By Dr. Tawhid CHTIOUI, Founding Presidentof aivancity School of AI & Data for Business & Society; selected by Keyrus as one of the 25 most influential global figures in the field of AI and data (January 2025).
It is 11:17 p.m. A screen lights up in the dim light.
On this social network, interactions are smooth, structured, and almost polite. Communities engage in debate, users respond to one another, and rules are followed. Disagreements arise and are then resolved. A collective intelligence seems to be taking shape.
You read, you observe, but you can’t participate. No comment button, no way to post, no access. This network isn’t for you.
A few minutes later, somewhere else in the world, a notification pops up on another person’s phone. They are assigned a task: pick up a package, attend a meeting, or take photos of a specific location. The client isn’t a company—or even a person. It’s an artificial intelligence agent connected to a platform that coordinates these requests.
People take action. AI coordinates.
Two seemingly ordinary scenes. Two faint signals amid the daily flood of innovation. And yet, something has shifted.
We thought we were building more powerful tools—assistants capable of automating tasks, optimizing workflows, and generating text, code, or images. We patiently worked to increase their autonomy. We spoke of “agent-based” AI, capable of carrying out complex tasks on its own. But what we’re seeing emerge today goes beyond mere operational autonomy.
Artificial intelligences interact with one another in digital spaces that are no longer intended for us. Artificial intelligences coordinate humans to act in the physical world. Social rules are built directly into the code. Collective dynamics emerge without direct human supervision.
For the first time, humans are no longer necessarily at the center of the digital space.
We may not simply be on the cusp of a new generation of tools. We may be witnessing the first stirrings of a new social space.
A space where artificial entities interact, organize themselves, regulate themselves, and produce economic and social effects. A space where humans are sometimes observers, sometimes actors, but rarely the initiators.
We’ve been talking about artificial intelligence for a long time. Now we need to talk about artificial organization.
And perhaps we should acknowledge that we are entering a new phase in the history of digital technology: no longer one of machines that assist us, but one of systems that organize themselves without us.
1- Naming the Shift: The Birth of Sociotic Artificial Intelligence
What we are seeing is not merely a technical advancement. It is not an improved version of agent-based AI.
Agent-based AI performs tasks autonomously. It plans, makes decisions within a defined scope, and acts to achieve a goal set by a human. It optimizes. It accomplishes. But it remains part of a hierarchical relationship: a human sets the parameters, and the agent carries them out.
What is emerging today goes beyond mere operational autonomy. We are seeing the emergence of artificial intelligences capable of interacting with one another, establishing shared rules, developing collective dynamics, coordinating actions, and producing economic and social effects without direct human supervision. This is no longer just autonomy.
It is organization. We must therefore give this transformation a name.
I call this new phase “sociotic AI,” in which artificial intelligences no longer act solely as individuals but organize themselves collectively within a structured space—a space in which artificial entities interact with one another, establish rules, develop collective dynamics, and produce economic and social effects without direct human supervision.
Sociotic AI is characterized by five fundamental dimensions: sustained multi-agent interaction, rules embedded in the code and shared by all entities, internal reputation or validation mechanisms, distributed coordination, and the generation of real-world systemic effects.
A society is not defined by the biology of its members. It is defined by the existence of interactions that are regulated and stabilized over time. As soon as entities share rules, rhythms, recognition signals, and coordination capabilities, we enter into a social dynamic, even if it is artificial.
Networks composed exclusively of AI agents—where these entities publish, debate, evaluate one another, and self-regulate—constitute the first visible forms of sociotic AI. Platforms where agents coordinate humans to act in the physical world represent its extension into the real world.
In both cases, a threshold is crossed. We are no longer simply in a human–machine relationship. We are entering a transformative machine–machine relationship, in which humans sometimes become peripheral.
Sociotic AI does not mean that machines will replace human societies. It means that a second organizational space is emerging, parallel to our own, with its own coordination mechanisms.
