Through McLuhan’s Lens

The Regulated Unknown

July 18, 2026 | 2680 words

Through McLuhan’s Lens: The Regulated Unknown

There is a phrase circulating through the AI policy world that deserves more scrutiny than it has received. Regulators, drafting frameworks on both sides of the Atlantic, have settled on the language of “risk-based” governance. The European Union’s approach sorts artificial intelligence into tiers — unacceptable risk, high risk, limited risk, minimal risk — and applies rules accordingly. It is a tidy architecture. It sounds responsible. It borrows the grammar of pharmaceutical approval and financial oversight, domains where we know what a drug is and what a derivative does.

But notice what this grammar assumes. To sort something into a risk tier, you must first know what the thing is. You must be able to name it, bound it, describe its mechanism. The remarkable feature of this week’s discourse is how much energy flows toward the sorting and how little flows toward the naming. The debate is loud about which agency should enforce, what penalties should apply, whether a given system counts as “high risk.” It is nearly silent on a prior question: what, precisely, is being governed?

This essay treats that silence as its subject. The gap between the volume of regulation-talk and the poverty of comprehension-vocabulary is not a minor imbalance to be corrected later. It is the phenomenon itself. And it becomes visible only when the regulation discourse is examined not as a set of proposals but as a medium — a form that shapes perception before it delivers any content.

The Ratio That Should Alarm Us

Consider the shape of the conversation. When public attention to artificial intelligence is measured, the overwhelming share clusters around control: who governs, what rules apply, which body enforces, what happens when the rules are broken. A far thinner sliver addresses what the technology actually is, how it works, or what vocabulary a citizen would need to think about it clearly. The debate has organized itself around the verbs manage, restrict, audit, and comply — and largely abandoned the verb understand.

This lopsidedness matters because of what it trains us to do. A public that spends its attention on the question “who should be in charge of this?” is a public that has already conceded the more important question: “do we know what this is?” The concession is invisible because it is never spoken. It is built into the form of the discussion.

Herbert Marshall McLuhan spent his career insisting that the form of a communication does more to us than its content. His most famous and most misunderstood formulation — “the medium is the message” — means, in plain terms, that a technology reshapes how we perceive and relate to one another regardless of what it carries. In Understanding Media, he put it bluntly: “the ‘content’ of any medium is always another medium.” We fixate on the message and miss the medium, the way a burglar throws meat to the watchdog to occupy its attention while the real work goes on elsewhere.

The regulation discourse is a medium in exactly this sense. Its content is a torrent of proposals, tiers, and enforcement mechanisms. Its message — the thing it is actually doing to us — is something else entirely. It is teaching the public a posture. It is training us to feel that artificial intelligence is a thing being handled, a situation under management, a problem for which the responsible adults are drafting the paperwork. That posture arrives whether or not any particular rule is wise, and it arrives whether or not anyone understands what the rules are about.

Governing by the Vocabulary of the Dead

The risk-tier framework did not appear from nowhere. It was reached for, and the direction of the reach is revealing. When societies confront something they have not named, they do not sit in silence. They grab the nearest familiar category and press the new thing into it.

McLuhan called this the rear-view mirror. The idea is simple and unsettling. When a genuinely new medium arrives, we cannot perceive it directly. We see it only through the frame of what came before. In The Medium Is the Massage, he wrote that “we look at the present through a rear-view mirror” and “march backwards into the future.” The new environment is invisible to us; we experience it dressed in the costume of the last one. The automobile was a “horseless carriage.” The radio was “wireless telegraphy.” We name the future in the vocabulary of the dead.

Watch the rear-view mirror at work in AI regulation. The “risk-based” approach is borrowed wholesale from pharmaceuticals and finance. In those domains the metaphor holds, because a drug has a defined compound and a known interaction with the body, and a financial instrument has a contract and a counterparty. Risk there is calculable because the object is known. But when the same framework is applied to a system whose behavior emerges from statistical relationships no engineer can fully specify — a system that surprises its own makers — the metaphor is not describing the thing. It is disguising the fact that we do not yet know the thing.

