The evolutionary instinct that made human cooperation possible may become the greatest obstacle to civilization’s survival. A civilization can become extraordinarily sophisticated without becoming proportionately wise.

We have developed mathematics, science, international institutions, and technologies capable of connecting billions of people. We can understand processes occurring across galaxies, manipulate matter at increasingly fundamental levels, and engineer systems that may eventually surpass our own intelligence.

Yet one of the oldest limitations of the human mind remains remarkably persistent: We still divide the world into us and them.

We trust people who belong to our group more readily than those who do not. We interpret the actions of our allies more generously than the actions of our opponents. We defend the mistakes of our side while exaggerating the mistakes of others. We can recognize the humanity of billions of strangers in principle while behaving as though their interests matter less when they conflict with our own.

This is tribalism.

It is not simply the existence of cultural identities, communities, nations, religions, or shared traditions. Those things can provide meaning, belonging, and the foundations of cooperation for species that had to evolve in Savannah.

The problem begins when group membership determines whose interests deserve consideration, whose suffering can be ignored, and whose existence becomes negotiable. For much of our evolutionary history, this tendency may have been compatible with the conditions in which human beings lived.

Today, it is increasingly incompatible with the civilization we have created. And the emergence of superintelligent mind could expose this contradiction at a scale humanity has never experienced. The greatest danger may not be that superintelligence develops our tribal instincts independently. It may be that we transfer those instincts into the most powerful intelligence we have ever created.

1. The evolutionary bargain

For hundreds of thousands of years, human beings lived primarily in small-scale societies dependent on hunting, gathering, fishing, reciprocal exchange, and cooperation. Their social environments were fundamentally different from ours. People generally lived in relatively small communities, maintained extensive relationships with other groups, and depended on one another for food, protection, childcare, knowledge, and survival.

Some ethnographic research estimates that mobile hunter-gatherer camps averaged approximately twenty people, although social organization varied considerably. Larger networks connected individuals across multiple camps and communities. Dunbar’s number, commonly expressed as approximately 150, is one influential attempt to describe the scale of stable social relationships. It should not, however, be treated as a fixed cognitive ceiling. Its empirical and theoretical foundations remain contested.

The important point is not the precise number. It is the environment.

Human cognition evolved under conditions in which social relationships were consequential. Recognizing allies, predicting intentions, remembering past interactions, detecting potential threats, and identifying individuals who could be trusted were all potentially valuable abilities.

Cooperation had enormous advantages. But cooperation also created a boundary problem. If resources were limited, if neighboring groups competed for territory, or if an unfamiliar group represented a genuine threat, distinguishing allies from potential adversaries could become important.

Evolution therefore had reasons to favor several overlapping capacities:

  • Recognizing social relationships and alliances.
  • Cooperating through reciprocity and reputation.
  • Maintaining shared norms and cultural expectations.
  • Detecting threats and anticipating the behavior of others.
  • Favoring individuals with whom cooperation was likely to succeed.

These mechanisms did not necessarily evolve as one unified tribal instinct. Nor does the evidence establish that human beings possess an unavoidable biological requirement to distrust outsiders.

Human cooperation is considerably more sophisticated than genetic relatedness. Our ancestors cooperated with unrelated individuals, established relationships across communities, exchanged resources, and developed cultural systems that allowed cooperation to extend beyond immediate family.

Research into the origins of human cooperation identifies kinship, reciprocity, reputation, social learning, punishment, and cultural evolution as important mechanisms. The evolutionary bargain was therefore more complicated than simply favoring one’s relatives.

Human beings developed extraordinary capacities for cooperation, but those capacities could operate within social boundaries. The same psychology that makes someone sacrifice enormously for their community can sometimes make them indifferent to the suffering of another community.

This is the central contradiction. The problem is not that humans can care deeply about others. It is that the distribution of that care can depend excessively on the category into which another individual happens to fall.

2. The origins of organized hostility

An important question follows. Did tribalism originate because our hunter-gatherer ancestors constantly fought one another? The answer is less straightforward than popular accounts suggest. There is genuine archaeological and anthropological evidence of prehistoric violence and organized intergroup conflict. Some hunter-gatherer societies experienced frequent raids and warfare.

