Through McLuhan’s Lens

The Frame War

June 21, 2026 | 2702 words

Through McLuhan’s Lens: Who Wins the Right to Name the Machine

This week, a single phrase did more political work than any policy paper. When Anthropic’s leadership described its models as “a country of geniuses in a datacenter,” and when other labs leaned into the language of AI as a “partner” or “collaborator,” they were not describing a product. They were fighting to install a metaphor. Meanwhile, a competing camp — safety researchers, labor advocates, and a chorus of newspaper columnists — pushed back with the older, harder words: tool, and its darker cousin, threat. The week’s discourse was not an argument about what AI does. It was an argument about what AI is. And the contest over that small verb — is — is the most consequential thing happening in AI right now, far more consequential than any individual benchmark or release.

This column treats that contest as its subject. Not the AI. The naming of it.

Marshall McLuhan spent a career insisting that we misjudge new technologies because we fixate on their content and ignore their form. His most famous and most misunderstood line — “the medium is the message” — means something plain when you strip the mystique away. It means the form of a thing shapes us more than the content it carries. A television changes a household’s life whether it shows opera or wrestling. The change comes from the screen’s presence, its temperature, its demands on attention — not from the program. McLuhan laid this out across Understanding Media, and it is the right instrument for this week’s phenomenon. Because the frame war is not a fight over accurate content. It is a fight over which form of perception gets installed in the public mind.

Each contending word — tool, threat, partner, transformation — is itself a medium. Each does something to the person who adopts it, independent of whether it is true. The job of this column is to make visible what these four words are doing to us while we argue about which one is correct.

Four Words, Four Machines

Start with the simplest move. Each frame names the new thing by pointing at an old one. McLuhan called this the rear-view mirror — our habit of seeing the present through an image of the past, the way a driver watches the road behind to navigate the road ahead. “We look at the present through a rear-view mirror,” he wrote in The Medium Is the Massage. “We march backwards into the future.” The trouble is not that the mirror is useless. The trouble is that it conceals. Every backward-facing name hides whatever the new thing does that the old thing never did.

Watch the four frames betray their mirrors.

Tool points backward at the hammer, the spreadsheet, the calculator. It reassures. A tool has no agency; it sits inert until a human picks it up. To call AI a tool is to insist that the human hand remains in charge, that nothing essential has changed since the first sharpened stone. This is comforting, and the comfort is the point. But the mirror conceals something specific: a tool does not generate language, does not negotiate, does not produce outputs its maker cannot predict. The hammer never surprised the carpenter.

Threat points backward at every prior catastrophe — nuclear weapons, pandemics, the runaway machine of science fiction. It is the mirror of the disaster film. It mobilizes fear and, with fear, attention and funding. But it conceals the ordinary: the millions of dull, useful, already-deployed interactions that look nothing like apocalypse.

Partner points backward at the colleague, the collaborator, the trusted human other. This is the mirror Anthropic and others reach for when they speak of “collaboration” and “geniuses.” It flatters the technology and, by extension, flatters the company that built a partner rather than a mere product. But the mirror conceals the asymmetry. A partner can refuse, can leave, can hold you accountable. A product owned by a corporation can do none of these. Calling it a partner smuggles in equality where there is none.

Transformation points backward at the printing press, the steam engine, electricity — the grand ruptures of history. It is the mirror of the textbook chapter titled “The Industrial Revolution.” It feels neutral, even humble: who could argue against history? But it conceals agency entirely. Transformations, in this telling, simply happen, like weather. Nobody decided. Nobody can be held responsible. The word launders human choices into impersonal destiny.

Here is the first thing none of the combatants can see while fighting: they are all driving by the mirror. Not one of the four words describes AI on its own terms. Each describes a familiar past object and asks the public to feel about AI the way it already feels about that object. The fight looks like a fight over the future. It is a fight over which piece of the past gets bolted onto the windshield.

The Temperature of a Verdict

McLuhan’s second instrument cuts deeper. He divided media into hot and cool. A hot medium delivers a finished, high-definition message that asks little of you — you receive it and you are done. A cool medium is low-definition; it leaves gaps, and you must fill them in. Film is hot; a cartoon is cool. A lecture is hot; a seminar is cool. The distinction matters because it tells you who is doing the work — the medium, or the person.