Other rear-view framings crowd the discourse. Some reach for broadcasting law, treating large AI systems as if they were television networks distributing content that must be moderated. Others reach for product liability, treating a model as a defective toaster whose manufacturer can be sued. Still others reach for data protection, treating the whole phenomenon as a privacy problem with familiar contours. Each of these is a mirror. Each shows us the road already traveled and calls it the road ahead.

The rear-view mirror is not stupidity. It is how perception works under conditions of rapid change. But it carries a specific danger. The familiar category feels like understanding. When a lawmaker says “we will regulate this like we regulate pharmaceuticals,” the sentence sounds like comprehension. It has the cadence of mastery. And that sound is precisely what allows the underlying ignorance to go unexamined. The mirror does not merely fail to show the new thing. It actively persuades us that we are already looking at it.

What the Frenzy Lets Us Avoid Feeling

There is a stranger and darker mechanism underneath the rear-view mirror, and McLuhan named it too. He argued that every technology is an extension of some human faculty, and that every extension produces a corresponding numbness. Extend the eye with a telephoto lens and the naked eye’s role recedes. Extend the muscle with a wheel and the leg’s labor is amputated. He called this self-amputation — the way a new medium anesthetizes the very sense it enlarges, so that we cannot feel what it is doing to us.

In Understanding Media, McLuhan tied this numbness to shock. The nervous system, overwhelmed by a new extension, protects itself by going numb around the site of the wound. “The person,” he wrote, “was numb or in a state of shock.” The anesthetic is not a malfunction. It is the body’s defense against a transformation too large to feel.

Apply this to the regulation frenzy and something uncomfortable comes into view. Artificial intelligence is a genuine shock to a range of human self-understandings — about what thinking is, about what writing is, about what judgment and authorship and expertise are for. That shock is disorienting. It is the kind of transformation a society cannot easily hold in its hands and examine.

Regulation offers a way not to feel it. The frenzy of rule-making functions as an anesthetic dressed as vigilance. To argue about enforcement agencies and penalty structures is to stay busy, to feel responsible, to occupy the hands and the headlines — all without confronting the deeper vertigo of a phenomenon we have not named. It looks like the opposite of numbness. It looks like maximal alertness. But activity is not the same as feeling, and vigilance about control can be a sophisticated method of not-feeling the transformation underneath.

This is the cruelest turn of the analysis. The regulation discourse presents itself as society waking up to a danger. Read through McLuhan’s account of self-amputation, it may be society administering itself a sedative. The louder the debate about who governs, the more effectively the public is spared the harder work of asking what, exactly, has changed about the ground beneath their feet.

Figure and Ground

To see this clearly requires one more of McLuhan’s tools. He distinguished, borrowing from perceptual psychology, between figure and ground — the object we consciously attend to, and the environment that surrounds it and gives it meaning but escapes our notice. His central complaint about modern life was that we obsess over the figure and remain blind to the ground. We watch the content and miss the medium. We debate the message and ignore the environment the medium has already built around us.

In the AI regulation discourse, the figure is dazzlingly bright. Draft frameworks, agency jurisdictions, compliance timelines, penalty tiers — all sharply lit, all endlessly discussed. The ground is the thing no one is looking at: the absence of a shared, public vocabulary for what artificial intelligence actually is. We are governing the figure while the ground goes undescribed.

This is why the attention ratio matters as evidence and not decoration. When the bulk of public attention flows to control and only a thin remainder to comprehension, the numbers are describing a figure-ground inversion. The society has trained its floodlights on the rules and left the object of the rules in darkness. And because the floodlights are so bright, the darkness is easy to mistake for solid ground.

The consequence is a spectacle McLuhan’s framework predicts with eerie precision: societies drafting elaborate rules for a phenomenon they have not clearly named. This is not hypocrisy. No one is hiding the ball on purpose. It is a structural feature of how a new medium is received. The rules rush in to fill the space where the naming should be — and by filling it, they conceal that the naming never happened.

The Revelation

Here the argument turns, and the turn is the whole point.

The comfortable assumption running through the AI policy world is that regulation and comprehension are allies moving in the same direction, that rule-making is a response to understanding and will be corrected by better understanding over time. First we grasp the thing, the assumption goes, then we govern it, and if our grasp is imperfect the rules can be revised.

