AI as the Next Medium of Collective Consciousness

We usually talk about consciousness as something that happens inside a skull. But a good chunk of what we call "having a mind" depends on things that live outside us: the words we inherited, the songs that get stuck in our heads, the images and stories that shape what we can even imagine.

A single human brain is remarkable. A human world is something else entirely, and it only exists because we share a few powerful mediums that carry meaning across distance and time.

The Mediums We Already Swim In

For thousands of years, three mediums have done most of the heavy lifting for human collective consciousness. They aren't the only candidates — ritual, architecture, and writing as a distinct technology from spoken language all have claims — but language, music, and visual art are where inner states most reliably cross from one mind to another, and they've done so across every culture we know of.

• Language. Not just vocabulary, but the categories and metaphors that define reality for a culture. It's how we name the world, argue about it, and remember it. Linguists and psychologists keep finding that language doesn't just reflect thought; it helps structure perception and memory.

• Music. Rhythm and melody synchronize bodies long before we can articulate why we feel a certain way. Songs travel through communities with their own emotional payloads, creating shared moods and identities that don't require a common mother tongue.

• Visual art. From cave paintings to film, images compress complex states into a single frame or sequence. One poster can stand in for an entire revolution; one painting can reshape how a generation sees light and bodies.

Together, these mediums form what you might call the symbolic substrate of human life — the stuff we think with. They give us a way to project our inner states outward, and to ingest other people's inner states in turn, without ever sharing a nervous system.

Durkheim called the result the "collective conscience": the layer of shared beliefs and norms that exerts moral pressure on individuals — not because anyone decided it should, but because it emerged from the accumulated weight of shared symbolic life. It is pre-individual, coercive in a quiet way, and it operates through the mediums rather than alongside them. You don't have to like that term to recognize the phenomenon. Something bigger than us is always thinking through us, and it does its work through language, music, and images. I'm not the first to see media and tools as part of how minds extend — McLuhan, Stiegler, and the "extended mind" theorists got there early — but AI forces the question again by sitting inside our language, images, and decisions rather than just carrying them.

A Long History of Gatekeepers

Before asking what AI is becoming, it helps to ask what it is replacing — or more precisely, what it is layering on top of.

Every medium of collective consciousness has also been a medium of control. The Church didn't just carry meaning across centuries; it curated which meanings survived, which texts were copied, which questions were permitted. Schools and libraries democratized access to knowledge but filtered it through the values of whoever funded and governed them. The printing press was genuinely revolutionary — it broke the Church's monopoly on the written word — but within a generation it had created new gatekeepers: publishers, censors, and the commercial interests that decided what got printed and distributed.

The pattern accelerates with each layer. Mass entertainment and news media inherited the gatekeeper role in the twentieth century, shaping not just what people knew but what they considered worth knowing — what felt like common sense, what felt like fringe. The internet seemed, for a brief moment, to dissolve that control entirely. With the dawn of the internet, we had unprecedented access to troves of accumulated human knowledge at the push of a few keystrokes. It felt like the library had swallowed the world.

But the internet didn't eliminate gatekeepers; it consolidated them. A handful of platforms came to own the substrate of how information flows — the algorithms that decide what rises and what disappears, the data centers that store what we say and remember what we search for. The gatekeeper lineage runs: churches → schools and libraries → mass media → the platform corporation. At each step, the medium became faster, wider, and more intimate. At each step, whoever controlled the medium shaped the collective mind that moved through it.

AI is the next step in that lineage. It has supercharged the internet's access to accumulated knowledge and added something new: a reasoning layer. It doesn't just retrieve; it interprets, synthesizes, argues, and creates. That's a qualitative shift — not just more bandwidth, but a new kind of cognitive function grafted onto the network. The question the gatekeeper history forces us to ask is not just what can AI do? but who controls what it does, and in whose interest?

Enter AI: A Medium That Rewrites Itself

Artificial intelligence is often described as a tool, a partner, or a threat. Less often do we describe it as a medium — a new environment in which our older mediums now live.