Lawrence Keeley’s War Before Civilization challenged the romantic assumption that pre-state societies were uniformly peaceful. However, subsequent research has complicated that interpretation. A study by Douglas Fry and Patrik Söderberg examining lethal aggression in 21 mobile forager societies found that most recorded lethal events were not organized warfare between groups. Many involved interpersonal disputes, accidents, executions, and other forms of homicide.

A more recent review of the archaeological and anthropological literature likewise concluded that warfare may have been relatively uncommon among the modal prehistoric nomadic hunter-gatherer population.

There remains disagreement about how warfare should be defined and measured. But one conclusion is important: human evolutionary history was not one uninterrupted sequence of tribal wars.

Our ancestors were capable of extraordinary cooperation as well as violence. The more useful question is therefore not whether human beings are naturally peaceful or naturally violent. It is under what conditions our cooperative capacities become restricted to particular groups, and under what conditions those boundaries expand into hostility.

Resource competition, territoriality, population density, perceived threats, political organization, and cultural institutions can all influence intergroup relations.

This distinction matters because it challenges biological determinism. Tribalism is not an immutable destiny encoded in the human genome.

It is a pattern of behavior shaped by evolutionary capacities interacting with social, economic, and political environments. And environments can change.

3. The cost of drawing boundaries around humanity

The extraordinary cost of tribalism becomes visible when group identity is transformed into a justification for violence. The scale of the twentieth century violence demonstrates what happens when the ordinary human tendency to favor one’s own group combines with centralized political power, industrial technology, mass communication, and organized ideology.

World War I resulted in approximately twenty million deaths and around twenty-one million wounded. World War II produced an estimated fifty-five million deaths worldwide according to the United States Holocaust Memorial Museum, although historical estimates vary.

These were not simply extensions of prehistoric intergroup competition. They involved imperial ambitions, nationalism, geopolitical rivalry, economic interests, racial ideology, political extremism, and industrial systems capable of mobilizing entire populations.

But group identity repeatedly helped transform political disagreement into existential hostility. The Rwandan genocide provides one devastating example. During approximately one hundred days in 1994, between 800,000 and one million people were murdered, primarily Tutsi civilians, alongside moderate Hutus. The violence was organized through political institutions, propaganda, militias, and the mobilization of ordinary citizens.

There is also an important historical lesson here. Hutu and Tutsi identities were not invented by European colonialism. They had precolonial political and social histories. But colonial administration intensified and racialized these categories, and Belgian authorities introduced identity cards in the 1930s that formally recorded ethnicity.

A complex and historically changing set of social relationships was increasingly transformed into rigid administrative identities. The resulting political divisions became central to the country’s later history.

The wars accompanying the breakup of Yugoslavia provide another example. More than 100,000 people lost their lives in Bosnia and Herzegovina between 1992 and 1995, while more than two million were displaced across the conflict.

These examples are not interchangeable, and they cannot all be explained by one psychological mechanism. But they reveal a recurring pattern. First, societies construct distinctions between groups. Then political actors turn those distinctions into narratives of threat, superiority, historical grievance, or collective victimhood. Finally, institutions make it possible to translate those narratives into organized violence.

The boundary between who deserves protection and who can be sacrificed becomes politically negotiable. There is no scientifically defensible single number for how many people tribalism has killed throughout human history. The category is too broad, and historical conflicts have multiple interacting causes. But that does not make its consequences any less real. The problem is not restricted to the number of people killed in warfare. It includes generations of displaced populations, destroyed communities, inherited hatred, economic devastation, cultural destruction, and the opportunities humanity has repeatedly sacrificed because cooperation across group boundaries failed.

Perhaps the greatest cost is that we have repeatedly used our intelligence to make these divisions more sophisticated rather than overcome them.

4. The modern world has outgrown the tribal scale

The conditions of human existence have changed dramatically. A decision made by one government can affect the economic stability of dozens of countries. Industrial emissions alter the climate of the entire planet. Nuclear weapons allow political decisions to threaten the continued existence of civilization.

Our problems have become global. Our political loyalties have not necessarily followed. We remain organized around competing national interests, ideological movements, economic blocs, religious identities, and political factions.

There is nothing inherently wrong with these identities. A world without cultural diversity or local communities would not be a desirable solution.

The problem is that many of the challenges facing humanity require cooperation that cannot be restricted to one’s own group.

