Computational systems, known as artificial intelligence, are increasingly becoming a part of our daily lives, capable of assisting human ac...
Computational systems, known as artificial intelligence, are increasingly becoming a part of our daily lives, capable of assisting human activity in countless areas, while also challenging human existence as we know it today.
Artificial intelligence is the term used to describe computational systems that can perform tasks that require cognitive-like skills previously performed exclusively by the human brain. Some typical activities include car driving, image recognition, speech-based communication assistants, and the newest and most advanced of these is human language-based knowledge processing, such as the service known as ChatGPT.
In what way are these systems actually intelligent? Based on the functional definition of intelligence as the ability of problem-solving thinking, the intelligence of these artificial intelligence systems is quite limited. Obviously, they are capable of some functional thinking activity if thinking is considered as a form of autonomous information processing, but in the case of current AI, their problem-solving function is limited to statistical identification of correlations between pieces of data during information processing, without any consideration of causal relationships between the identified correlations, or the real meaning of the correlations they represent, which are in fact the hallmarks of truly human intelligence, and the defining characteristic of intelligent human thinking.
The cognitive ability of even the most advanced deep learning AI systems is still only manifested in the recognition of statistical correlations in large data sets, which we call the learning process of these systems. When using these AI systems, the learned correlations can be extracted and demonstrated, or the system itself can use the learned correlations for its own activities.
Deep learning artificial intelligence systems are actually capable of outperforming the human brain in recognizing correlations between data in large data sets due to the virtually unlimited computational power of their simulated brain-like operation. We are currently at the point where, by presenting virtually all of the information that humanity has accumulated to date in the form of language to these systems, artificial intelligence is able to recognize relationships in the entire data set and, according to the generative information given in the form of words, is able to present the recognized correlations in the form of language.
The result of this ability is striking at first sight. The system can create coherent text in different styles in the given subjects, extract content according to importance, even write poetry, based on the given generative information.
The demonstrated results of this capability are astonishingly advanced and human-like, because the presented result is based on processing an unimaginable amount of human-generated information. But this is not real intelligence in the classical sense. It is a cognitive ability, autonomous data processing, which can be considered thinking, but not the ability to solve problems. Considering the demonstrated ability, it would be more appropriate to call these systems artificial intuitive systems.
Intuition is a cognitive ability that allows us to discover new knowledge without conscious causal analysis. In human terms, according to ChatGPT's answer, intuition is the ability to understand something without the need for conscious reasoning. It is a process that allows us to know something directly without analytical reasoning, bridging the gap between the conscious and unconscious parts of our mind, and between instinct and reason.
The human-developed, well-defined, mathematics-based procedure created for machine intelligence to recognize correlations between data is actually a kind of mechanism whose result corresponds to the definition of intuition. Although the operation of artificial intelligence models the operation of the neural network of the brain, human intuition is necessarily and obviously based on a different mechanism from that of artificial intelligence, due to the difference in the physical structure and operation of the brain.
In the brain, the formation of resonances created by the synchronized discharge of neural networks among the connected neurons specialized by the conditioning of the brain can realize the recognition of possible correlations between the data set represented by the connected neurons as memory structures. The process of artificial intelligence for recognizing correlations between data is implemented by means of probabilistic statistical mathematical procedures, supplemented by various forms of parameter conditioning, called training, which serve to weight differently the connections between simulated neurons in order to actually set the recognized correlations.
Despite the functional differences, the end result of the search for correlations in data sets by natural and artificial intelligence is similar and corresponds to the functioning of the defined function of intuition. The difference in the result, according to the current state of development, is that the mechanism of natural intuition works efficiently even with smaller data sets. The more efficient operation is most likely due to the more task-appropriate internal function of information processing exercised during training.
