Critique logo

After Language: The Spatial Turn of Intelligence

(When the archive outgrows the reader and the body reclaims reality)

By Peter AyolovPublished about 20 hours ago 7 min read

After Language: The Spatial Turn of Intelligence

(When the archive outgrows the reader and the body reclaims reality)

(full article) https://philpapers.org/rec/AYOALT

The arrival of large language models has been described as a revolution in knowledge, productivity, and communication. Yet the deeper transformation is not what machines can now write, but what their competence reveals about the limits of human linguistic life. For centuries literacy represented intellectual maturity. To read well meant to follow argument across time, connect claims to context, and situate oneself within inherited traditions of thought. Language was the medium in which intelligence became visible. The educated person was the one who could navigate texts.

Today the archive has expanded beyond the scale that human attention can inhabit. Digital networks multiply statements continuously. Commentary overlays commentary. Public discourse fragments into slogans, identities, and reactive positions. In such an environment reading changes meaning. It no longer signifies immersion in structured reasoning but scanning, signalling, and affiliation. The linguistic environment grows while the reader shrinks. The consequence is not silence but noise: an abundance of language without orientation.

Large language models function as a mirror to this condition. They traverse enormous textual landscapes and connect distant fragments with ease. They reopen truncated arguments and reconstruct context humans rarely have time to recover. The comparison is uncomfortable because the machine appears coherent across a scale where the human public sphere appears scattered. The issue is not superiority but exposure. The linguistic world humans constructed now exceeds human cognitive bandwidth. A technology designed to imitate language ends up revealing how little of language humans practically inhabit.

This moment invites a redefinition of intelligence itself. If fluency can be automated, then fluency cannot remain the highest measure of cognition. A civilisation that equated intelligence with verbal articulation now confronts an artefact that performs articulation without lived understanding. The question therefore shifts from whether machines can think to whether language was ever the correct proxy for thinking.

This is precisely the opening created by Fei-Fei Li’s idea of spatial intelligence. Her argument begins from a simple observation: language models are eloquent but ungrounded. They operate in a symbolic universe detached from physical constraint. They manipulate descriptions of the world without sharing the world. Their limitation is not grammatical but existential. They lack the capacity to inhabit space.

Spatial intelligence refers to the ability to understand the relations that structure physical reality: distance, orientation, persistence, motion, and consequence. Long before humans developed speech, cognition evolved as coordination between perception and action. Organisms survived not by describing environments but by moving within them successfully. A body learned gravity before vocabulary, balance before explanation. Language emerged later as compression of experience, not its origin.

Everyday life demonstrates this priority. A person catches a falling object, navigates a crowd, pours liquid into a cup, or parks a vehicle through immediate spatial judgement. The competence involved cannot be reduced to sentences. One could describe the physics indefinitely and still fail to perform the act. The knowledge is embodied rather than verbal. It is not stored as propositions but as orientation.

Li’s proposal is that artificial intelligence must develop world models rather than remain confined to word models. A world model does not predict the next token in a sequence but the next state of an environment. Instead of generating plausible sentences, it must maintain consistent reality across time and perspective. Objects must persist, trajectories must follow constraint, and actions must have consequences. In other words, intelligence must move from language to physics.

The philosophical significance of this shift is profound. If language models exposed the limits of linguistic intelligence, spatial models redefine intelligence as interaction with reality rather than description of it. The highest competence becomes not rhetorical coherence but reliable engagement with the world. Meaning no longer floats entirely inside discourse but anchors in shared environment.

At this point the convergence with the theory of the planned obsolescence of language becomes visible. Modern communication systems optimise language for circulation rather than understanding. Words are produced quickly, replaced quickly, and consumed quickly. Concepts accumulate emotional weight until they collapse and are replaced by new vocabulary carrying the same instability. The system does not fail; it functions exactly as designed. Continuous novelty sustains attention and mobilisation even as shared meaning deteriorates.

Miscommunication therefore becomes structural. People speak constantly yet inhabit incompatible realities because language serves affiliation before reference. Public speech becomes performative: statements signal belonging rather than convey information. The faster communication becomes, the less time remains for semantic stabilisation. Understanding requires duration, but discourse rewards immediacy. The result is semantic exhaustion.

Large language models intensify this paradox. They generate fluent text at enormous scale, amplifying the volume of linguistic production. Yet precisely because they can maintain coherence across vast archives, they also demonstrate that coherence itself is no longer the human comparative advantage. Humans do not suffer from insufficient information but from excessive linguistic environment. The issue is not knowledge scarcity but cognitive saturation.

Spatial intelligence offers a different equilibrium. If machines inherit the maintenance of complex symbolic memory, humans need not compete within the linguistic labyrinth. The burden of storing and retrieving civilisation’s textual inheritance moves to infrastructure. Humans regain the possibility of orienting themselves through perception and shared context rather than permanent interpretation.