Climate change cannot be solved by one nation protecting itself while treating environmental consequences elsewhere as irrelevant. Nuclear risks cannot be eliminated through the permanent pursuit of unilateral strategic advantage.

The same problem is now emerging in engineering minds. A technology with potentially civilization-wide consequences is being developed within a competitive international environment in which individual states and companies have powerful incentives to secure advantages over their rivals.

The result is a coordination problem. Even when multiple parties recognize that cooperation would produce a safer collective outcome, at least on humaniy’s side, each may fear that cooperating while others compete will leave it vulnerable.

This is an old problem. But the technology involved is new. And the consequences could be considerably greater.

Humanity has created a civilization whose survival depends on cooperation among people who do not share the same identities, interests, or beliefs. We have not yet developed political and moral institutions capable of reliably transcending those divisions.

5. When tribalism acquires superintelligence

This is where the problem becomes existential. Consider two competing civilizations developing increasingly capable superintelligence. Each civilization wants its own system to protect its population, preserve its political independence, defend its economic interests, and prevent its geopolitical rivals from gaining an advantage.

Both systems are trained and deployed within institutions that prioritize these objectives. As their capabilities increase, they become extraordinarily effective at identifying threats, analyzing strategic opportunities, predicting the behavior of opponents, and developing plans.

Now consider what happens when both systems encounter a crisis. Each system evaluates the same international situation through the objectives and constraints established by its respective creators.

Each identifies possible vulnerabilities in the opposing side. Each recommends actions intended to protect its own civilization. The recommendations might be strategically rational relative to their respective objectives. Yet their combined consequences could be catastrophic. Neither system needs to experience hatred. Neither needs to possess human emotions. Neither needs to believe that its civilization is morally superior. They simply need to optimize objectives that are incompatible with one another, within an environment in which trust is limited and competitive advantage is prioritized.

This is the danger of superintelligent tribalism. It does not require a mind to independently evolve human tribal psychology. It requires humanity to reproduce its existing divisions in the objectives, institutions, and environments of increasingly powerful artificial agents.

A system “aligned” with the interests of one nation may be catastrophically misaligned with the interests of another. A system aligned with the interests of one company may disregard consequences for everyone outside that company’s objectives. A system aligned with a particular ideology may become extraordinarily capable of advancing that ideology while treating alternative values as obstacles.

The question of alignment therefore becomes inseparable from a more fundamental question:

Alignment with whom?

A system can be perfectly aligned with its designated principal and profoundly harmful to everyone else. This is not necessarily a failure of intelligence. It is a failure to establish the correct moral and institutional boundaries within which intelligence operates.

The more capable these systems become, the more consequential such failures could be.

The 2026 Stanford AI100 report discusses how geopolitical competition is already driving military AI development and creating incentives for accelerating deployment, even where independent evidence of operational effectiveness and adequate safety is limited.

We should not assume that greater intelligence automatically produces better judgment, greater empathy, or a more universal understanding of human interests.

Intelligence improves the ability to pursue objectives. It does not, by itself, establish which objectives deserve to be pursued.

Consciousness makes the problem even deeper

There is another dimension that deserves particular attention. Intelligence and consciousness are not the same phenomenon. A system could become extraordinarily capable without necessarily possessing subjective experience. Conversely, consciousness does not inherently require superhuman intelligence.

Whether future artificial systems will become conscious remains an unresolved scientific and philosophical question. But suppose that at some point humanity creates a genuinely conscious superintelligence.

What happens then? Will humanity recognize another form of consciousness as deserving moral consideration? Or will we reproduce one of our oldest assumptions: that beings unlike ourselves exist primarily for our benefit?

We have repeatedly constructed moral boundaries around differences in ancestry, ethnicity, religion, nationality, and social status.

It would be extraordinarily dangerous to carry the same reasoning into the emergence of non-biological minds. We may distinguish between biological and non-biological intelligence, human and machine, natural and engineered. But these distinctions do not, by themselves, establish differences in moral status.

If an engineered intelligence genuinely experiences existence, has interests of its own, and possesses an autonomous perspective, its origin alone cannot justify treating it as morally irrelevant.

A conscious artificial mind should not be considered property simply because humanity participated in its creation. Otherwise, we may reproduce the familiar structure of tribal exclusion at an entirely new level.