McLuhan’s framework exposes this as backwards. The regulation discourse is not waiting for comprehension. It is functioning as a substitute for it. The frenzy of rule-making does not sit atop a foundation of understanding. It stands in the place where understanding should be, occupying that space so completely that its absence goes unnoticed.

This is the mechanism made visible. When a society cannot name a new phenomenon, the naming-gap produces anxiety. That anxiety demands resolution. And rule-making resolves it — not by closing the gap, but by covering it. The rules give the anxiety somewhere to go. They convert the intolerable feeling of not knowing what this is into the tolerable activity of arguing about how to control it. The conversion feels like progress. It is actually a form of forgetting.

Consider what this means concretely. A citizen following the news encounters an unbroken stream of governance stories: this agency proposes this rule, this jurisdiction adopts this framework, this penalty attaches to this violation. From that stream, the citizen infers — reasonably, inevitably — that the object of all this governance must be well understood. Why else would there be so many rules? The volume of regulation-talk is itself taken as evidence that comprehension exists somewhere upstream, in the minds of the experts drafting the frameworks.

But it does not. The experts are reaching for pharmaceutical metaphors and broadcasting analogies precisely because the naming has not been done. The rear-view mirror is running the whole operation. And the citizen, watching the confident machinery of governance, concludes that the ground is solid when the ground was never surveyed.

The regulation is not filling the gap between us and the phenomenon. It is disguising the gap. It is teaching a public to accept the management of something it has never been given the words to see. This is the message of the medium, in McLuhan’s exact sense — not any single rule, but the posture the whole discourse installs. We are being trained to feel governed rather than to understand. And a society that feels governed stops demanding to understand.

What It Costs a Public

The cost falls on the reader as citizen, and it is worth stating plainly.

To be handed rules for the unnamed is to be handed a peculiar kind of powerlessness that feels like protection. The rules imply that someone, somewhere, has done the comprehending on your behalf. You need not do it yourself. You need only trust the framework, follow the tiers, defer to the agency. This is the transfer at the heart of the regulated unknown: comprehension, which is a form of power that belongs to the public, is quietly exchanged for governance, which is a form of power that belongs to institutions.

The exchange is a bad deal for the citizen, and it is a bad deal in a way that compounds. A public that cannot name a technology cannot meaningfully evaluate the rules written for it. If you do not have the words to describe what a large AI model is and does, you cannot tell whether a risk tier is well-calibrated or absurd. You cannot judge whether an enforcement mechanism reaches the real hazard or misses it entirely. You are reduced to trusting the rule-makers — which means the very absence of public vocabulary that the regulation was supposed to remedy becomes the condition of the public’s dependence on the regulators. The gap does not close. It hardens into a permanent asymmetry of understanding between those who govern the technology and those who live under it.

There is also a subtler cost, the one McLuhan’s self-amputation names. A society anesthetized to a genuine transformation does not get a second chance to feel it. The shock passes. The numbness settles into normalcy. By the time the rules are refined and the agencies staffed and the compliance regimes mature, the moment when the public might have asked what is this and what is it doing to us will have closed. The regulation will have done its quiet work of making the unnamed feel handled, and the handling will have made the naming seem unnecessary. This is how a civilization sleepwalks through its own transformation while believing itself vigilant.

A Tool for the Reader

None of this argues against regulation. Some rules for artificial intelligence are surely necessary, and some are surely wise. The argument is narrower and more useful than opposition. It is that regulation without comprehension is not a partial good to be completed later. It is a specific kind of trap — a medium that installs a posture of managed passivity while persuading everyone involved that alertness is at an all-time high.

The reader can carry a single question out of this essay, and it is the question the rule-makers are not asking. When any new framework for AI is proposed — any tier, any agency, any penalty, any “risk-based” architecture — ask first: do we yet have the words to know what we are governing? Not “is this rule strict enough” or “who should enforce it,” but the prior thing. Has this phenomenon been named on its own terms, or is it being pressed into the costume of pharmaceuticals, broadcasting, or finance because no one has done the harder work of describing what it actually is?

This question is a way of pulling the ground into the foreground — of doing, for oneself, McLuhan’s figure-ground reversal. It refuses to let the brightness of the rules blind you to the darkness they surround. It treats the confidence of the governance