Multipolarity and Agentic Systems
My Notes from the Fiker Institute in Dubai
Yesterday I was at the Fiker Institute library in Dubai for the Foresight & Global Security Symposium.
The panels focused on foresight, security, and shifts in the international order.
A consistent message emerged across the discussions:
Multipolarity today is not just “more major powers.”
It is countries reclaiming space in a system that never fully served them.
No one used the language of “agentic AI” or “architectures.”
So let’s apply this here.
How We Used to Understand Global Power
Traditional geopolitical categories assumed:
Unipolar → one dominant power (post–Cold War United States)
Bipolar → two superpowers (United States and Soviet Union)
Multipolar → several major powers competing and cooperating
And they assumed:
states are the primary actors
power flows through military, diplomacy, and markets
alliances reflect shared interests or values
But the world we live in now doesn’t fit these assumptions.
What Multipolarity Really Means Today (Based on Yesterday’s Panels)
Across the sessions, speakers made a clear argument:
Multipolarity is not fragmentation. It is reclamation.
Countries across Asia, Africa, and the Middle East increasingly view themselves not as the “Global South,” but as the Global Majority:
they represent most of the world’s population
they carry many of the world’s resources
they want leverage, not alignment
they want to set norms, not simply follow them
Underneath this was a more direct point:
The post-WWII system was never neutral.
It privileged the countries that built it.
But another recurring theme followed immediately:
Reclaiming agency has not removed structural dependency.
Countries still rely heavily on:
United States and European financial infrastructure
Chinese mineral processing and manufacturing
Gulf energy and compute centres
Korean industrial capacity
Silicon Valley cloud and AI platforms
One panelist used a concrete example: several Southeast Asian countries created a shared QR-code payment network so citizens could make cross-border payments without relying on U.S. or European financial systems. It was a small step toward digital sovereignty. But Washington later labelled the system a trade barrier because it bypassed established Western payment infrastructure. The point was about how quickly “sovereign” systems can run into the limits of larger architectures they do not control.
The world is politically multipolar,
but structurally interdependent.
And that tension is driving many of the security and governance challenges we’re seeing now.
Now let’s add the AI layer ourselves, because this is key to defence discussions.
Why This Definition Breaks Down in the Age of AI
Nothing in the symposium addressed agentic AI directly, but the implications flow naturally from the conversations.
Traditional geopolitics assumes that
states shape systems.
But in the AI era:
systems shape states.
As AI systems become more agentic, meaning they can act, plan, and coordinate, national decision-making becomes increasingly mediated by:
who controls the compute
which models filter information
how automation is integrated
which safety and escalation logic is embedded
which architecture interprets the world
This means:
Multipolarity is shifting from a political concept to a computational one.
It no longer maps to blocs like “United States,” “China,” “European Union,” or “Global Majority.”
It maps to AI architectures.
The World Is Splitting Into Multiple Architecture-Driven AI Civilizations
Within the next decade, regions and tech ecosystems are building distinct AI operating systems with their own:
memory structures
reasoning patterns
safety logic
alignment assumptions
escalation thresholds
narrative frames
data and compute foundations
This goes far beyond “different models.”
What is forming are what we might call agentic civilizations:
systems that interpret reality differently at the machine level.
Examples already emerging:
United States commercial architecture (OpenAI, Anthropic, Google)
China state-aligned architecture (Alibaba, Baidu, Huawei)
European Union compliance-driven architecture
UAE, Korea, Singapore sovereign compute corridors
Open-source federated reasoning ecosystems
As governments embed these systems into intelligence, military planning, diplomacy, and crisis response, these architectures will directly influence national behaviour.
This is the next phase of multipolarity.
(Yesterday was a reminder that foresight isn’t about certainty, it’s about influence - so let’s keep going)
The New Risk: Cognitive Misalignment Between Architectures
Traditional multipolarity caused:
miscalculations
competing alliances
economic friction
Architectural multipolarity introduces something new:
machine-level divergence in how events are interpreted.
Two allied countries could be politically aligned
but computationally misaligned.
One architecture might:
escalate faster based on pattern recognition
classify a threat as imminent
recommend action
While another might:
classify the same signal as noise
recommend restraint
delay warning
This is not a diplomatic disagreement.
It is incompatibility at the system layer.
And it carries real security risk.
What Needs to Be Built
The instinctive answer “one global model” will not happen.
Different regions have sovereign reasons to maintain their own systems.
Which means the real work is:
How do we coordinate across incompatible agentic systems?
Not unification.
Translation.
We need:
cross-architecture interoperability
transparent reasoning bridges
shared uncertainty formats
machine-to-machine negotiation pathways
escalation mediation logic
cognitive translation layers
This is the infrastructure missing today.
Without it, political multipolarity becomes computational fragmentation. What does that look like? More to follow. I’m heading back into the conference now.