Apply the thermometer to the four frames and they sort themselves cleanly.

Threat runs hot. It arrives as a completed verdict: AI is dangerous, full stop. There is nothing for the citizen to fill in. You receive the alarm and you react. Hot media, McLuhan noted, produce specialists and spectators rather than participants — you watch the experts manage the danger, and your role shrinks to applause or dread.

Transformation runs equally hot. It too delivers a sealed conclusion: this is the next great rupture, inevitable and total. The verdict is complete on arrival. You are not invited to deliberate about whether the transformation should happen, only to brace for it. Both hot frames hand the citizen a finished story and leave no door for participation.

Tool and partner run cooler. Tool invites the question, “to do what?” — it leaves a blank the user must fill. Partner invites the question, “on what terms?” — it implies an ongoing relationship to be negotiated. These frames demand the citizen supply something: a purpose, a boundary, a judgment.

Now ask McLuhan’s most reliable question, the one this column exists to keep asking: who benefits from each temperature?

The hot frames benefit those who want the public in the grandstand. Threat serves the safety-funding ecosystem and the regulatory entrepreneur — the louder the alarm, the larger the mandate. Transformation serves the incumbent who wants change to feel inevitable, because inevitability shuts down the question of whether anyone chose this and might choose otherwise. The data this week underlines how effective inevitability has become as a rhetorical instrument. Surveys of public attitudes show large majorities now expect AI to reshape daily life within a decade — and that expectation arrives detached from any sense that the public had a vote in it. When a Pew Research Center reading finds that a majority of Americans say they feel more concerned than excited about AI in daily life, the threat frame is not describing that concern. It is manufacturing and harvesting it.

The cool frames, by contrast, would seem to serve the citizen, because they demand participation, and participation is power. But here the analysis turns, because the cooler frames carry their own anesthetic.

What the Winning Frame Numbs

McLuhan’s most unsettling idea was that every extension of ourselves is also an amputation. The wheel extends the foot and amputates the leg’s labor. The phone extends the voice across distance and amputates the body’s presence. And crucially, the nervous system protects itself from the shock of each extension by going numb at the point of contact. He called this auto-amputation — self-numbing. We do not feel what the new technology cuts off, precisely because the cutting-off is how we cope with the extension. “The price of eternal vigilance is indifference,” he wrote in Understanding Media. We grow indifferent to exactly what we ought to watch.

Each frame, then, is not just a description. It is an instruction about what to feel and what to stop feeling.

The tool frame extends the sense of human mastery and amputates the awareness that the tool now produces outputs no human fully authored. When a citizen accepts that AI is “just a tool,” that citizen stops feeling the strangeness of a system generating arguments, images, and decisions at scale. The numbness is the comfort. And the comfort is sold. This week’s data shows just how widely the tool frame has penetrated working life: adoption figures continue climbing, with Pew Research Center reporting that a substantial and growing share of U.S. workers have used AI chatbots for tasks on the job. Every one of those uses is conducted under the tool frame — pick it up, get the output, move on — and every one quietly normalizes a faculty being extended and a vigilance being amputated. The worker does not feel deskilled. That is precisely what the numbness accomplishes.

The partner frame is subtler and, for the reader, more dangerous. It extends the sense of companionship and amputates the awareness of ownership. If the system is your partner, you forget to ask who owns your partner, who sets its objectives, who profits from the relationship, and who can alter or terminate it without your consent. The word partner is doing the work of obscuring an employment relationship in which the citizen is not the employer. This is the frame the labs have the most commercial reason to win, and the temperature analysis explains why: partner runs cool, it invites the user to fill in trust and intimacy, and the trust the user supplies becomes the product’s moat. The citizen does the emotional labor of believing in the partnership, and that belief is monetized.

This is the figure-and-ground move that McLuhan prized, and that serves the reader directly. The figure is what each frame puts in the foreground — mastery, danger, companionship, destiny. The ground is everything the frame pushes into the invisible background — ownership, labor, asymmetry, choice. The whole art of the frame war is the management of grounds. Whoever wins the frame decides what the public will not see.

