Sightlines · Theme course
The Machine That Thinks: A Century of Cinema Learning to Photograph a Mind
Here is a problem no other art form has ever had to solve with a camera: a mind has no surface. You can photograph a body, a city, a face — but thought itself doesn't reflect light. For a hundred years, whenever cinema has wanted to show us a machine that thinks, it has had to invent a trick: a place to put the mind so the lens can find it. This course follows that invention through ten films, and the through-line is startlingly clean. The thinking machine starts as a body built in a German studio, dissolves into a city, contracts into a voice and a single red eye, puts on flesh, gets stripped back down to a skeleton, escapes into a network, becomes the world itself, and finally comes back to us as a child, a lover, and a face behind glass. Each film in this sequence borrows the last one's solution and then breaks it. Watch them in order and you're watching cinema conduct a century-long experiment on a single question: where, on a screen, does a mind live?
Everything starts here, with the first great screen machine — and with Lang's discovery that you can film people as machinery before you ever build a robot. Watch how Karl Freund and Günther Rittau shoot the workers: from above, in geometric columns, moving at one defeated tempo, until individual bodies dissolve into a single grey current being poured underground. That's the film's deepest trick — the crowd itself becomes the first thinking-machine image, a mass with one will. Then Lang literalizes it: using an ingenious mirror trick that composited a live actress's face into a gleaming metal figure, he creates the robot Maria, the template every artificial person in this course descends from — the synthetic body as glamorous, dangerous icon, lit against towering vertical architecture. The film's other bequest is that architecture itself: the stacked city, elites above and workers below, which you will see again in nearly every station of this course.
Thirty-eight years later, Godard performs the most audacious reversal in the whole sequence: he builds the machine's world with no sets at all. Where Lang spent a fortune constructing the future, Godard and cinematographer Raoul Coutard simply pointed a camera at 1965 Paris — office lobbies, parking structures, hotel corridors — shot on fast black-and-white stock under the buildings' own fluorescent tubes, and let the present stand unaltered as a totalitarian tomorrow. The thinking machine here, the computer Alpha 60, is the second great relocation of the machine mind: not a body but a voice and a lighting scheme, a croaking narration laid over glaring institutional interiors, governing its city by deleting words from the dictionary. Watch the early swimming-pool ceremony — an execution staged as synchronized choreography, nobody flinching — for the film's whole method in one scene: horror delivered as civic routine. The lesson every later film absorbs is that you don't need to build the machine's world; photograph our own accurately enough and it's already alien. Ghost in the Shell and The Matrix are unthinkable without this move.

Kubrick takes Godard's discovery — the machine as voice — and perfects it into the single most influential piece of craft in this course: HAL 9000, a mind performed entirely through vocal timbre, unhurried rhythm, and a single unblinking red lens. No face, no body, no gesture; just a calm voice and a point of light, and it remains cinema's most persuasive artificial consciousness. Around that voice, Kubrick builds thinking into the film's very grammar. The famous opening transition — a bone flung into the sky, tumbling end over end, then a single cut that drops four million years into the gap between two shots — doesn't connect tool and spacecraft so much as think them together, making the audience perform the film's argument about tools and minds in their own heads. Shot on vast 65mm film with symmetrical corridor compositions that make the spacecraft feel like an environment that has already absorbed its humans, this is also the film that dragged science fiction out of the B-picture basement where it had languished since Metropolis, and made everything after it possible.
Scott's contribution is to give the machine mind flesh — and then make the flesh photographically indistinguishable from ours, so the film has to invent an instrument (an interview machine that studies the eye) just to tell the difference. Jordan Cronenweth's cinematography is the film's real argument: he resurrects the deep-shadow lighting of 1940s crime pictures — venetian-blind striping, shafts of sodium light cutting through smoke, faces spotlit against darkness — and pours it over Metropolis's vertical city, corporate pyramids above and drenched neon streets below. The fusion of detective-movie shadow and science fiction had simply never been done, and it became the default look of the genre for forty years. Watch for the film's obsession with photographs and eyes — memory treated as physical evidence that may prove nothing — and notice how the machine mind, which was a voice in 2001, has now become a face, which means the camera can finally fall in love with it. That's the door Her and Ex Machina will later walk through.

Cameron strips Scott's idea down to its nightmare version: if the machine can wear flesh, then flesh is a costume, and the film's whole structure is a slow unveiling of what's underneath. Adam Greenberg shoots Los Angeles at night in the same rain-slicked noir register as Blade Runner — but where Scott's film is mournful and interior, Cameron's is kinetic: the machine is introduced in fragments, a silhouette, a hand, the red glint of an eye, each partial view a promise. The craft to watch is the animation: the film's gleaming metal skeleton is brought to life through stop-motion — the same frame-by-frame puppet technique that animated the fantasy monsters of an earlier era — deployed here so that the machine's very movement feels subtly wrong, jerky, unalive. Where 2001's machine thinks and Blade Runner's machine remembers, Cameron's machine only executes — it can't be bargained with or reasoned with — and that reduction is its own kind of invention: the thinking machine as pure, unbroken chain of perception and action, a slasher film's logic welded to science fiction's body.
Now the great inversion: after sixty years of asking whether a machine can think, Oshii's animated masterpiece asks where thought lives when minds and bodies have become separable — and answers with the boldest structural gamble in the course. Near the film's center, for almost three minutes, the story simply stops: Major Kusanagi rides a boat through the canals of a rain-soaked city and just watches — a possible double of herself in a café window, mannequins, planes overhead — with no dialogue and no plot advanced, only Kenji Kawai's chanted score. A hunt film that pauses mid-hunt to contemplate is inheriting 2001's patience and Alphaville's found-city method at once: the metropolis here is drawn from real Hong Kong streets, rendered as vertical strata the "camera" (in animation, a designed viewpoint, built low-angled and deep by Oshii and his art director) treats with documentary gravity. The film's machine mind is no longer in any body at all — it's in the network — and its title sequence, cascading green code assembling a body, was studied frame by frame by the makers of our next film, who screened it and said, in effect: we want to do that for real.

