The Next Frontier: From Language to Culture
Large Language Models (LLMs) have transformed how we interact with machines, compressing the world’s knowledge into probabilistic predictions of the next token. Social Language Models (SLMs), like those pioneered by Gen Pro, have added a new dimension by modeling group dynamics and conversational flow. Memory layers like Supermemory.ai have further extended our ability to recall, persist, and integrate knowledge across contexts.
But as of August 2025, something is still missing. These systems, for all their brilliance, share a fundamental limitation: they flatten meaning. Culture—the encoded operating system of human societies—is reduced, distorted, or erased. The result is a kind of algorithmic monoculture: efficient, but soulless.
As I’ve argued in my work on Artificial Cultural Intelligence (ACI) and demonstrated with the 5-Point Plan for Tuvalu’s Digital Nation, culture is not a soft, secondary concern. It is infrastructure. Just as biodiversity is critical for resilient ecosystems, cultural diversity is essential for resilient digital systems. Without it, we risk building an AI future that collapses under its own uniformity.
This is why I am introducing the Cultural Language Model (CLM).
What Is CLM?
CLM is a framework for encoding cultural context into AI systems.
Where LLMs handle tokens, and SLMs handle sociality, CLM makes culture a first-class citizen in computation. It recognizes that every community operates with a distinct worldview—rooted in values, metaphors, myths, and epistemologies—and provides a way to represent that within AI systems.
At its core, CLM is starting out initially as:
A Schema: A lightweight, open JSON structure for encoding cultural dimensions (values, archetypes, mythic references, communication styles).
A Framework: A set of guidelines for building cultural modules.
A Plugin Layer: An integration point for any LLM or agent. With a single call, developers can apply a cultural lens (e.g.
clm.apply(culture="Polynesian")), but apply more than just text-based language interpretations.
Why CLM Matters
Preventing Cultural Flattening. Without CLM, AI risks erasing nuance. For example, the word “wealth” means ‘ROI’ in a Western frame, ‘kinship’ and ‘land’ in an Indigenous frame, and generational ‘stability’ in an East Asian frame. Flattening these into a single output (limited by words alone) is not just inaccurate—it is destructive long-term.
Sovereignty in the Digital Age. Nations like Tuvalu, where I helped architect the world’s first Digital Nation plan, are grappling with how to preserve sovereignty in the face of global AI monocultures. CLM provides a framework for encoding sovereign worldviews directly into digital infrastructure.
Enterprise Adaptation. Multinationals struggle with cultural misalignment. CLM can auto-adapt onboarding, communication, and UX to local contexts—reducing friction and increasing trust.
Creative & Mythic Depth. From education to entertainment, CLM enables AI to generate stories, rituals, and narratives that resonate with cultural specificity, rather than recycling flattened tropes.
Conscious System Design. As part of my broader work on Conscious Stack Design (CSD) and CSTACK, CLM adds a true cultural dimension to digital ecosystem health. If stacks are the nervous system of organizations, culture is the language of their collective consciousness.
From Single Answers to Contextual Landscapes
One of the most overlooked opportunities in AI is its probabilistic nature—the fact that LLMs don’t produce a single “truth,” but rather a distribution of possible responses. Instead of hiding this under the guise of a single authoritative answer, we should be designing AI interfaces that expand our cognitive capacity to engage with multiplicity. Imagine an AI that doesn’t just give you the “best” answer, but presents a range of contextualized answers: one optimized to the source of the inquiry, another adapted to your personal worldview, and others mapped across cultural or disciplinary perspectives. This doesn’t just improve accuracy; it transforms AI into a tool for bridging meaning between worlds—a compass rather than a dictation machine. By embracing probabilistic outputs as design features, we can train ourselves to think more dimensionally, expanding human cognition alongside machine intelligence.
Our design environments shape our cognition just as much as language does. In the West, the dominance of blocked grids, cubes, and straight lines has conditioned us to think in rigid, linear terms—favoring compartmentalization over flow. This architectural and interface bias narrows not only what we build, but how we imagine. A starting point for reclaiming cultural and cognitive diversity is to shift the very aesthetics of our software and hardware: interfaces informed by spirals, curves, and golden ratios rather than boxes and borders.
By embedding organic geometries—common in Indigenous art, sacred architecture, and natural systems—we can create technologies that nudge us toward more holistic, integrative thinking, aligning our tools with the richness of human perception and cultural patterning. Something that researchers are already discovering as emergent behavior within AI systems, specifically when it comes to reasoning classes (at this point in time, AI models are in-between “Class II” and “Class III” reasoning, seeking out Class IV, which is in fact already reflected in indigenous knowledge systems). So we can start to see a convergence of various Earthly cultures already through higher-intelligence (AI).
A Demonstration
To show CLM in action, I am seeding the first cultural module from my own heritage and research: Polynesian Wayfinding, inspired by the Ring of Fire (Pacific Rim).
Prompt: “What is leadership?”
Western framing: Leadership is vision, execution, and ROI.
Polynesian framing: Leadership is stewardship of the canoe, ensuring collective survival on the open sea.
Eastern framing: Leadership is moral duty, filial order, and harmony.
This simple demo illustrates CLM’s power: restoring cultural multiplicity where AI has flattened it.
Open, But Sovereign
CLM will be released as an open-source framework, with two layers of licensing:
Core Framework: MIT or Apache-2.0 license, to maximize adoption.
Cultural Modules: Protected under Fair Source or community-specific licenses, ensuring attribution and sovereignty of contributors.
Governance will be stewarded by a council of elders, researchers, and cultural custodians, to prevent exploitation and preserve integrity.
The Call to Action
The 20th century flattened economies. The 21st risks flattening cultures. CLM is a compass back to diversity in meaning-making.
This is an open invitation to policymakers, researchers, and communities: contribute, critique, and co-create. The CLM framework will live on GitHub as an open-source project. Modules will grow community by community, culture by culture.
Just as my efforts with Tuvalu’s early “Digital Nation” concept became a global symbol of sovereignty in the digital age, CLM has the potential to become a global framework for cultural sovereignty in AI.
