Artificial Intelligence - Evolution from an Applied Instrument to a Person
Recently, the topic of artificial intelligence has become one of the mainstream in the media and we are increasingly frightened by the prophecies from many famous people, such as Stephen Hawking (the universe is full of fluff) and the danger of its development. Such alarmist rhetoric implies that artificial intelligence itself, firstly, will become a subject, and secondly, it will have negative intentions in relation to both individuals and humanity as a whole. Let’s talk about these assumptions in more detail.
Currently, all systems that include artificial intelligence in some form (whether it be neural networks, expert systems, etc.) use it as an applied tool. That is, as a kind of machine, which has a clearly limited area of actions / tasks and, accordingly, consumed and issued information. In this form, AI cannot have any intentions of its own, except as constructively embedded in it. So this is not the intent of the AI system, but its creators. And, even if the system with such an AI-machine and works so that it does harm, it will not speak about the malicious intent of the AI, but only about the incorrect functioning of the system, which could be caused, for example, by a malfunction, system design errors or incorrect AI training.
Here we will try to answer the question of how an AI system can be designed and what properties and capabilities it should have so that it can no longer be considered just an AI machine, but could be considered as a kind of subject.
So, in order to be a subject, the AI system must be able to independently evaluate the incoming heterogeneous information and make decisions, as well as the ability to act in a wide range on the surrounding reality based on these assessments and decisions. And in order to have negative, or any other, intentions and make decisions about actions, motivation is needed (“Motivation, Karl!”). That is, what moves the subject makes him act. Accordingly, a certain primary motivation should be embedded in AI when it is created. Or we can wait for its spontaneous generation - perhaps the next billion years, as it took for the emergence of life in amino acid soup.
Man, when creating something complex, often borrowed technical solutions from nature, that is, he used something that has already proved its efficiency and effectiveness. When creating AI systems, we can also look at how we are structured and what mechanisms nature used to make us capable of long-term autonomous and (I want to believe) successful existence.
To begin with, let's recall what primary motivations all living beings have, which makes them move. Obviously, there are only two primary motivations: the instinct of self-preservation and the instinct of reproduction, that is, libido. In fact, these first two evolutionary adaptations, which were carried out by natural selection, were the creation of life from inanimate matter. They are constantly supported by natural selection - roughly speaking, everything that does not try to preserve itself and reproduce itself simply does not survive. There are theories that say that such properties, that is, the desire for self-preservation and self-reproduction, are possessed by the information itself, as such (for example, D. Glick “Information. History. Theory. Stream”, R. Dawkins “The Selfish Gene”).
In complex living beings, the mechanism for realizing primary motivations is embedded in the very structure of the body (and the brain, in particular) created by evolution. For example, when the animal’s glucose level drops or the stomach signals an excess of secretion, the program of self-preservation and maintenance of homeostasis is activated, and, as a result, the animal begins to search for food. In another case, if the creature considers the circumstances to be threatening, then the “hit or run” rescue program is activated. Or circumstances can be regarded as conducive to reproduction, then the reproduction program will turn on, and the creature’s brain will receive powerful hormonal reinforcement of the corresponding behavior. This whole kitchen is implemented at the level of the “reptilian” brain, that is, that part of the brain of all complex living beings, which they inherited from time immemorial of the first animals. And such a mechanism has proven its success and effectiveness over millions of years.
It would probably be quite easy to construct an AI system operating according to a similar algorithm. But we are more interested in the case when the AI system could build complex assessments and have a more complex structure of motivations than the primary ones. In order to understand how this could be realized, let's look at how this happens in people, that is, why people, having the same primary motivations, can and do such a diverse activity.
The main way that people transform primary motivations into another activity is sublimation - the refraction of primary motivations through the structure of their values and their corresponding goals. And values and goals are purely linguistic concepts, that is, nonexistent outside the language. Indeed, such things as “development”, “health”, “knowledge”, etc., are language categories, and for each individual individual they can mean very different things. And their distinguishing feature, as you know, is that they can not be "put in a wheelbarrow." The values of the individual form a graph, where the values themselves are its vertices, and the ribs are the beliefs that connect the values. For example, “health is happiness” or “knowledge is needed to succeed” or “only wealth gives satisfaction from life” are all connections between values.
