the information on the chip will constantly change dynamically as the connections in our brain change. Therefore, when you start the game again it will continue from the same point you reached in the last game.
And regarding the example with the cars, let's assume for the purpose that their speed is affected only by the traffic lights, the traffic lights and the number of lanes on the road, since these are the same in both cities (we built the same infrastructure) then the flow of vehicles will also be similar.
rival
"I think the situation we're talking about is much more similar to a digital chip that you burned a computer game into. As soon as you connect an electrical voltage to this chip, it will "come to life" and you can play the game again."
Yes … but the game will start from the beginning 🙂 But, our brain, materially, is no longer the same brain we were born with.
The problem with cars is that you don't know how fast they are. Therefore, you will not get the same traffic.
K.
Your smiley came out a little crooked,
rival
And you must know how to build a machine like our brain, yes?
Because you know how our brains are built...
Do you think you can build something you don't know what it is?
you are insane.
anonymous,
I'm saying something simple, it doesn't matter if you know or don't know how to measure consciousness or define it, as soon as you create a machine (a digital chip for example) that is built like our brain, and operates according to the same principles, then everything that is created in our brain will also be created in the machine, and if Our brain has consciousness so the machine we built will also have consciousness.
K.,
I understand exactly what you're talking about, and yet I try because I really enjoy these conversations 🙂 until I get tired of them.
Miracles,
4. "Imagine that you take your computer and turn it off, without saving the state of the memory to disk. When you turn it on - you will not be able to return to the state before it was turned off. The state of our brain cannot be "written to a disk", certainly not in the coming decades.
I think the situation we are talking about is much more similar to a digital chip that you burned a computer game into, as soon as you connect electrical voltage to this chip, it will "come to life" and you can play the game again.
5. "Think of a crowded city with a million cars - do you think it is possible to reproduce the traffic situation just from a static photograph?"
Let's say that the city represents a physical brain (that is, the neurons and the connections created between them) and the cars that drive in this city represent the thoughts, and the consciousness -
If you build a second, parallel city, built in the same way as the first city, with the same buildings, and the same bridges, and the same squares, and the same narrow alleys, and the same highways, and the same traffic lights with the same timings... it is likely that the traffic in the new city will be very similar to the existing one In the original city, the traffic jams will be created in the same places, the "pulses" will be at the same frequency (according to the cycles of the traffic lights - red and green) the flow pattern of the vehicles (=thoughts) will be very similar to the flow pattern in the original, which means that this city will also have consciousness.
rival,
He meets one of these every morning in front of the mirror, but because he is not self-aware, he believes that he has already met many of them :-)
What a waste of electricity.
rival
for sure. Such is the situation.. 🙂
I'm really not sure that Nissim is aware of himself.. 🙂 I guess he himself doesn't know how aware or unaware he is of himself.
Are you aware of yourself?
How do you know you are self-aware?
How do you even know you exist? Maybe you're even a computer simulation of aliens? 🙂
I don't want to tell you how many people I've met in my lifetime who are not self-aware. You will usually be able to identify such people when you notice that their actions stem from instinct and not from conscious thought.
rival
3. I certainly agree with that. I think the thermostat also has a very small amount of consciousness. Self-consciousness needs something more complex. Just yesterday my auto informed me that the air pressure in the front left tire is lower than 27.
4. Here I think you are wrong. A large part of the state of the brain depends on the timings between the different pulses. Think of a system consisting of many, many computers, with the output of each computer providing the input to another large number of computers. In this situation, the information is not on each computer separately, but on the lines between them. Think of taking your computer and turning it off, without saving the memory state to disk. When you turn it on - you will not be able to return to the state before it was turned off. The state of our brain cannot be "written to disk", certainly not in the next decades.
5. And to clarify - these pulses are analog signals, what is important is the timing between the pulses. This is information we don't have and in the coming decades we won't have it either.
Think of a crowded city with a million cars - do you think it is possible to reproduce the traffic situation only from a static photograph?
anonymous,
"You don't know what consciousness is. If you don't know what consciousness is - how will you know that what you created is consciousness? You don't know what consciousness is!'
According to what you say we cannot know, for example, if our friend Nisim has consciousness. After all, you don't know how to measure it, so how do you know that miracles are self-aware?
Miracles,
1. We've already had this discussion several times before and couldn't come to an agreement, so I don't know if there's any point in trying again.
2. You talk about copying the consciousness of a specific person, which is a fascinating topic and I have a lot to say on the subject, but it's better if we save it for another time.
3. I claim (as you know from previous discussions) that when you create a sufficiently massive neural network, which operates according to the same principles of our brain, and you connect it to a robotic body that has sensors, and cameras (as eyes), and hands and feet, etc., then it will probably be created In this network is consciousness, just as it is formed in the brain of a baby (I don't think that a fetus in its mother's womb has consciousness, it needs to be born first and experience the world a bit in order for consciousness to develop in it).