The Revelation: The War Is the Medium

Now the build pays off. Step back from the four words and look at the activity of fighting over them. McLuhan’s master insight applies not only to AI but to the discourse about AI. The medium is the message — which means the form of the frame war shapes its participants more than any frame’s content.

And the form of the frame war is this: a forced, permanent, high-speed contest in which every public statement about AI must declare allegiance to one of four words. That is the real medium. And it is doing three things to its participants that none of them can see while inside it.

First, it has eliminated the option of not naming. The frame war makes a neutral, descriptive, wait-and-see stance impossible. You cannot enter the conversation without picking a word, and picking a word conscripts you into a camp. The citizen who simply wants to understand AI on its own terms — slowly, provisionally, without metaphor — finds there is no seat at that table. The medium has already decided that everyone must be a partisan. This is the McLuhan point in its purest form: the content of your position matters less than the fact that the medium has structured the room so that only positions, never patient observation, can be expressed.

Second, it has accelerated everyone into reaction. A war demands speed. You must answer the other camp’s frame before it sets. McLuhan observed that electric-speed environments collapse the time for reflection — “we have to consider all the consequences instantly,” he noted in Understanding Media, and the consequence of instantaneity is that judgment is replaced by reflex. The participants in the frame war believe they are thinking. The medium has them reacting. The hot frames win in this environment not because they are truer but because heat travels faster than nuance. A finished verdict outruns an open question every time.

Third, and most importantly, the frame war has changed who gets to speak. This is the revelation the data supports most directly. When the public’s dominant emotional register toward AI is concern rather than excitement, and when that concern is the very fuel the threat and transformation frames burn, the conversation tilts toward whoever can generate the most heat. And heat is expensive. It requires platforms, megaphones, research budgets, and communications teams. The four words are not competing on equal footing. They are competing on a battlefield whose terrain — speed, scale, emotional intensity — was built by and for the largest players. The citizen does not get to coin a fifth word. The citizen gets to choose among four words already on the menu, each placed there by an institution with something to gain.

So the combatants think they are arguing about AI. They are being shaped by the act of arguing. “Winning the frame” is itself the medium, and the medium has already done its work before any frame wins: it has trained the public to receive AI as a thing to be labeled and reacted to rather than observed and governed. The deepest amputation of the frame war is not what any single frame cuts off. It is what the war as a whole amputates — the public’s capacity to relate to a new technology without a metaphor, slowly, on terms it sets itself.

This is why the question “which frame is true?” is a trap. It keeps the citizen inside the medium, busily adjudicating content while the form does the governing. McLuhan would point out that the people most certain they have chosen the right word are the most thoroughly captured, because their certainty is the proof that the medium has succeeded in making them a partisan rather than an observer.

Thinking One Level Above the War

What does this mean for the ordinary citizen — not the faculty member, not the policymaker, but the person trying to think clearly while four words compete for the inside of their head?

It means the first act of intelligence is to decline the menu. When a headline, an executive, or a columnist hands you one of the four words, the reader’s move is not to argue for a better word. It is to ask the figure-and-ground question: what is this word putting in the foreground, and what is it pushing into the dark? When a lab calls its product a partner, ask what the word partner conceals about ownership. When a columnist calls it a threat, ask what the alarm conceals about the boring, useful, already-present reality. The frame’s power dies the moment you can name what it hides.

The second act is to check the temperature of your own reaction. If a statement about AI delivered you a finished verdict and asked nothing of you, you were handed a hot medium, and you should ask who benefits from your being a spectator. If a statement left you a gap to fill, you were handed a cool one, and you should ask what belief or trust or labor you are being invited to supply, and to whom it accrues.

The third act is the hardest, because the medium is built to prevent it: refuse the speed. The frame war runs at electric pace and rewards the fastest reaction. The citizen’s defense is the unfashionable one of slowness — the willingness to say “I don’t yet have a word for this,” and to keep observing the thing itself rather than the metaphor draped over it. This is not indecision. It is the only stance the medium cannot easily mon