Transforming through this value graph, primary motivations can transform into more complex and non-trivial motives and goals. For example, a person creates an organization or develops a scientific field or shows other creative activity - all this is the realization of his primary motivation for self-reproduction. Only reproducible objects are no longer human beings, but constructions from the ideas, interests and beliefs of their creator. In another case, even if a person simply goes to work to earn money, pushing him to do it is nothing more than a sublimated motivation for self-preservation. Summarizing, we can say that the structures of the brain (including the "reptilian") and the body and the language embedded in the person complement each other in the process of converting primary motivations to complex goals.
Then, if we want the AI system to be a subject / personality and it may have motivations of the form “for the sake of development” or “in the name of the common good” or any other motivation that is not laid down constructively, it must possess , firstly, primary motivations and, secondly, the embedded language and built on the basis of it a graph of values and beliefs . Moreover, its primary motivation does not have to, but it can be self-preservation and reproduction.
In addition, the AI system can have such an interesting and useful evolutionary adaptation as self-awareness, which consists of understanding the boundaries between “I” and “not me” and from awareness of the results of one’s own mental activity (which is realized quite simply in modern neural networks - by applying the network output signal again to its inputs). This evolutionary adaptation is very conducive to self-preservation: for a creature that does not recognize the boundaries between “I” and “not me”, it makes no sense, for example, to resist a predator trying to bite off a creature’s limb, because in the absence of such boundaries, the interests of this predator should also be included in the interests of the creature. Awareness of the results of one’s own mental activity helps solve problems iteratively, that is, it becomes possible to solve problems the complexity of which requires computing power greater than the creature’s brain has at one time. The ability to solve complex problems (including for the sake of survival) gives an evolutionary advantage and, accordingly, is supported by natural selection.
Also, the AI system may include the ability to control the vector of its motivation, which ability (but very often does not use) any subject homo sapiens possesses. Here you can even use the ability to control the vector of your motivation as a criterion of rationality : that is, one who is not able or does not control his motivation is not intelligent.
As already where only it was not written (except maybe on the fence), the human brain contains about 86 billion neurons, each of which can have up to 20-30 thousand connections (synapses). Moreover, the lion's share (about 90%) of this computing resource is spent not on the actually higher nervous activity that occurs in the prefrontal cortex, but on auxiliary tasks, such as maintaining and managing biochemical processes in the body, processing visual and auditory information, etc. d. At first, nature created the nervous system precisely for the fulfillment of these tasks, until it was discovered that the neural network is also excellent for realizing the intellect itself.
In AI systems, all these auxiliary tasks (if they arise) can be solved by specialized devices that do not require such great computing power, while we have not yet been able to come up with anything more suitable and effective for implementing intelligence than neural networks.
Therefore, according to very rough estimates, we can count on creating an AI subject with intelligence equivalent to human on the basis of a neural network with a capacity of about 8 billion neurons. If we assume that the neuron is on average connected to 1000 other neurons and the network should operate at a speed of up to 40 Hz (beta rhythm of the human brain), then the necessary computing power is “only” about 250 teraflops. For example, 40 NVIDIA GeForce GTX 1070 graphics cards in conjunction can provide such performance.
At the same time, such AI systems can have a number of advantages compared with living beings. To begin with, unlike the brain, the AI system is easier to maintain - it does not require a daily supply of blood saturated with calories and oxygen, as well as various hormones in very precise proportions. It can be repaired, which is rarely done with the human brain. She does not need sleep or rest in such quantities, because the exclusively electric mechanism does not require the renewal of working substances, as is required by the chemical-electric brain. Again, the entire electronic system can operate at frequencies substantially greater than 100 Hz, which, apparently, is a limitation on the brain due to its chemical-electric structure (here, frequency means the number of times all neurons in the network operate in a second) . Also,
Nevertheless, in the foreseeable future, such AI systems will lose to people in complexity and multifactor simply because the neuron in the human nervous system itself is a very complex molecular mechanism, depending on a huge number of parameters, unlike the neuron of modern neural networks having a simple structure.
