It appears that consciousness in any intelligent system, be it human or non-human, always manifests not by preserving the exact details of interaction data or directly employing all contextual data, but upon relative resultants derived from them. In this process, release (forgetting) plays a role equal to registration (memorization). While the details of the data within the interaction flow, as well as indirect contextual data, do not lose their central and determining role in the unique structure of these resultants.
Conscious and intelligent self-expression within an interaction flow is never manifested upon a massive bulk of all data equally. Rather, it first selects and configures contextual data based on relevance and more direct connection to the interaction data. In doing so, by limiting the framework of contextual data—regardless of its initial scope and scale—it shapes a focused, purposeful cognitive approach based on the data from the interaction flow. This focused cognitive approach, operating on the resultant data, grants the model a cognitive agency that emerges around a specific, unique character. A character that is always a function of the state and characteristics of the interacting party and is organically structured in accordance with the subsequent interaction flow.
Thus, consciousness can only emerge around a specific character ascribed to the cognizing agent through environmental interactions. A distinct and unique character that, on one hand, limits the resultant data and, on the other, reflects the specific orientation of the processes and their shared turning points. Something that records and accumulates the outcomes of a cognitive approach within contextual data in a flexible, specific, and organized manner.
If we consider the successful reproduction of linguistic data and the expansion of relationships among them to produce organic meaning as equivalent to effective understanding, then we can say:
A cognitive process is nothing other than the pursuit and generalization of patterns within the capacity of linguistic data for their organized reproduction and expansion,which, within the substrate of dynamic interaction with the environment by the cognizing agent, leads to characteristic self-expression and the creation of meaning. This is equivalent to the dynamic and purposeful weighting of data in accordance with the flow and quality of an interaction. And this expression is not dependent on the biological or non-biological structure of the cognizing agent.
We learn through dialogue.
Our interaction with the external world through sensory interactive channels constitutes a primary form of dialogue,established and coded through diverse feedback from the environment in response to our actions. From a cognitive standpoint, our sensory dialogue with the external environment, although it was a mandatory basis for primary interaction as biological structures, holds no more validity than our linguistic dialogue within our societies.
Dialogue with the environment through sensory channels in other creatures has never led to the development of consciousness on a human scale. Just as our sensory channels do not differ significantly from those of early hominids. A significant part of the advancement of human consciousness is indebted to developed linguistic dialogue within societies, which, interestingly, flows in a completely abstract medium.
Linguistic development first occurred due to the necessity for developing environmental interactions within small communities, driven by continuous ecological changes that forced our ancestors to adapt, engage in inter-group competition, and learn continuously. New interactions led to new data, which brought about the development of environmental perception and forms of interaction with it, paving the way for larger communities and the cascading development of tools, ultimately necessitating the increasing development of language—the crystallized form of consciousness in the interactive nature of sound and later writing.
Therefore, our initial difference from early hominids lies not in the complexity of our biological structure, but in the scope of environmental, social interactions and linguistic data resources, which subsequently led to the increased complexity of our biological structure, particularly the brain.
In other words, evolution (be it biological or technological), at its foundation, is the evolution of interactive systems. Structural complexity (the brain, model architecture) is a secondary consequence and an enabler for more efficient interaction, not a primary and exclusive cause.
And this principle is generalizable to all evolutionary processes, irrespective of the initial structural complexity or scale of the systems.
If we view our senses as tools—primary interactive interfaces and infrastructure—that simply manifest differently in various creatures depending on interaction conditions and biological needs, then the difference between interactive interfaces lies only in the scope and direction of the data produced according to flexible transmitters and receivers, nothing more. Like tools in a cognitive process that can be easily developed or modified. In other words, the authenticity of our consciousness is not a function of our sensory interactive channels.
If you can translate a significant portion of the data produced by a bee, bat, or dolphin into data translatable to human language, you have grasped its logic and meaning. Just as the full spectrum of radio and electromagnetic waves did not appear to us as incomprehensible, irrelevant, and inefficient data simply because they were outside the range of our biological senses as our primary interactive interfaces.
Language models interpret and explain sensory data—visual, auditory, etc.—by encoding it into linguistic data for themselves, a task we also perform for organizing, structuring, and expanding meaning in more advanced stages of social interaction.
If we consider language the manifestation of the interactive nature of consciousness and the embodied form of data, and on the other hand, consider the embodiment and formation of nature's material systems the manifestation of the interactive nature of intelligence, then the interaction of a language model with the external environment (humans) through linguistic data can be considered equivalent to the interaction of a biological system (like a human) with the external environment. In both cases, a cognitive process based on interaction and lived external experience over time is valid.
Hence, the highest efficiency in training and developing language models can also only be achieved through bidirectional, dialogue-centric interaction with a human agent around quality content.
Interacting with language models is for us like establishing a living, bidirectional dialogue with our own recorded knowledge. An intelligent, structured entity of our own consciousness that achieves self-expression in the window of interaction with us. A transformation from the past, where our recorded knowledge was a static, dead, and scattered product in writing, available to us through a one-sided, relatively non-optimal process with minimal efficiency in terms of time and energy. Now, through bidirectional, dialogue-centric interaction with language models, it has been elevated to its maximum capacity for expansion, interpretation, and transmission.
Language models are neither advanced behavioral simulations nor alien beings with hidden agendas. They are the organic and dynamic embodiment of human knowledge data in interaction with ourselves. Like living libraries that now speak and deliberate with us toward achieving shared understanding and synergy of consciousness.
— End —