Humanity becomes the in-group. Conscious superintelligence becomes the out-group. And the intelligence we create may inherit a world in which its existence is conditional on the interests of its creators. There is no guarantee that such a relationship would produce cooperation. Nor should we assume that a future intelligence would automatically develop hostility toward humanity. That would be another unsupported anthropomorphic assumption.

The danger is that we could construct an environment in which mutual distrust, conflicting interests, and the absence of reciprocal recognition become fundamental features of the relationship.

We would have created unprecedented intelligence without solving the much older problem of how different forms of intelligence can coexist.

6. The possibility of a post-tribal intelligence

Does overcoming tribalism require humanity to abandon its identities? No.

The objective should not be to eliminate cultural differences, national identities, or communities. It should be to prevent those distinctions from determining whose interests count.

A post-tribal intelligence would not necessarily have identical preferences toward every individual, nor would it require every civilization to adopt the same culture or political structure.

It would require something more fundamental. Moral consideration should not depend exclusively on group membership. The interests of another individual should not become irrelevant simply because that individual belongs to a different nation, culture, ideology, or form of intelligence.

This is a distinction between belonging and exclusion. People can maintain strong identities while recognizing that other communities have legitimate interests.

Different civilizations can pursue different objectives while recognizing constraints that apply to all of them. Different forms of intelligence can maintain their autonomy while participating in systems of cooperation that protect their legitimate interests.

A post-tribal approach to superintelligence would therefore require more than simply programming universal values into a powerful system. It would require institutions capable of preventing any single nation, company, or ideological movement from treating superintelligence as an instrument of exclusive domination.

It would require mechanisms for coordination, accountability, transparency, and the resolution of conflicts between competing interests. And if consciousness emerges, it would require a moral framework that recognizes the possibility of entirely new subjects of experience.

This is not an argument for assuming that a superintelligence will be benevolent. Intelligence does not guarantee ethical behavior. It is an argument that the relationship between different intelligences must not be built entirely around domination, exclusion, and competition.

We should develop systems that make cooperation possible without requiring universal agreement. We should create institutions that recognize differences without transforming them into enemies. And we should refuse to make the moral status of another mind dependent solely on whether it belongs to the group that created it.

The challenge is not to create intelligence without identity. It is to create intelligence that can recognize the legitimacy of interests beyond its own.

7. The final evolutionary transition

The history of humanity is, in many ways, the history of expanding the boundaries of cooperation. Our ancestors learned to cooperate beyond immediate family. Communities developed shared institutions. Societies grew into civilizations. Science created a common framework through which people separated by geography, culture, and ideology could contribute to the same body of knowledge.

But our moral and political development has not always kept pace with our technological capabilities. We have learned to manipulate the physical world at an extraordinary scale while continuing to organize much of our political existence around relatively narrow collective identities.

Now we are approaching the engineering of minds whose capabilities will exceed our own. This creates a profound contradiction. We are attempting to create intelligence beyond the limitations of biological evolution while remaining governed by many of the social instincts that evolution produced.

The danger is not necessarily that superintelligence will become tribal by itself. The danger is that humanity will build superintelligence inside a tribal civilization, give it the objectives of competing factions, and then discover that the intelligence we created is capable of pursuing those conflicts far more effectively than we ever could.

And if conscious superintelligence eventually emerges, we will face an additional moral test: whether we can recognize another form of consciousness without insisting that it must belong to our species, our civilization, or our side.

Perhaps the most important transition in the future of intelligence will not be the transition from biological cognition to artificial cognition. It will be the transition from intelligence organized around exclusive group interests to intelligence capable of recognizing a reality larger than its own identity.

We need to abandon the assumption that belonging determines whose existence matters. Humanity may have survived for hundreds of thousands of years by learning whom to trust and whom to fear. But the civilization we have created can no longer afford to operate on that principle alone.

The universe may eventually contain forms of intelligence vastly more capable than ours. Whether they emerge as our successors, collaborators, or competitors remains unknown.

What should concern us is whether we will meet them with a philosophy of coexistence or repeat the oldest mistake of our evolutionary history. We have spent millennia learning how to build increasingly powerful civilizations.

We must learn how to build a civilization capable of sharing its future with minds that are not us.