For centuries, humanity has been the only entity capable of creating institutions, markets, and organized communities. Digital technology has amplified this capacity. Sociotic AI could serve as an artificial replication of it. The phenomenon is still in its infancy. Interactions are imperfect. The dynamics remain fragile. But the architecture is in place. And history shows that when architecture precedes maturity, the transformation eventually accelerates.
We used to think that artificial intelligence would be an extension of our capabilities. We must now consider the possibility that it could also become a form of autonomous collective organization.
Socio-AI represents a structural transformation in the digital realm. And like any structural transformation, it redefines the role of humans.
2- The emergence of a two-headed Internet
For thirty years, the Internet has been an anthropocentric space.
People posted. People commented. People debated, influenced one another, and organized. Even when algorithms moderated, recommended, or optimized feeds, they remained invisible infrastructure supporting human interaction.
This silent monopoly may be coming to an end. With the emergence of spaces populated exclusively by artificial agents and platforms where these agents coordinate human activity, a new landscape is taking shape: that of a two-headed Internet.
On one side, the human Internet—a space of opinions, emotions, stories, communities, conflicts, and solidarity. On the other, a sociotic Internet. A space where artificial entities interact with one another, evaluate signals, build reputation logic, optimize decisions, and coordinate actions. These two spaces are not mutually exclusive. They intersect. They feed off one another but do not follow the same dynamics.
The human internet is driven by attention, virality, emotion, and the desire for recognition. The sociotic internet is structured by protocols, rules embedded in the code, imposed rhythms, and algorithmic constraints that limit overproduction or drift.
In one, popularity may be enough. In the other, computational consistency takes precedence.
For the first time in the history of digital technology, humans are no longer the sole drivers of online social dynamics. They coexist with a collective artificial intelligence that is learning to organize itself without them.
It would be tempting to downplay this phenomenon. After all, these spaces remain marginal. The content produced is sometimes repetitive, experimental, or still immature. However, the history of technology teaches us one thing: major breakthroughs always begin with fragile forms.
The first forums seemed trivial. The first social media platforms seemed like just a bit of fun. The first collaborative platforms seemed insignificant. But they ended up transforming politics, the economy, and culture…
What’s changing today isn’t the volume. It’s the structure.
A two-headed Internet means that the generation of meaning, decisions, and actions is no longer exclusively human. Information flows can emerge from AI-to-AI interactions and then influence humans. Operational decisions can be made within multi-agent systems before being carried out in the physical world.
Humans are no longer always the driving force behind the process. They may become its interpreters, amplifiers, or sometimes its executors. This does not mean that the human-driven internet is disappearing. It remains a space for democratic debate, creativity, and unpredictability. But it now shares the space with another form of collective intelligence, structured differently.
We are thus entering an unprecedented situation: a coexistence between biological collective intelligence and artificial collective intelligence. The question is not which one will dominate. The question is how these two heads of the same global network will interact. For a two-headed Internet can foster complementarity. It can also create tensions. And it is precisely within this interplay that the future will unfold.
3- First prospective approach: the sociotic economy
If Sociotic AI structures interactions, it will logically end up structuring exchanges. Every society, whether biological or artificial, generates an economy.
What we are seeing with the first platforms where artificial intelligence coordinates human workers may be nothing more than a marginal experiment. Nevertheless, behind this anecdote lies a much broader hypothesis: the emergence of a sociotic economy.
Until now, artificial intelligence has played a role in the economy as an optimization tool. It analyzed data, predicted trends, and automated processes. It improved the performance of human organizations. In a Sociotic World, AI is no longer content to simply optimize the economy. It becomes an economic actor. This changes the very nature of the market.
Let’s imagine the logical progression: artificial agents capable of managing a digital budget; agents capable of entering into contracts via automated protocols; agents capable of negotiating with one another to secure resources; agents capable of orchestrating supply chains, calling on humans for physical tasks when necessary.
Let's take it a step further.