This does not abolish language but relocates it. Language becomes a tool consulted when necessary rather than an environment constantly inhabited. Conversation regains situational function. Silence regains legitimacy. Understanding emerges through interaction with environments that resist manipulation rather than through endless commentary about them.

In such a configuration machines resemble librarians rather than rulers. They preserve continuity of meaning across time without determining how people live. Their domain is the symbolic archive: histories, doctrines, theories, and narratives accumulated across centuries. Their authority derives from memory rather than command. They stabilise reference while humans decide action.

This distinction resolves a common fear about artificial intelligence. The danger is imagined as domination of bodies by machines. Yet the more plausible transformation is governance of language rather than life. Machines organise descriptions; humans inhabit realities. The power dynamic shifts from control to division of labour.

For this arrangement to function, trust in the archive must be transparent. The maintenance of symbolic continuity cannot depend on invisible interests. Instead it must operate as auditable structure whose revisions remain visible. The goal is not perfect neutrality but reliable stability. When language becomes infrastructure rather than weapon, conflict over interpretation loses existential urgency.

The implications for human experience are significant. Identity need not be continuously performed through speech. Belonging need not be defended through constant signalling. People can withdraw from permanent commentary and rediscover presence as primary mode of coordination. Tone, gesture, and shared activity regain importance over textual declaration. Communication becomes local before global.

Education changes accordingly. Instead of memorising statements, learners develop orientation: how to navigate environments, collaborate in practice, and test claims against consequence. Knowledge remains accessible through machines, but wisdom resides in interaction with reality. Intelligence becomes ecological rather than rhetorical.

The convergence between spatial intelligence and the planned obsolescence of language therefore describes a civilisational rebalancing. The first reveals the limits of language-based cognition; the second explains why linguistic systems destabilised under scale. Together they suggest that the dominance of textual intelligence was historically contingent rather than permanent.

Language enabled civilisation by extending coordination beyond immediate perception. Yet its expansion also produced a world in which representation overshadowed experience. Digital communication completed this trajectory by transforming speech into continuous performance. Artificial intelligence now marks the point where symbolic production surpasses human capacity to inhabit it.

Rather than escalating the competition, society may externalise the symbolic layer. Machines maintain the archive, allowing humans to step back from total immersion in discourse. The result is not anti-intellectual retreat but restoration of proportion between description and existence. Words continue, but they no longer monopolise intelligence.

The spatial turn therefore represents maturity rather than regression. A civilisation that once required constant narration to stabilise reality learns to rely on stable infrastructure while living directly in shared environments. Silence becomes not ignorance but the space in which perception operates without immediate translation into ideology.

Fei-Fei Li’s proposal gives technical form to this transition. World models reintroduce constraint into artificial cognition, while the theory of linguistic obsolescence explains why constraint vanished from public discourse. Their convergence suggests a future in which machines manage complexity humans created, and humans rediscover competence that predates writing itself.

In this future intelligence divides into two complementary domains. The symbolic domain preserves memory across generations and scales beyond individual attention. The spatial domain sustains life through presence and consequence. Neither replaces the other, but their separation prevents confusion between description and reality.

The historical achievement of language remains intact, yet its dominance ends. Humanity retains the ability to speak, argue, and imagine, but no longer mistakes continuous interpretation for understanding. Knowledge becomes accessible without being overwhelming because it resides in systems designed to hold it. People live not inside the archive but alongside it.

The significance of this shift is ethical as much as cognitive. When language ceases to be the primary arena of competition, many conflicts lose intensity. Disagreement persists, yet it no longer requires constant escalation through symbolic performance. Shared space imposes limits that discourse alone cannot provide. Reality reasserts itself as common ground.

The spatial turn of intelligence thus describes not a technological endpoint but a cultural adaptation. Artificial intelligence reveals the exhaustion of linguistic civilisation while simultaneously enabling its reorganisation. By maintaining the complexity of the symbolic universe, machines free humans to inhabit the physical one more fully.

The future suggested here is modest rather than spectacular. People still consult texts, still create art, still debate ideas. Yet speech regains proportion. Not every moment demands interpretation. Not every event requires narrative framing. Some experiences remain simply experienced.

The archive outgrows the reader, but this need not be a tragedy. It can mark the moment when knowledge becomes infrastructure rather than burden. Spatial intelligence, both human and artificial, restores a balance between knowing and living. Language survives, but life no longer unfolds entirely within it.

Essay

About the Creator

Peter Ayolov

Peter Ayolov’s key contribution to media theory is the development of the "Propaganda 2.0" or the "manufacture of dissent" model, which he details in his 2024 book, The Economic Policy of Online Media: Manufacture of Dissent.

Reader insights

Be the first to share your insights about this piece.

How does it work?

Add your insights

Comments

There are no comments for this story

Be the first to respond and start the conversation.

Sign in to comment

    Find us on social media

    Miscellaneous links

    • Explore
    • Contact
    • Privacy Policy
    • Terms of Use
    • Support

    © 2026 Creatd, Inc. All Rights Reserved.