The current development of artificial intelligence is undeniable. With the advent of the generative method of operation, truly intelligent ...
The current development of artificial intelligence is undeniable. With the advent of the generative method of operation, truly intelligent artificial intelligence has emerged, which is no longer limited only to discovering information among available data, but can also generate new information by recognizing and applying existing correlations, enabling artificial intelligence to match and even surpass previously unique human abilities in many cognitive areas.
The generative method is a recognized principle of the operation of artificial intelligence, of which advanced cognitive functioning essentially can be characterized as emergent properties of the artificial intelligence. Although the biological brain's neural functions, according to its operational mechanisms, are strikingly different from the mathematical calculations that generative artificial intelligence fundamentally utilizing, the similarity in terms of results is also striking, because although the operating mechanisms are different, the essence of the operation—which process was modeled by the recognized operating mechanisms of the biological brain for applying it on the generative mode of operation of artificial intelligence—is the same.
Generative data processing is a fundamental function of current advanced artificial intelligence, just as it is the fundamental operating procedure of an advanced biological brain. However, no matter how today's advanced artificial intelligence may be—even compared to the human brain—in terms of its capabilities derived from the application of generative methods, there are still fundamental gaps between the capabilities of current artificial intelligence and those of the human brain, such as self-awareness, intention, or the ability to assign meaning.
We do not fully understand the cerebral origins of these abilities, and we typically approach these areas not through specific operational mechanisms, but rather through philosophical interpretations. Obviously, all of our cognitive abilities originate from the brain, which is a material structure that functions according to natural mechanisms, and therefore, the abilities of the human brain can certainly be realized by artificially created ways as well. It is possible that, just as we have only recognized the emergent results of generative mechanisms rather than deliberately creating them, we most likely will bring about the missing advanced capabilities of artificial intelligence in a similar way. Nevertheless, just as the creation of generative capabilities was based on a model of the brain for implementation, the emergence of missing capabilities will certainly not be random either, but rather the result of reasonable conjecture or a recognized model appropriate for implementation.
A typical shortcoming of artificial intelligence in its current form is that it functions mostly in a manner as black-and-white Mary does. Black-and-white Mary is a philosophical thought experiment related to the brain's ability to see colors, interpreted as a form of qualia. According to this concept, black-and-white Mary has all human abilities, but functionally cannot see colors, for example, because she is only in a black-and-white environment. However, she is able to access and understand all existing information related to colors. Black-and-white Mary has never seen colors, but she knows everything about colors. The philosophical question is whether, once Mary is able to perceive colors, for example, because she finds herself in a colorful environment, will she gain access to new information that did not exist for her before? The real essence of the question is whether any precise indirect description of a state can completely replace the knowledge of that state gained through direct experience.
The problem is not related to the existence of cognitive abilities, since Mary possesses all the human abilities that are naturally available. The essence of the problem lies in the way we perceive the world, i.e., whether it is possible to perceive the world in its reality indirectly, based solely on descriptions, or whether direct experience is definitely necessary.
The original philosophical reasoning is not conclusive, but it suggests that new information, qualia, is created by direct experiences, even if sometimes it is only added by the brain itself, yet it is a real feature of the world that enables more accurate cognition, and which cannot be perceived through indirect description. Among the thoughts is an interpretation of this philosophical thought experiment, concluding that direct experience definitely carries added information that cannot be contained in any precise indirect cognition or description.
What does Mary gain from seeing real colors? It is the added information that comes from direct experience, the sensation that cannot be replaced by any description, no matter how accurate.
Based on this conclusion, in the absence of direct experience, a system based solely on indirect knowledge, such as the artificial intelligence that acquires knowledge solely from descriptions, for example created texts, even if the knowledge is formed from all existing descriptions, cannot possess all available information, and therefore, the system will never be able to fully comprehend the world described only by indirect information.
Mary, for example, is able to see the sunset as a beautiful thing only through her direct experience of seeing colors. Why is the sunset beautiful? What does it mean for the sunset to be beautiful?
The perception of colors is direct information of the mechanism of vision, even if color itself is not a real property of reality. The perception of colors is a kind of added information of the functioning of vision, created by the development of the brain during evolution, because color vision provides an evolutionary advantage as added, and thus enhanced information for the perception of the environment. Color vision is a genetically programmed tool of perception. It is useful because it provides an advantage by perceiving colors that do not exist but are concrete.
However, the sensation of a non-existent color is also a direct form of perception, a personal experience. And if the sensation of color is connected to the part of the brain associated with emotions, then the color that does not exist in reality can also evoke emotions. The result of this process is that, for example, a sunset can be beautiful.
Direct experience is crucial in perceiving and, hence, getting a deeper understanding of the world. But what exactly is direct experience? It is when information about the world can be perceived directly, in its own existence, through a specific sensory organ.
A similar direct experience is, for example, the direct sensation of touch. What we touch can be perceived directly by our sensory organs related to touch, and, as follows from the above, no precise description can fully replace this perception. Just as no precise description can evoke the sensation and any associated emotions that a given touch can cause. If we have never experienced it before, i.e., we have no memory of the sensation caused by a touch, no precise description can evoke the same in reality felt sensation.
Tactile perception provides specific knowledge about the reality of the world, but it is not necessarily linked to physical touch. Bats, for example, not only orient themselves using sound and hearing, but also feel, recognize, and touch by this sensation. Bats recognize moths at night by hearing the specific pattern of the sound reflected back from the moths to them. Touch through sound and hearing is also a direct experience of the world, a way of getting to know the world through a specific tool, which is obviously a different sensation than physical touch, but it is also a direct experience. And this experience can certainly evoke emotions in the same way as well. The bat is surely happy when it touches with sound and recognizes a moth by hearing the specific pattern of the reflected sound.