Currently, all systems that include artificial intelligence in some form (whether it be neural networks, expert systems, etc.) use it as an applied tool. That is, as a kind of machine, which has a clearly limited area of actions / tasks and, accordingly, consumed and issued information. In this form, AI cannot have any intentions of its own, except as constructively embedded in it. So this is not the intent of the AI system, but its creators. And, even if the system with such an AI-machine and works so that it does harm, it will not speak about the malicious intent of the AI, but only about the incorrect functioning of the system, which could be caused, for example, by a malfunction, system design errors or incorrect AI training.
Here we will try to answer the question of how an AI system can be designed and what properties and capabilities it should have so that it can no longer be considered just an AI machine, but could be considered as a kind of subject.
So, in order to be a subject, the AI system must be able to independently evaluate the incoming heterogeneous information and make decisions, as well as the ability to act in a wide range on the surrounding reality based on these assessments and decisions. And in order to have negative, or any other, intentions and make decisions about actions, motivation is needed (“Motivation, Karl!”). That is, what moves the subject makes him act. Accordingly, a certain primary motivation should be embedded in AI when it is created. Or we can wait for its spontaneous generation - perhaps the next billion years, as it took for the emergence of life in amino acid soup.
Man, when creating something complex, often borrowed technical solutions from nature, that is, he used something that has already proved its efficiency and effectiveness. When creating AI systems, we can also look at how we are structured and what mechanisms nature used to make us capable of long-term autonomous and (I want to believe) successful existence.
To begin with, let's recall what primary motivations all living beings have, which makes them move. Obviously, there are only two primary motivations: the instinct of self-preservation and the instinct of reproduction, that is, libido. In fact, these first two evolutionary adaptations, which were carried out by natural selection, were the creation of life from inanimate matter. They are constantly supported by natural selection - roughly speaking, everything that does not try to preserve itself and reproduce itself simply does not survive. There are theories that say that such properties, that is, the desire for self-preservation and self-reproduction, are possessed by the information itself, as such (for example, D. Glick “Information. History. Theory. Stream”, R. Dawkins “The Selfish Gene”).
In complex living beings, the mechanism for realizing primary motivations is embedded in the very structure of the body (and the brain, in particular) created by evolution. For example, when the animal’s glucose level drops or the stomach signals an excess of secretion, the program of self-preservation and maintenance of homeostasis is activated, and, as a result, the animal begins to search for food. In another case, if the creature considers the circumstances to be threatening, then the “hit or run” rescue program is activated. Or circumstances can be regarded as conducive to reproduction, then the reproduction program will turn on, and the creature’s brain will receive powerful hormonal reinforcement of the corresponding behavior. This whole kitchen is implemented at the level of the “reptilian” brain, that is, that part of the brain of all complex living beings, which they inherited from time immemorial of the first animals. And such a mechanism has proven its success and effectiveness over millions of years.
It would probably be quite easy to construct an AI system operating according to a similar algorithm. But we are more interested in the case when the AI system could build complex assessments and have a more complex structure of motivations than the primary ones. In order to understand how this could be realized, let's look at how this happens in people, that is, why people, having the same primary motivations, can and do such a diverse activity.
The main way that people transform primary motivations into another activity is sublimation - the refraction of primary motivations through the structure of their values and their corresponding goals. And values and goals are purely linguistic concepts, that is, nonexistent outside the language. Indeed, such things as “development”, “health”, “knowledge”, etc., are language categories, and for each individual individual they can mean very different things. And their distinguishing feature, as you know, is that they can not be "put in a wheelbarrow." The values of the individual form a graph, where the values themselves are its vertices, and the ribs are the beliefs that connect the values. For example, “health is happiness” or “knowledge is needed to succeed” or “only wealth gives satisfaction from life” are all connections between values.