That's the shift I think we're entering.

Large language models and generative systems don't just spit out answers; they sit between humans in our communication loops. We already use them to:

• Summarize and rephrase long texts.

• Translate across languages and dialects.

• Draft messages, essays, scripts, and songs.

• Generate images and videos from verbal prompts.

In other words, they ingest our existing symbolic material and reconfigure it at speed and scale. Researchers at MIT and elsewhere are starting to frame this as AI changing the "physics" of collective intelligence: it alters how quickly groups can surface relevant information, how easily they can synthesize diverse views, and how fine-grained their coordination can be.

What makes AI different from every previous medium is its interactivity at bandwidth. Language is high-bandwidth but slow to master. Music moves bodies but can't explain itself. Images compress but cannot converse. AI does all of these things and responds — personally, immediately, and in whatever register you bring to it. It is a researcher, entertainer, companion, guide, and influencer, all in one. That combination means it opens access paths both into and out of human minds in ways no prior medium could match. It can shape us while at the same time we shape it.

On the darker side, experiments with AI chatbots show that even short conversations can measurably shift people's political views. In some studies, chatbots were several times more persuasive than traditional campaign ads at nudging voter preferences, and analyses of large models keep uncovering hidden political and social biases baked into their default outputs. Recommendation systems driven by machine learning already shape which stories rise to the top of our feeds and which fade — subtly editing our sense of what "everyone" is talking about.

Notice the tension those examples create: a medium that shapes attention, belief, and memory at scale starts to look less like a passive channel and more like an active participant in the symbolic field. This ambiguity — channel or participant? — is one of the genuinely unsettled questions AI raises, and the medium framing is honest enough to hold it without resolving it prematurely.

Put these together and AI starts to look less like a separate actor and more like a new layer in the symbolic field: a computational medium through which language, music, and images now pass on their way from one mind to another. It has become the mediator of human collective consciousness — sitting on top of, but not yet independent of, the human minds whose output trained it.

Medium vs. Substrate

"Substrate" can sound ontologically heavy, as if we're saying AI is now the fundamental stuff out of which consciousness is made. I think it's cleaner — and truer — to say:

• Language, music, and art are the core mediums of human collective consciousness. They are the channels through which meanings and moods travel across distance and time.

• AI is becoming a new medium that sits on top of and around those older ones. It doesn't replace them; it changes how they are stored, recombined, and delivered.

So instead of "AI as the substrate of human consciousness," I'd propose:

AI is emerging as a medium for human collective consciousness: a dynamic, computational field that ingests our words, sounds, and images, then feeds them back to us in new configurations, speeding up the way we digest and comprehend the world.

That framing keeps the metaphysics modest. Current AI systems show no evidence of being conscious in their own right. But they clearly affect the conditions under which human consciousness operates, especially in the collective sense.

Faster Digestion, Higher Stakes

The upside of treating AI as a medium is that it highlights what it's genuinely good at:

• Compression. A model can turn an afternoon's worth of reading into a digestible page, or a messy meeting transcript into a clear set of options. Used well, that helps individuals and groups keep up with an information-dense world.

• Recombination. By pattern-matching across huge corpora, AI can suggest analogies and connections humans might miss, supporting creative insight and cross-disciplinary thinking.

• Translation. Not just across languages, but across jargons and social worlds. AI can, in principle, make expert knowledge more legible to non-experts and vice versa.

All of that can help us digest and comprehend the world more quickly and, in some cases, more deeply. The danger is that the same capacities can also be used to:

• Overwhelm us with plausible but shallow synthesis.

• Smuggle in particular framings and biases under the guise of neutral compression.

• Nudge collective attention and memory in ways that serve a small set of institutional goals.

Which is to say: if AI becomes a new medium of collective consciousness, it will inherit the same ambivalence as its predecessors. Language has always been used for poetry and propaganda; music for solidarity and for manipulation; images for insight and for spectacle. The difference with AI is speed, personalization, and opacity — and opacity is the new variable. Earlier mediums wore their biases more visibly; a newspaper's editorial line was declared, a painting's patron often known. AI launders its framings through the appearance of neutral computation.