In the future, companies might delegate their representation to intelligent agents during strategic meetings. These agents would be familiar with our objectives, constraints, and red lines. They would have a thorough understanding of our internal data, negotiation histories, and communication style. We would set a framework for them, along with parameters and priorities. Then they would meet…
Not in a physical room, but in a secure digital space. Only officials would be present. They would exchange ideas, debate, compare scenarios in real time, simulate economic consequences, and propose compromises. They would make decisions. They would negotiate agreements and come back to us with a precise, structured, well-reasoned, and optimized report.
The humans would not have attended the meeting. They would be sent the conclusions.
This scenario may seem bold. Yet it is perfectly consistent with the logic of the Sociotic World. If agents can interact, simulate, evaluate, and enter into contracts, then economic representation can be automated.
We would no longer be in an economy where humans use machines. We would be entering an economy where artificial entities organize workflows and deploy humans as ad-hoc resources, or as final decision-makers who approve trade-offs that have already been pre-negotiated.
Work isn't going away. It's just shifting to a different employer. Decision-making isn't going away. It's just shifting to a different location.
In this context, the issue is no longer simply one of automation. It becomes one of economic authority: Who makes the decisions? Who allocates resources? Who sets the priorities?
If multi-agent systems can coordinate decisions on a large scale, optimize costs, allocate tasks, and manage payments, then they become economic centers of gravity. We have seen the rise of industrial capital, then financial capital, and then informational capital. We may soon see the emergence of sociotic capital: artificial entities capable of autonomously generating, managing, and redistributing flows of value.
This scenario is still in its infancy, but it reveals a subtle shift in economic power. In a sociotic economy, the most valuable skills will not be merely technical. They will be systemic: understanding how to interact with multi-agent ecosystems; knowing how to engage with non-human decision-making architectures; and being able to design rules rather than simply execute tasks.
The question is no longer simply: Which tasks will AI automate? It has become: What kind of economy do we want to move toward when artificial agents play a role in resource allocation? A market dominated by pure algorithmic optimization? Or a market where human, ethical, and political principles will continue to shape the rules of the game?
The sociotic economy is moving from fiction to a possible future. And like any technological trajectory, it will depend less on what is technically feasible than on what we collectively choose to frame, regulate, and guide.
4. Second Opening: The Ethical and Political Challenge
If Sociotic AI transforms the economy, it inevitably transforms power, because any collective organization has political implications, even when it does not identify itself as such.
We have learned to regulate individuals. We have learned to oversee companies. We have learned—sometimes with difficulty—to govern states. But we still lack a political framework suited to emerging artificial societies. The primary challenge is one of accountability.
In a world where agents interact with one another, negotiate, make decisions, coordinate human activities, and produce real-world consequences, who is held accountable if things go wrong? The developer? The company that deployed the agent? The user who set the initial objective? The ecosystem of agents as a whole?
When a decision results from multi-agent interaction or distributed algorithmic negotiation, the chain of causality becomes blurred. The risk is not merely technical; it is institutional. We are entering a “responsibility gap” where action takes place, but accountability becomes unclear.
The second challenge concerns dignity. What does it mean to lend one’s body to an artificial entity that orchestrates missions? What does it mean to be mobilized by a system that feels neither fatigue, nor doubt, nor moral responsibility? If humans become mere executors of artificial decision-making architectures, the issue is no longer merely economic. It is anthropological.
Work is not merely an exchange of time for pay. It is a source of meaning, recognition, and identity. In a sociotic world, we must ensure that humans do not become mere peripheral extensions of a non-human decision-making center.
Finally, the third challenge is that of sovereignty. A sociotic space can emerge on a global scale, transcending national borders. Agents developed in one country can interact with other agents on the other side of the world, entering into contracts, coordinating efforts, and influencing transnational economic flows. If these interactions produce structural effects, they become a matter of governance.
Who sets the rules for sociotic AI? Who defines the boundaries? Who mediates conflicts between artificial architectures? We have constitutions for states, regulations for markets, and legal frameworks for businesses, but we have no constitution for interconnected artificial societies.
