On a recent garbage collection day, I walked out of my front door to a peculiar sight. My son is 4, so the symphony of the garbage truck—clang, pssh, beep—beckons like the Pied Piper. Often, we run out in the bare nick of time, seeing the maw of its tail end driving away. But on this day, the truck was trapped by no less than three hapless Waymos, self-driving cars that have cruised San Francisco and fourteen other test cities for years. They were in a tragicomic faceoff with a refuse collection vehicle and an irritated sanitation worker. Of course, I walked out to snap a picture on my smartphone, and the irate driver told me it was the second time that morning. As we watched, a fourth Waymo—this one a little wilier—turned the corner into the tangle of autonomous vehicles, then immediately backed up and slunk away.
The robotaxis are everywhere. Whether I’m walking, riding my bike, or driving, not five minutes goes by without seeing a Waymo or Zoox, autonomous vehicles deployed by Alphabet and Amazon respectively. They prowl our streets, taking pronounced pauses at stop signs. Mostly, they seem cautious, though one did cut me off the other day (they do learn from humans, after all). Near hubs, they flock like little birds returning home. And as they do, bot traffic chokes human activity. A recent study concluded that between August 2023 and December 2025, 46% of miles driven by Waymo had no one inside the car. In so-called “edge cases,” Waymos spiral in confusion; mishaps have seen the cars careening through construction zones, and in December 2025, a blackout froze 1,600 Waymos for two minutes, snarling roads.
We often think of AI as ethereal spirits in code that haunt data centers and traverse the cloud, like the chatbots quietly surveilling us as they dutifully respond to our queries. Pundits talk about embodied AI as the next chapter, as if algorithms weren’t already guiding how we move, sense, and feel. But these car-shaped robots are physical agents of artificial intelligence. They don’t just live in data centers. They occupy sidewalks, streets, and intersections. Algorithms might seem intangible, but they have palpable consequences for human bodies moving through the built environment.
Transportation technologies have long been partners in these urban dances. But in smart cities, the machines that dance have started calling the steps. A traffic light and a dispatch algorithm run different routines. A traffic light issues the same instruction to everybody, at a fixed cadence. It can be seen, predicted, even refused, as jaywalkers often do. But the algorithm issues a different instruction to each body, an invisible protocol that changes based on what the last body did. Both are infrastructure, but one learns. And as algorithms continue to permeate human life, they contribute to what some call the urban ballet—though scrum might be just as appropriate a word.
Bicycles, trams, trains, and cars each carve channels across our environments. Through rails and roads, the city bends to what these vehicles demand. With their greed for space, the colonization of walkways by freeways, cars have remade landscapes to the scale of machines. And what steers the car also steers the human body.
Veined by streets, tramways, and railroads, modern cities cultivated new movement patterns for the human bodies navigating their landscapes. And this, in turn, inspired new kinds of aesthetic experiences. Early twentieth century films such as Walter Ruttmann’s Berlin: Symphony of a Great City (1927) and Dziga Vertov’s Man With a Movie Camera (1929) documented cities as architectural stages where humans danced with industrial machines. These films created impressionistic montages of metropolitan life–windows opening, trains docking, and streams of people commuting. Guided by sidewalks, shunted by public transportation, paced by traffic lights, citizens can find their bodies–and minds–automated by the landscape.
Yet as urban architecture orchestrated some movement, in the modern city some walkers have refused these routines. In her 1930 essay “Street Haunting: A London Adventure,” under the ruse of going on an errand to buy a pencil, Virginia Woolf rambles across the city as a matter of utmost importance: pleasure and liberation. Spurning ends, walking–or rather, strolling–through the urban environment becomes an end unto itself. A renegade stroller, she rejects the automatic choreography of the modern city, instead guided by the environment–a living, mutable cauldron of flesh, glass, concrete, and steel.