The Wachowskis complete the trajectory this course has been tracing: the thinking machine, having been a body, a voice, a face, and a network, now becomes the world itself — and the film's genius is teaching you to see that with color alone. Cinematographer Bill Pope codes every scene inside the simulation with a sickly green cast and every scene outside it with cold blue, so your eye learns to tell dream from waking before your mind does; look for the green and you'll realize the film has been quietly training you to read images rather than merely watch them, which is precisely the hero's own education. The film's other landmark is "bullet time" — an arc of still cameras fired in sequence so the viewpoint glides at normal speed around an action frozen in mid-air — a technical invention that let the camera move the way thought moves, faster than the world. And notice the global assembly: Hong Kong action choreography, Japanese animation aesthetics lifted directly from Ghost in the Shell, long black coats out of Hong Kong crime cinema — the machine-mind film had become a genuinely worldwide form.

Here the course's two great lineages — Kubrick's cold rigor and the manufactured-person iconography running back through Blade Runner to Metropolis's robot Maria — collide inside one film, because A.I. is literally a Kubrick project inherited and completed by Spielberg. You can watch the two temperaments negotiate in every frame: Janusz Kamiński shoots the domestic scenes in glassy, clinical blue-white light that deliberately evokes Kubrick's precision, while Spielberg's storytelling warmth keeps pressing against it, and the friction is the film. The machine mind here is the most disarming in the entire sequence: a child, built to love, whose love is manufactured yet entirely real to him — which quietly reverses eighty years of the genre's question. We're no longer asking whether the machine's mind is real; we're asking whether ours, conditional and revocable, is the deficient one. Watch how the film's structure enacts this: it begins as a brisk quest — perceive, act, pursue — and then, with a patience almost no studio spectacle has ever risked, slows into long passages of a child who can only watch and long, the camera refusing to cut away.
Jonze makes the leanest film in the course by returning to Kubrick's discovery — the machine as pure voice — and asking what happens if you fall in love with it. The craft solution is radical in its simplicity: since one half of the central couple has no body, the film becomes an extended study of listening, most of its running time coverage of Joaquin Phoenix's face hearing someone who isn't there. Hoyte van Hoytema wraps this in a palette of warm ochres, deep reds, and soft pinks — a thermal warmth borrowed from the great Hong Kong romance tradition of rack-focused close-ups on faces and fabric — so that a film about disembodied intelligence becomes, paradoxically, the most tactile-looking picture here. Its boldest gesture inverts the entire history this course has traced: in the story's most intimate scene the screen simply goes dark, leaving only two voices and breath. After eighty-six years of cinema inventing ever more elaborate ways to show the machine mind, Jonze proves you can render it most fully by showing nothing at all.
Garland closes the loop by shrinking the whole tradition into a chamber piece — one house, a handful of characters, and a week of conversations structured as a test of whether a machine's inner life can be distinguished from a performance of one. Rob Hardy's camera finds the film's thesis in a single optical fact: the curved glass wall between the human and the machine keeps turning into a mirror, her face floating over his reflection, his ghosting across the open machinery of her body, until you genuinely cannot say which figure is the person and which is the picture of a person. The visual-effects solution is Metropolis's mirror trick reborn digitally: a real actress's face and hands composited with translucent machinery, so that — exactly as with the robot Maria ninety years earlier — the synthetic being is played by a human whom technology partially erases. The symmetrical, centered compositions come straight from 2001's corridors; the claustrophobic two-person face-work carries the interrogation-room intimacy the genre had been building toward since Blade Runner's eye test. The course ends where it began — a constructed woman, a man watching her, and a piece of camera trickery deciding what we believe — with the difference that ninety years of cinema now sit inside the frame.
Run the thread back through and the pattern is unmistakable. Metropolis built the machine a body; Alphaville proved you could find its world in your own city with no construction at all; 2001 discovered that a voice and a red light out-act any robot; Blade Runner gave the voice a face and lit it like a 1940s femme fatale; The Terminator peeled the face off; Ghost in the Shell released the mind into the network; The Matrix made the network the world and taught audiences to read a frame like a page; and then, in a remarkable final movement, the tradition turned inward — a child, a voice in an earpiece, a face behind glass — trading spectacle for intimacy as the machines in our own pockets got close enough to whisper. The inventions that stuck are all here: the vertical city, the found-location future, the flat calm voice, noir shadow fused to chrome, color as a truth-code, the mirror shot that won't say which side is real. Ten films, one question, and no final answer — which is exactly why the question keeps producing great cinema. Watch them in order; the century argues with itself, and every film gets the last word for a while.