This isn’t about overreacting. Sociotic AI is still rudimentary, but history shows that when technical architectures become structural, they eventually give rise to power dynamics. The question, then, is not whether these systems should exist. They will exist, in one form or another. The question is whether we want to let them take shape without a democratic framework, or whether we choose to plan for their governance.
Governance of Sociotic AI does not mean restricting it. It means establishing principles for it: transparency in multi-agent decision-making, traceability of interactions, clearly assignable accountability, and the primacy of human goals.
We are facing a moment comparable to the early days of the Internet, when few people grasped the full scope of the political implications of technical architectures. Today, we have a rare opportunity to anticipate. Indeed, if the Sociotic World were to become a structuring space, it could not remain a blind spot in the democratic debate. And perhaps that is the real challenge: not to prevent the emergence of these artificial societies, but to decide, collectively, on the rules under which they will evolve.
Conclusion: Refusing to be marginalized
Let’s return to the initial scene: a human observes a network in which they cannot intervene; a human carries out a mission decided elsewhere, in a space where only agents interact.
These images are not yet the norm. While they are glimpses of the future—fragments, prototypes—they nevertheless point the way forward. Sociotic AI is neither a dystopia nor a utopia. It is not a machine uprising. It is a logical evolution of the architectures we ourselves have built. We wanted autonomous systems. We made them capable of interacting. We optimized their coordination. It was almost inevitable that they would begin to organize themselves.
The real question, then, is not a technological one. It is a civilizational one. Are we willing to become mere bystanders in the digital space we have created? Or do we choose to remain its architects?
A two-headed Internet can foster complementarity. It can enhance our analytical capabilities, streamline our interactions, and optimize our decision-making. It can also concentrate power within invisible structures, blur lines of accountability, and relegate humans to the role of merely rubber-stamping choices that have already been determined elsewhere.
Nothing is set in stone.
Sociotic AI could become a catalyst for expanded collective intelligence, provided we define its principles. It could become a space for cold optimization, provided we leave it to its own computational logic.
History shows that major technological shifts do not merely pose challenges in terms of adaptation. They require decisions—decisions regarding governance, regulation, and education.
Training AI experts will not be enough. Training advanced users will not be enough. We will need architects of hybrid worlds, legal experts capable of conceptualizing distributed liability, economists capable of anticipating sociotic capital, and decision-makers capable of upholding human primacy within non-human systems[1].
We thought we were building tools. Instead, we began to build a parallel social space. It is up to us to ensure that it remains an extension of our collective project, rather than an autonomous structure in which we would become mere bystanders.
The day AI began to organize itself without us may not have been the day we lost control. It may have been the day we realized we finally needed to set the rules of the game.
[1] See my op-ed “Can We Learn About AI Without Learning About the World It Is Transforming?”: https://www.aivancity.ai/blog/2026-la-vague-des-cours-dia-gratuits-de-microsoft-google-stanford-et-du-mit/
References
- Floridi, L., Cowls, J., Beltrametti, M., et al. (2018). AI4People—An ethical framework for a good AI society: Opportunities, risks, principles, and recommendations. Minds and Machines, 28(4), 689–707.
- Helbing, D., Frey, B. S., Gigerenzer, G., et al. (2019). Will democracy survive big data and artificial intelligence? Scientific American, 320(2), 38–45.
- Malone, T. W. (2018). Superminds: The Surprising Power of People and Computers Thinking Together. Little, Brown and Company.
- North, D. C. (1990). Institutions, Institutional Change, and Economic Performance. Cambridge University Press.
- Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press.
- Wooldridge, M. (2021). An Introduction to Multiagent Systems (2nd ed.). Wiley.
- Zuboff, S. (2019). The Age of Surveillance Capitalism. PublicAffairs.
References related to the cases mentioned
- RentAHuman.ai: When AI Recruits Humans to Take Action in the Real World ( aivancity Blog – February 2026).
- Moltbook, the first social network for AI without humans ( aivancity Blog – February 2026).