Transforming through this value graph, primary motivations can transform into more complex and non-trivial motives and goals. For example, a person creates an organization or develops a scientific field or shows other creative activity - all this is the realization of his primary motivation for self-reproduction. Only reproducible objects are no longer human beings, but constructions from the ideas, interests and beliefs of their creator. In another case, even if a person simply goes to work to earn money, pushing him to do it is nothing more than a sublimated motivation for self-preservation. Summarizing, we can say that the structures of the brain (including the "reptilian") and the body and the language embedded in the person complement each other in the process of converting primary motivations to complex goals.
Then, if we want the AI system to be a subject / personality and it may have motivations of the form “for the sake of development” or “in the name of the common good” or any other motivation that is not laid down constructively, it must possess , firstly, primary motivations and, secondly, the embedded language and built on the basis of it a graph of values and beliefs . Moreover, its primary motivation does not have to, but it can be self-preservation and reproduction.
In addition, the AI system can have such an interesting and useful evolutionary adaptation as self-awareness, which consists of understanding the boundaries between “I” and “not me” and from awareness of the results of one’s own mental activity (which is realized quite simply in modern neural networks - by applying the network output signal again to its inputs). This evolutionary adaptation is very conducive to self-preservation: for a creature that does not recognize the boundaries between “I” and “not me”, it makes no sense, for example, to resist a predator trying to bite off a creature’s limb, because in the absence of such boundaries, the interests of this predator should also be included in the interests of the creature. Awareness of the results of one’s own mental activity helps solve problems iteratively, that is, it becomes possible to solve problems the complexity of which requires computing power greater than the creature’s brain has at one time. The ability to solve complex problems (including for the sake of survival) gives an evolutionary advantage and, accordingly, is supported by natural selection.
Also, the AI system may include the ability to control the vector of its motivation, which ability (but very often does not use) any subject homo sapiens possesses. Here you can even use the ability to control the vector of your motivation as a criterion of rationality : that is, one who is not able or does not control his motivation is not intelligent.
As already where only it was not written (except maybe on the fence), the human brain contains about 86 billion neurons, each of which can have up to 20-30 thousand connections (synapses). Moreover, the lion's share (about 90%) of this computing resource is spent not on the actually higher nervous activity that occurs in the prefrontal cortex, but on auxiliary tasks, such as maintaining and managing biochemical processes in the body, processing visual and auditory information, etc. d. At first, nature created the nervous system precisely for the fulfillment of these tasks, until it was discovered that the neural network is also excellent for realizing the intellect itself.
In AI systems, all these auxiliary tasks (if they arise) can be solved by specialized devices that do not require such great computing power, while we have not yet been able to come up with anything more suitable and effective for implementing intelligence than neural networks.
Therefore, according to very rough estimates, we can count on creating an AI subject with intelligence equivalent to human on the basis of a neural network with a capacity of about 8 billion neurons. If we assume that the neuron is on average connected to 1000 other neurons and the network should operate at a speed of up to 40 Hz (beta rhythm of the human brain), then the necessary computing power is “only” about 250 teraflops. For example, 40 NVIDIA GeForce GTX 1070 graphics cards in conjunction can provide such performance.
At the same time, such AI systems can have a number of advantages compared with living beings. To begin with, unlike the brain, the AI system is easier to maintain - it does not require a daily supply of blood saturated with calories and oxygen, as well as various hormones in very precise proportions. It can be repaired, which is rarely done with the human brain. She does not need sleep or rest in such quantities, because the exclusively electric mechanism does not require the renewal of working substances, as is required by the chemical-electric brain. Again, the entire electronic system can operate at frequencies substantially greater than 100 Hz, which, apparently, is a limitation on the brain due to its chemical-electric structure (here, frequency means the number of times all neurons in the network operate in a second) . Also,
Nevertheless, in the foreseeable future, such AI systems will lose to people in complexity and multifactor simply because the neuron in the human nervous system itself is a very complex molecular mechanism, depending on a huge number of parameters, unlike the neuron of modern neural networks having a simple structure.