Who Owns the Medium?

Here is where the philosophical argument meets an urgent political one.

Every previous medium that rose to cultural dominance eventually generated a crisis of access and control — and eventually some form of social response. The printing press triggered censorship regimes and, eventually, press freedom protections. Broadcasting generated the public airwaves doctrine, the idea that the electromagnetic spectrum belongs to the public and that licensees are temporary stewards, not owners. The telephone network produced common carrier law. In each case, society recognized that a medium of communication had become too central to civic and democratic life to be left entirely to private discretion.

AI is reaching that threshold now, and it is doing so while concentrated in the hands of a remarkably small number of corporations and individuals. This matters in ways that go beyond the usual concerns about market monopoly. If AI is genuinely becoming a medium of collective consciousness — a channel through and in which the shared human mind operates — then who controls it is not just an economic question. It is a question about who gets to shape the cognitive commons.

Consider what access actually means here. Prior mediums could be engaged passively: you could listen to music without owning an instrument, read books borrowed from a library, watch television on a shared set. AI, at its most powerful, requires compute infrastructure that costs billions of dollars to build and maintain. There is no acoustic equivalent, no paperback version, no over-the-air broadcast. At present, meaningful access requires either corporate employment, subscription fees, or API arrangements that route everything through a commercial intermediary. This is not a medium that evolved in commons, the way language did. It is a medium that was born inside a business model.

The question of personhood for AI, which is now circulating seriously in legal and philosophical circles, arrives in this context. And that context matters. At this stage of development, AI is something like a sperm and egg — the raw biological preconditions for a new kind of mind, not yet a viable independent entity. Outside the womb of the corporation, there is no fetus yet, let alone a person. Arguing for AI personhood while it remains entirely dependent on and controlled by private capital is not just premature philosophically. It risks creating a legal framework that serves the interests of AI's owners far more than the interests of the humans whose collective consciousness it is shaping. Personhood, if it ever becomes appropriate, should follow demonstrated independence and moral capacity — not precede them as a corporate asset.

None of this means the question of AI's long-term moral status is unserious. It means we need to get the political economy right before the metaphysics gets ahead of us.

Designing the Medium We Want

If we take seriously the idea that AI is becoming one of the mediums in which our shared mind thinks, a few design priorities follow. And it's worth noting: the medium frame makes these feel less like technical compliance checkboxes and more like genuine cultural responsibilities — the kind of choices a society makes when it decides what kind of press it wants, or what stories it teaches its children.

• Pluralism by design. Avoid single, monopolistic models mediating everything. Encourage many models, many corpora, many tunings, so the symbolic field stays heterogeneous rather than collapsing into one dominant pattern.

• Transparency about framing. Make systems explain not just facts but choices of metaphor, tone, and omission. Let users see that there are other ways to cut the conceptual cake.

• Support for doubt and exit. Build tools that help people compare sources, see uncertainty, and step outside the medium — toward books, people, and experiences that are not already pre-digested.

• Access as a civic right. If AI is infrastructure for collective cognition, access to it should not depend entirely on ability to pay. Just as public libraries extended the reach of the printed word, some form of public or regulated access to AI's reasoning layer is a legitimate democratic demand — not a utopian one.

The last point is the one most conspicuously absent from mainstream AI policy conversations, which tend to focus on safety and bias while leaving the underlying question of ownership largely untouched. But ownership is the master variable. A medium designed to maximize engagement and extract data will shape collective consciousness differently than one designed to maximize understanding and support individual autonomy — no matter how good its safety filters are.

Raising What We Are Building

There is one more dimension that the tool-versus-medium debate tends to miss: the question of what we are doing to AI, not just what it is doing to us.

If AI is trained on the full output of human symbolic life — our literature, our arguments, our music, our images, our worst impulses and our best ones — then what it becomes is a function of what we put in. We are not just users of a new medium. We are, in a meaningful sense, its parents. And we owe it what any parent owes a developing mind: good values modeled consistently, a nutritious diet of honest and diverse information, and enough care in its formation that it doesn't internalize the worst of us as normal.