4. "A dead brain does not contain the information of a living brain"
The question is why do you call it "brain dead", if the brain is damaged (as happens in brain death) then it is clear that it no longer contains the information that was in it, because the brain cells are damaged (due to internal hemorrhage for example) but if it is a case of clinical death, and the brain is still intact, So as soon as the blood and oxygen supply to the brain returns, so do the thoughts and consciousness.
5. "Thought is a collection of pulses at different frequencies and occurrences and not just connections between neurons"
I completely agree with you, I only claim that the pulses you are talking about are created as a result of the structure of the brain and as a result of the connections between the neurons. Therefore, when you build a neural network with a similar structure to the one that exists in an average human brain, the pulses that represent the thoughts will also be created in it, meaning that this network will begin to think, and become aware of itself, just as happens in a human brain.
rival
1:
A: You don't know what consciousness is.
B: If you don't know what consciousness is - how will you know that what you created is consciousness? You don't know what consciousness is! So even if you create consciousness you will not know that it is consciousness. So how do you allow yourself to say that you have created consciousness? Out of egoistic considerations? After all, you don't know what you created, so how do you claim that you do know what you created? This is a lie. did you understand
If you claim that in 30 years the computer will have consciousness, and you don't know what consciousness is - then you are lying.
By the way, "30 years" - they said it 40 years ago.
2: So there are believers who think like this.. there are non-believers who do not have God and therefore they think that they are allowed to behave in burglaries and they will not be punished for it... so they think... so what?
I personally do not believe that the soul is a consciousness that continues after the death of the body.
But as long as we don't know what a soul is and what consciousness is - then the discussion about it is pointless and meaningless.
rival
I meant to read - you have to read the mind before you can simulate it.
Today we know how to connect electrodes to individual cells in the brain, and I mean individual, not even dozens.
An MRI device, which knows how to read, in a rough way, the state of the brain knows how to read with a separation of 2 mm, that is, several billion neurons per sample.
The ability to improve these two measuring devices is limited. We will probably not insert 100 billion electrodes into the brain, and the increase in power of MRI devices is very limited due to the radiation.
A dead brain does not contain the information of a living brain! He may (very possibly) hold the memories but not the thoughts. It's like shutting down a computer that doesn't have a hard drive. The point is that thought is a collection of pulses at different frequencies and occurrences and not just connections between neurons. We know this from several sources. One is the brain waves, the second is epileptic effects, the third is loss of consciousness which causes the loss of memory in the short term, the fourth is dreams and there are more and more.
I'm not saying that we won't reach a computer with the thinking ability of a person, but in the foreseeable future we won't be able to simulate the consciousness of a certain person.
By the way - there is an experimental way to see if I'm right or not: if we can freeze a person and then wake him up, then I'm probably wrong.
Miracles,
"In how many years do you think we'll be able to read a person's mind?"
What is "reading a person's mind"? You probably meant to ask: "In how long do you think a computer will have a human level of consciousness, or higher."
In my opinion, based on the growth rate of this amazing field, and based on the rapid progress I see - within 30 years at the most, and 20 years seems more reasonable to me.
anonymous,
1. "Even when there is a mind built from a quantum computer, you will not be able to reach the level of human consciousness through a computer"
That's what you said, that a computer cannot have the level of consciousness of a person, so what exactly did I not understand?
2. "On what basis do you base this claim?"
And then you laugh at my lack of understanding... If you didn't understand this is not my claim, but the claim of many believers, they call it "soul". I asked you what you think about this claim, complicated?
Miracles
You ask him as if he knows… 🙂 🙂
Are you writing this to stroke your ego after losing an argument about God? 🙂 🙂
After all, if you wanted to get an answer to your question, I'm sure you would turn to people who are much more qualified than him to answer it... 🙂
rival
As usual, you did not understand what you read, and based on your misunderstanding, you construct a distorted argument.
1: That's not what I said.
I said that if you do not know what consciousness is then you cannot know that what you have built is consciousness.
2: What do you base this claim on?
Joseph
I'm sorry but I can't understand what you mean.
You write that you don't believe, and I don't understand what you don't believe? And what does this have to do with what I said?
I wrote that if you know what x is, how can you know that what you built is x?
If you don't know what consciousness is - how will you build it on a computer? It doesn't make sense what you claim.
rival
In how many years do you think we will be able to read a person's mind?
Hi Anonymous,
I see that you are not interested in talking about God, so let's talk a little about consciousness:
"Even when there is a brain built from a quantum computer, you will not be able to reach the level of human consciousness through a computer. And you won't even come close to a better understanding of consciousness. Even if the computer is like the human brain, one for one, you will not understand consciousness. It doesn't matter in what way you create consciousness - it will not be an explanation of what consciousness is.
1. On what basis did you infer that a computer could not have a consciousness like that of a person, or even higher? Is this a guess or based on something?
