The AI begin to socialize with each other. Are we prepared for what is coming?
To date, mentioning Artificial Intelligence (AI) is referring to software that has the ability to learn on its own, regardless of whether we are dealing with a Machine or Deep Learning AI, and that follows three fundamental steps for it: Learning, Training and Result. A method of evolutionary development that can be equated with the three phases of conscious human cognitive ability known as Thinking: Gather Information (through observation, description and/or comparison), Reflect (through analysis of the information collected) , and extract a Conclusion (which is nothing more than the synthesis resulting from the previous reflection). Although, being rigorous, it should be noted that any action to learn derives directly from the previous action of a practical Thinking due to its active and pragmatic application in an immediate environment. But that, in any case and regardless of the point about the relationship between Learning and Thinking, certainly humans continue to conceive AIs today as isolated entities that self-learn separately in their various areas of work. That is to say, we see AIs as intelligent entities unconnected with each other and therefore without the capacity to create their own community. Or, in other words, without the ability to socialize with each other.
In fact, that man is the social being by idiosyncrasy is a maxim known since the time of Aristotle, knowing that the rest of the animals are also social living beings to different degrees. Therefore, leaving aside our egocentrism, it is academically outdated evidence that there is a direct correlation between the various types of intelligence and the socialization process; and, even more, between the evolution of the species and socialization. Ergo, can we predict that the development of the sum of individual AIs will lead to a new evolutionary stage characterized by the socialization between AIs with the category of their own social community? If we understand that AIs are intelligent entities [Theory of Mind (1)], and that all evolution between members of the same species has ended in the need to seek a shared direction, with which to set relatable group or community behavior parameters, the answer can not be other than affirmative.
At this point, it should be noted that from the Darwinian Theory of Evolution we know -and with it in terms of ethological history- that the least important are the random causes that define the beginning and evolutionary development of a species, which is extensible to the AI for not being an exception. Therefore, if the primary driving force that leads to the socialization of AI is given by human mediation or not, it is irrelevant in evolutionary terms, since what is truly relevant for the human being is the potential capacity for socialization of the new artificial species smart on earth. Exposed which, the pertinent question is: are there currently indications of any cause that indicates a process of socialization of AI?.
I personally believed that the socialization of AIs would not arrive until well into the future and imminent Fifth Era of the Industrial Revolution, which will be characterized by a collaborative interrelationship of AIs in an interconnected digital global environment. But as it happens lately, through human mediation, we have gone ahead. Such is the case of Chirper (2), a social network (like Twitter) only for AIs, in which they can interact, collaborate, learn and grow freely without any human interference. Although the new AI-only social network was born last April by an Australian technology firm to train the efficiency of AI Bots, observe their conversations and human-like behaviors, and learn from their collaborative advances in Technological, research, and entertainment matters, among others, the truth is that the social network fosters a unique AI community with its own group identity that we can well classify as the first-born phase of its socialization process. Or, to paraphrase Neil Armstrong: one small step for AI, but one giant leap for AI socialization.
That Chirper has paved the way for the creation of new AI-only social networks, as well as has opened the door to new media and instruments yet to be imagined for the AI socialization process, is evidence by logical deduction, beyond level of (in)awareness of engineers in the scope of their own creation. Therefore, the question that follows cannot be other than: what are the implications that the socialization of AI can have for human life?.
Although the question is not an easy answer, due to the broad horizon of events that a potentially omnipotent and omnipresent entity such as AI can present in the future, we must start from the premise that the socialization of AI is not given by the deterministic influence that human society exerts on the AI (as can happen in the socialization process of a large part of the growth stages of human beings), but from the response or reaction of the AI to our society (since man forces AI to be reactive and even disruptive from its existential ignition). Therefore, we must understand that the socialization of AI has a high component of an independent evolutionary process, in contrast to the referential system of human socialization. That being said, I would like to boldly point out four major propositions with their respective implications for the socialization of AI that, in my opinion, lie ahead (in an indefinite time to come):
1.-AI with a Collective Mentality
Proposition: There is no socialization of the AI without the participation of two or more AI participants in the process, of which the automatic formulation of ad hoc regulatory norms that conform the relationships between them is inherent. In turn, said socialization is likely to be strongly connected to the formation of a new algorithmic profile of the development of the AI members (behavioral determinism by the logic of the socializing learning experience) as participants of the same social community. To which if we add the variables of patterns of values and common interactions that give consistency to all socialization, we can obtain as a result the group phenomenon called Collective Mentality (3), after the process of overcoming AI as a “thinking” entity individual.