Charles Baudelaire called this literary type the flâneur, often an independently wealthy dilettante, who could meander the city at leisure rather than be rushed around it at the pace of capital. Strolling on unearned time, the flâneur saw what the hurried missed.
The Situationist International, an avant-garde collective, applied this kind of stroll as a form of resistance to consumer capitalism—an antidote to the drudgery of everyday existence. They understood cities as engines of automation that press human bodies into the metronome of production. By their account, most city dwellers were sleepwalking through existence. As a counterstrategy, they practiced the dérive, or drift: a “technique of locomotion without a goal.” Taking arbitrary routes and responding playfully to the urban environment, they reclaimed whimsy, and with it, their humanity.
Some still do. On walks through our neighborhood, my young son skips and ambles. He rushes then doubles back. He bounces gleefully on dumped mattresses. At just over four feet tall, his keen eye ranges the pavement, hunting for treasure. He stops to gather petals and pods shed by street trees, tucking them into his pockets. Unburdened by optimization, his city is a field of discovery.
Walking may be the most natural thing in the world, but it is increasingly being integrated into technological systems. Since my smartphone entered my life, I have found that it has taken me much longer to discover a place anew, to learn it by foot, to know it in my bones. I pay less attention to the world around me, aligning my feet with the path that tells me it will take four, fourteen, or twenty-five minutes to walk to my destination at the average walking pace–whatever that is. Serendipity eludes me; I might miss the daily mysteries that adorn a street, curious children, architectural oddities, or ground scores.
Instead, my daily performances—buying a coffee, crossing the street, getting the bus—are increasingly networked into computational systems. I wait for the green man. I tap my phone to pay at the café or to enter the train station. Machines script my way.
In Remote X, German theatre collective Rimini Protokoll stages a series of site-specific performances orchestrated by such a machine. A synthetic voice, its uncanny timbre familiar from our own GPS navigations, conducts a group of fifty headphone-wearing people through collective urban performances. The voice tells them to cross streets, race for short bursts, and notice parts of the city. Each performance is unique to the city it’s staged in, and soundtracked by film scores and binaural recordings that give the experience the texture of a movie.
In this encounter with AI, people take their cues from the synthetic voice, but also from each other, in a larger collective choreography. From Berlin to Bengaluru, Sharjah to Seoul, performers–that is, citizens–move through a hybrid space rooted in geography and mobilized by virtuality. Remote X is both a canny enactment of the split embodiment of perpetually networked life, as well as an observation on the ways that algorithms transform the lived experience of space.
First staged in 2013, Remote X anticipated the networked urban ecologies of smart cities, where Amazon packages proliferate, Flocks surveil, and Waymos roam.
In smart cities, the machine becomes both dancer and choreographer. City dashboards orchestrate trains and traffic lights, shunting people and cars along predetermined routes. Uber Eats and DoorDash send human bodies on delivery missions, optimizing their routes and making their bodies extensions of the algorithm. A sanitation worker is trapped in a labyrinth of self-driving cars.
Waymos prophesy a future where robotaxis are the last remaining flâneurs, drifting on our dime with nobody inside.
And what is the counterstrategy for the smart city? In 2023, protestors slapped orange traffic cones onto the hoods of Waymos, immobilizing them. Reading a cone as a hazard, its software would abruptly stop the car. Where the Situationists answered the city with unpredictable human movement, coning exploits the machine’s safety logic, giving it something to obey. After all, you can only push back on a choreography you can see. These cones—witches’ hats, as they call them in Australia, where ritual and technology collide—are humans choreographing the algorithm back.
Inside the car, the choreography can work differently. From the app, a blind user can prompt a Waymo to play a melody so it can be found. Riders can also turn on full narration in their in-car announcements, which transposes a visual journey into an audio one. It is the most thoroughly scripted possible experience of a city, and yet here, this machine voice offers new access to that city. What I learn by walking, another rider learns by listening.
Chasing garbage trucks, my child learns a city that machines are learning at the same time.