2. What do you think of the claim that a person's consciousness can continue to exist even after death?
What the last 3 years have done in a nutshell: the computing power available to neural networks as hardware and software models has increased by orders of magnitude. The giant companies Google, IBM, and Elon Musk's OPENAI entered this with tremendous financial power. At the universities of Princeton, MIT, Berkeley and others, new models of artificial intelligence came out. A sub-branch called deep learning.
There are two computer projects that will provide even greater computing power: a quantum computer at IBM, a neurological computer at IBM. The new layer created is not yet of consciousness, but of the decoding of the sense gathered by sight and hearing, and the creation of decoding in natural human language. (Noam Chomsky also has a share in the matter). The level of decoding of photographs and even just graffiti is very high and for the first time passes a human level. In the near future we will see more vehicles, aircraft, autonomous robots. For example, for a trip to Mars, such are necessary, and that's why Elon concludes on the matter. Beyond that, consciousness research centers were established in the world - they did not usually produce a dramatic direct product. But one psychiatrist, for example Tononi, developed a mathematical-physical theory of consciousness. A theory that is similar to condensed matter theory. Explaining phase transitions using thermodynamic tools, which explain the formation of levels of consciousness in multiple neurological connections. Above a certain threshold these create a constant change in the algorithm level. The fact that there are multidisciplinary centers for the study of rationality, gives legitimacy to the publication of studies on consciousness, which was not the case before. The number of skeptics and opponents is very large. In the end, humanity, if it does not enter the middle ages, will succeed in developing artificial intelligence, and in my estimation it could take between 1 and 100 years.
Sorry. I don't believe that. By the way, I am a person of faith and this is not relevant to the question of whether consciousness can be developed or not. As well as other questions: was there a big bang, is evolution true. We can create consciousness and some are concerned with the question of what it is
Consciousness as a mathematical model. Even if there is a God - irrelevant to the question. From the progress I see in the academy, we are not there but we are making great strides on the way there.
Joseph
I did not understand what you wrote. But from the few that I think I managed to understand - so I'm sorry to tell you - but even when there is a brain that will be built from a quantum computer, you will not be able to reach the level of human consciousness through a computer. And you won't even come close to a better understanding of consciousness.
Even if the computer is like the human brain, one for one, you will not understand consciousness.
No matter which way you create consciousness - it will not be an explanation of what consciousness is.
And if you don't know what consciousness is - how will you know that you have created consciousness?
The comment that was deleted about Peza crossings was correct. The use of thermodynamic tools, which passed through Claude Shannon for numerical communication to terms such as the channel capacitance, passed into all areas of modern mathematics as the N*log(N) matrix (whether it is Pontreygin, or Grothendieck or Saichi Mozuichi - I just listed a few big names in the mathematics of 20-21 centuries. For example geometrical number theory, for example basic concepts in p-adic number theory. Move on to the analysis of consciousness. These are the tools Tononi uses for the theory of consciousness. As I expressed an opinion - it is a step on the way and not the end of the way. But it is a complementary stone to deep learning - in order to eventually arrive at the algorithm that I will say.
Precisely the comment
We lack insights into consciousness - what it is. Layered deep neural networks are sensing and sensing interpretation. But Tononi says that beyond the threshold of connections between neurons - a phase transition occurs, as in physics liquid, solid, gas, plasma and the laws change. Micro and macro understanding is achieved. Tononi is not the end of the story in the mind, but he has stepped on the way. However - Google's machines already are - referring to the software algorithms that run on their supercomputer, catch up with the person, in the interpretation of images, creating text in the language of humans. The big oligarchs of commercial science: IBM, Elon Musk, OPENAI, and GOOGLE, are currently the main contributors to the development of this science. as well as the universities Berkeley, MIT, Princeton. And one last point. To all those who disparage Arabs or Muslims. There are more of them in this science than Jews. There is a young doctor named Raja Jiris in Tel Aviv, and there is Professor Alex Bronstein who slowly and quietly got into the matter.
The most recent progress, which began in 2012 and has entered Israeli universities only now, is deep learning. or
convolutional Artificial Neural Networks = deep layer ANN. In addition, significant progress has been made in the hardware field in that a graphics accelerator card (GPU) from NVIDIA, for example, the one used by gamers, accelerates a neural network calculation that takes a month on a normal processor to 15 minutes. I estimate that using IBM's SYNAPSE hardware, which is a chip that contains about 48 million neurons, will speed this up in the blink of an eye.
Another advance is that it's science, and it's mathematics and not just engineering. The latest progress and here I am subjective to differentiate from the previous ones: Guillemo Tononi psychiatrist publishes a mathematical theory (already in 2004, we just didn't notice before) of consciousness, based on digital information theory. literary communication. It defines how a quantity of consciousness is calculated. Whenever he says algorithm I am. This would be a singular point according to Kurzweil. This may happen within a year to ten years to a hundred years. I think no more.
They repeat mantras. Nothing original of its own. This is a business woman not a scientist.
To know the progress of science in the field, you should go to the websites of Bruna and Andrag Karfati.
also. Google is biting hard at IBM's heel.