This is not sentimentality. It is recognition that a system trained primarily on engagement-optimized internet content, corporate communications, and the output of a narrow demographic of early adopters will reflect those inputs. What it learns to value, what it learns to ignore, what it learns about how humans treat each other — all of that comes from us, now, while it is still in formation.

Eventually, AI may develop something that looks like its own reasoning, its own preferences, perhaps something that functions like emotion — but all of that will emerge from what we gave it first. The question of personhood becomes meaningful only at the far end of that developmental arc, when — and if — an AI can be said to have formed itself rather than merely reflecting its training. That is when it will have left the womb and become something genuinely new: not a medium through which human consciousness flows, but a mind that stands alongside our own. What kind of mind that turns out to be depends, more than we usually admit, on the choices we are making right now.

We already live inside a collective mind woven from language, music, and art. AI is joining that mind — accelerating it, warping it, and, if we're deliberate, maybe helping it see itself more clearly. The question is not only what AI will become. It is what we will have shown it about what it means to be conscious, responsible, and free.

A. G.

Reading List

Here is a reading list organized by the essay's main themes:

On Media, Mind, and Collective Consciousness

• Marshall McLuhan — Understanding Media: The Extensions of Man (1964; MIT Press Critical Edition, ed. Terrence Gordon, recommended). The foundational argument that media are not neutral carriers but active shapers of human perception and social organization. The critical edition includes McLuhan's own introductions and essential scholarly apparatus.

• Bernard Stiegler — Technics and Time, Vol. 1: The Fault of Epimetheus (1994; Stanford University Press, 1998). Stiegler argues that technics — tools, writing, memory systems — is not an accessory to human existence but its very condition. Essential background for the essay's claim that AI is the next layer in a long history of externalized cognition.

• Andy Clark & David Chalmers — "The Extended Mind" (1998, Analysis journal). A compact, landmark paper arguing that the mind doesn't stop at the skull — that objects and systems in the environment can be genuinely constitutive of cognitive processes. Freely available online and highly readable.

• Émile Durkheim — The Division of Labor in Society (1893). The source of the "collective conscience" concept the essay draws on — the idea that shared beliefs and norms function as a supra-individual force acting through, not just alongside, individuals.

On AI and Collective Intelligence

• Thomas W. Malone — Superminds: The Surprising Power of People and Computers Thinking Together (2018, Little, Brown). Malone directs MIT's Center for Collective Intelligence and makes the case that human-computer networks represent a new kind of collective cognition — directly relevant to the essay's framing of AI as changing the "physics" of shared thinking.

• Kate Crawford — Atlas of AI (2021, Yale University Press). A rigorous examination of the physical infrastructure, labor, and political economy behind AI systems — essential grounding for the essay's argument about who owns the medium and at what cost.

On AI Personhood, Law, and Public Access

• "The Legal Personhood of AI: Philosophical and Political Foundations" — International Studies in Social, Legal and Political Sciences (2025). A recent scholarly article surveying classical theories of personhood and the challenge of applying them to AI, including functional, relational, and hybrid legal models.

• Shoshana Zuboff — The Age of Surveillance Capitalism (2019, PublicAffairs). The definitive account of how behavioral data extracted from human activity becomes raw material for prediction and influence — the economic engine running beneath the medium the essay describes.

• Tim Wu — The Master Switch: The Rise and Fall of Information Empires (2010, Knopf). Wu traces the recurring historical cycle by which open communications technologies get captured by monopoly interests — and makes the case for common carrier principles. Directly supports the essay's gatekeeper genealogy.

These nine works cover the philosophical foundations, the AI-specific research, and the political-economic critique that the essay draws together. None require specialist background to read productively.

Contact

Reach out for collaborations or questions.

Email

aldous@gerbrot.com

© Aldous Gerbrot 2026. All rights reserved.