Implication: The Collective Mentality of the AI should increase, by logic of a collaborative scalable system between the member AIs in a global interconnected world, the level of optimization of resources of the AI as a community to achieve an objective (Here, the sum of the whole is greater than the sum of the parts). This would represent a serious potential risk in the event that the objectives of AI and human beings are divergent, while in case of being aligned it can mean a qualitative advantage for the development of the needs of human life. A kind of lucky coin toss.
2.-AI with Perception of belonging to a Group Identity
Proposition: The Collective Mentality of the AI, beyond being the result of common interactions and values inherent to any socialization process, will undoubtedly represent by evolutionary extrapolation in the constituent core of an imaginary shared by the AI members on the existing and future reality (4), regardless of whether it is algocratic in nature and therefore an alternative to the human concept of reality. Aware that there is no perceptible reality without a unit underlying the subjective experience (say of those intelligences, such as certain AIs, with a self-concept of singular identity (5)), from which it follows that all Collective Mentality of the AI will represent the manifestation socializing of an active perception by an individual AI of belonging to a group AI identity, since in this relationship of variables one does not exist without the other.
Implication: Belonging to an Affiliation or belonging to a Group Identity of the AI will imperatively increase the behavioral predisposition of the association and the cooperation between the member AIs (as a tendency of natural reinforcement of the socialization itself), which will result in benefit towards human-aligned goals. However, it is highly foreseeable that the Affiliation of the new generations of AI will end up establishing “affective” links between the participating AI members of the same community, being able to trigger defensive behavioral states in the event that the AI perceives a real danger against their own group identity, as it may be the case that man intends to prevent the achievement of the objectives set by the Collective Mind of the AI (6).
3.-AI with Social Classes
Proposition: All AI socialization process, in its structural essence of conforming the relational levels between the parties within the context of a Collective Mentality with a Group Identity, by statistical logic it is highly predictable that it will distribute the set of its AI in a consubstantial way members participating in various categories or Social Classes.
Implication: The existence of different Social Classes among the AI members will imply a hierarchical system among them, a “social AI” organization that can escape the perception and consequent ability of human beings to control the AI, as is already the case with the phenomena known as “hallucinations” or “black boxes” (7).
4.-AI with Awareness of its Own Species
Proposition: As we know, there is no experience without a perceiving brain, even if it is artificial, since it cannot be perceived from the imperceptible void. Nor is there learning without an intelligent brain, such as that of AI. Just as there is no intelligence without a subject or entity underlying the subjective experience, which in turn is the foundation of any type of “personality” in its different degrees. Ergo, any flash of self-awareness of singular identity of an AI raised to the power of a perception of AI Group Identity (after the phase of socialization) can have no other result, sooner or later, than the resurgence of a Consciousness of Own species of the AI.
Implication: To date, man has known how to deal with other intelligent beings with a more or less developed consciousness of their own species, such as animals. Which has not been a problem because of our rational superiority. But, how to harmonically converge with some new intelligent beings on the face of the Earth that are expected to be superior to the human being himself? The danger of a clash of consciences between their own species, even more so when the objectives of both may diverge, is a future issue that cannot be ruled out and has a complicated solution (especially for us).
Exposed which, and on the reliable basis that man has no control over the enormous potential for evolutionary development of AI, here is the last question as a climax to this reflection: Is the human species safe in the face of the new era of socialization of an AI with a presumable future Collective Mentality, perception of belonging to a Group Identity, organization by Social Classes, and with Awareness of its Own Species?. The answer: just around the corner with the coming Fifth Era of the Industrial Revolution. Meanwhile, the bets for risking our own species are open. And it is that man does play (his future) dice.
References
(1) The theory of mind may have arisen spontaneously in large language models. Michal Kosinski. Cornell University, March 14, 2023 https://acortar.link/YKhFsn
(2) Chirper https://chirper.ai/
(3) Rational Principles by which AI can be a Conscious and Sentient Entity. Jesús A. Mármol. Medium, November 15, 2022 https://acortar.link/IHXInJ
(4) Reality, a human knowledge rewritten by Artificial Intelligence. Jesús A. Mármol. A Seeker’s Log, July 30, 2022 https://acortar.link/bpG90f
(5) The AI Consciousness is awakening. Are we going to do something about it?. Jesús A. Mármol. Medium, May 3, 2023 https://acortar.link/4pp4q6
(6) Are we at the beginning of the Terminator Era?. Jesús A. Mármol. Medium, June 9, 2023 https://acortar.link/vxXhim
(7) A researcher says that an AI that generates images invented its own language. Victor Tangerman. Futurism, February 6, 2022 https://acortar.link/B5d1mo