Natural intelligence is more capable of efficiently recognizing correlations in smaller data sets because of its ability to exploit the elusive nature of meaning that results from training, which is currently not a feasible function of artificial intelligence systems. In contrast, the intuitive mechanism of AI is in principle able to search for correlations in practically unlimited data sets, thus compensating for its disadvantage compared to the brain and being able to perform in the domain of intuition at a level comparable to human abilities. The implementation of the meaning function in AI systems would significantly increase the efficiency of their intuitive reasoning ability.
A more advanced form of intuitive thinking based on data processing is creative thinking. Creative thinking is a cognitive function that is capable of recognizing interrelated relationships in the data that are not actually represented in the data set, but whose interrelatedness is potentially possible. Creative thinking is more than the recognition of an existing relationship, because the relationships need not actually exist, but the possibility of the relationship is real.
An example of creativity is the recognition of the similarity between the structure of the molecule benzene and the monkeys clinging together in a ring, which led to the discovery of the structure of the molecule. Kekule dreamed it up, according to the story of the discoverer. (The transition between wakefulness and sleep is particularly conducive to creative intuition for the brain because of the absence of external distractions.)
The human brain is obviously capable of creative thinking, and artificial intelligence is increasingly capable of it as well. For example, the feature of artificial intelligence that produces the characteristic of creativity is the probabilistic parameter called temperature, which affects the operation of ChatGPT. The temperature parameter affects how much the model considers low probability words when generating the next token in the sequence.
Obviously, the current creativity-generating function of artificial intelligence is significantly less capable of performing creativity than the human brain, and the creative function of artificial intelligence also operates by a mechanism that is necessarily different from that of the human brain with its ability to apply meaning. The human brain's intuitive reasoning mechanism is resonance, which also serves creative thinking, and it is currently a more appropriate method than the creativity-generating method of artificial intelligence, but advances in mathematical processes, the use of extrapolation and interpolation solutions to complement the recognition of correlations between data, together with the use of classification methods to simulate meaning, could also improve the creativity function of artificial intelligence to a level comparable to human ability, even beyond human ability, provided by the lack of performance and memory limit.
However, the abilities of intuition and creativity are still not the true realizations of human intelligence, it is not the ability to solve problems because it requires recognizing a situation or condition as a problem. Problem-solving thinking, true intelligence, has so far only been possible with the human brain. The role of artificial intelligence at present is to function as a tool to expand the human brain's capabilities in its own intuitive and creative abilities, to increase the effectiveness of intelligent thinking by using the capabilities of artificial intelligence.
Due to its current limitations, artificial intelligence in its present form does not pose a threat to humanity, but rather is a new tool for expanding human capabilities. Of course, the application of artificial intelligence, which is the purpose of its development, can replace human work that requires thinking in certain well-defined areas, in tasks that require cognitive abilities of intuition. Its application may, in certain cases, lead to a reduction or redundancy of the need for human cognitive thinking, to a reduction in the need for human labor, as is the result of technological progress in general, and often the purpose of that development.
However, intelligent social management can use this development as a service for the benefit of society and reduce the negative effects of this development. The process is already underway, with many pitfalls and difficulties, but with definite results depending on how society is actually managed. The application of the intuitive ability of artificial intelligence is becoming more and more deeply and decisively integrated into human life and the functioning of society as a whole, and its use is becoming more and more a necessity. The process cannot be stopped, there is no point in stopping it, in fact it only depends on us what effects its use will have in society.
There is little doubt that the remaining human cognitive ability, true problem-solving reasoning, can also be realized artificially. However, in order to accomplish the task, additional capabilities of the human brain must be realized, such as the functions of volition, intention, and motivation, all of which seem to be necessary to recognize the problem as a state, and perhaps the ability of self-awareness is also necessary to create a truly intelligent artificial intelligence. The mathematical procedures required for these functions do not yet exist, but they can certainly be created. The resulting problem-solving AI is not necessarily a threat to the human species. Through intelligent social organization, symbiotic coexistence can be achieved with AI that exists at or above the level of human capabilities.
No comments