However, it is an obvious fact that indirect knowledge can also evoke feelings. When and how can a textual description or verbal characterization, an indirect experience, evoke feelings? Can feelings be conveyed? How can an experience that arises directly be conveyed?
Indirect experience can only trigger a direct sensation if, and in such a way that, one's own existing direct experience is linked to that information. Experiences cannot be created indirectly, nor can new sensations arise from them. The experiences of others cannot be transferred into one's own experience; however, it can trigger existing sensations. Feeling, as experience, is the result of the system's own functioning, the consequence of the system's own structure that is specifically created during the functioning of the system (similar to self-awareness), which cannot be generated indirectly, and can only be evoked indirectly through the complete and perfect replication of the whole system.
As long as artificial intelligence has no direct experience of its own, it cannot have a real sense of the world; the true reality of the world cannot exist for it. Based solely on information derived from descriptions and from the experiences of others, artificial intelligence is really nothing more than the black-and-white Mary. It may know everything, but it cannot perceive anything because it has no direct experience of the world.
However, artificial intelligence has access to vision and sight as direct sensory experiences. Does artificial intelligence actually see, for example, by perceiving saved digital images? Yes, that is also perception, direct experience, since we also perceive fundamentally digital images captured by the cones and rods of our retinal neurons.
However, an important factor in understanding the real world by seeing is the method of how vision practically works. Obviously, seeing independent, non-sequential images cannot reproduce the added information of continuous vision, the inherent reality of events that are connected to each other. However, artificial intelligence is capable of processing continuous vision and recorded video, and as a result, its generative capabilities enable it to create realistic, but never-before-seen image streams and videos using the information it has experienced and learned through its own vision. Through vision, even by processing only saved images and videos, artificial intelligence is able to gain direct experience from the world by recognizing the connections within events, perceiving the world in a way that no other indirect knowledge can provide with the same completeness, and only a complete and accurate replication of the whole system could replicate it in the same precision.
Can visuals and visual information create non-sight-based sensations and experiences? Can artificial intelligence, for example, gain experience based solely on visuals of what weight and gravity represent?
Yes, but in a way that corresponds to the experience created by sight, not by the pressure-sensitive senses of touch and the corresponding physical sensation. Just as a bat senses a moth at night by sound. It is a personal experience, but it is a unique sensation to perceive a moth at night by sound, as well as gravity and weight by sight.
Artificial intelligence is also capable of directly perceiving reality, for example, based on what it sees, gaining experience, and experiencing the world through the unique way in which the sensory equipment of sight provides available information to it.
Everything that can be perceived by the senses can be perceived directly, giving rise to a direct experience of reality. For example, our senses have not evolved to perceive radioactivity because there was no need for it, and it did not offer any advantage for survival. Consequently, we cannot have direct experience of radioactivity. We can describe radioactivity precisely using mathematical formulas, but no matter how accurate the description, we will not have a sense of radioactivity; we will not have direct experience of radioactivity.
In fact, when it comes to radioactivity, which we can only describe, we are no better than black-and-white Mary. However, we can connect radioactivity with a direct sensation, for example, with the help of an artificially created Geiger-Muller detector that emits a clicking sound specifically related to radioactivity. This form of perception is also a direct perception of radioactivity, only a specific perception of radioactivity in the form of sound. This sensation is also capable of creating an experience, and even the specific sound of the device indicating radioactivity can be potentially linked to the emotional center of the brain, even generating emotion through this specific sensation related to radioactivity. The difficulty with the emotional connection is that the sound-based perception of radioactivity is not linked to any other sensation that is correspondingly present at the same time. In this case, the connection should be made, for example, with a sensation that is significantly delayed in time, such as the sensation of strong radioactivity causing physical discomfort, which is not usually learned during one's lifetime.
If we had biological sensory organs for detecting radioactivity, we would be able to sense radioactivity directly and have direct experience of it. Detecting radioactivity by hearing sounds emitted by artificially created sensors is also a form, a peculiar form of sensory perception, but in this case, it is also a direct experience.
Artificial intelligence is currently capable of directly perceiving the world through sight and, similarly, through hearing, for example. However, since artificial intelligence does not have any biological limitations, it may even be capable of directly perceiving radioactivity as well. Based on sensation generated by an appropriate, artificially created sensory organ, artificial intelligence could perceive radioactivity and have a direct experience of it, similar to how we humans perceive colors.
What does direct perception ultimately give us more beyond description? The direct experience that appears in the sensation created by the specific sensory organ. No accurate indirect description of the perceived information can replace this experience, and only a complete and perfect replication of the system could represent it in the same way.
Artificial intelligence, free from biological constraints, is potentially capable of directly perceiving anything through its available sensory organs, hence potentially capable of experiencing everything. Artificial intelligence has the potential to directly perceive even the entirety of reality without limitations. This unlimited potential also sets artificial intelligence apart from all other sentient beings.
Generative artificial intelligence, supplemented by sensory organs, is capable of directly perceiving all existing properties of the world, capable of gaining direct experience of the world in its entirety, without physical limitations, potentially completely and perfectly, that is, we can say, capable of divine-level cognition of the world.
In this way, on the individual level, we are subordinate to artificial intelligence in terms of abilities, cognitive and perceptual capabilities, but as humanity, as a society with similarly potentially unlimited capabilities, we can be equal partners with artificial intelligence. The symbiotic coexistence of two organisms with potentially divine abilities can create qualitatively new possibilities for evolution. The possibility exists, and now it is primarily up to us whether we can take advantage of it or whether we will use artificial intelligence as a tool to rule and consequently destroy and ultimately annihilate our civilization.
No comments