This article is a collaboration between WIRED and Epic Magazine.

Photographs by Cait Oppermann

Love in the Time of Robots

Hiroshi Ishi­guro builds androids. Beautiful, realistic, uncannily convincing human replicas. Academically, he is using them to understand the mechanics of person-to-person interaction. But his true quest is to untangle the ineffable nature of connection itself.

It is summer 2002, mid-morning in a university research lab on the edge of Osaka, Japan. Two girls—both dressed in pale yellow, with child-puffy cheeks, black shoulder-length hair, and bangs—stand opposite each other under fluorescent lights. More precisely: One is a girl, 5 years old; the other is her copy, her android replica. They are the same size, one modeled on the other, and they are meeting for the first time. ¶ The girl stares hard into the eyes of her counterpart; its expression is stern and stiff. It seems to return her gaze. ¶ A man is videotaping the pair—he is the father of one, creator of the other—and from off-camera he asks, “Would you like to say something?” ¶ The girl turns to him, disoriented. She turns back to the android. ¶ “Talk to her!” he says. “Hello.” ¶ The girl repeats the word, quietly, to her robot-self. It nods. ¶ Her father feeds her another line: “Let’s play.”

The android wiggles its head. Her father chuckles behind the camera. But the girl does not budge. She simply stares at her double, the look on her face one of focus and perhaps concern.

Each member of this pair continues making the barely there gestures that serve, through reflex or ruse, as signs of life: Each blinks at regular intervals; each tilts her head from side to side. One is processing, in the raw, sensory-overload manner of a human child; the other is performing a series of simple movements made possible by the servomotors installed inside the silicone casing that is its skin.

“Is it difficult to play with her?” the father asks. His daughter looks to him, then back at the android. Its mouth begins to open and close slightly, like a dying fish. He laughs. “Is she eating something?”

The girl does not respond. She is patient and obedient and listens closely. But something inside is telling her to resist. 

“Do you feel strange?” her father asks. Even he must admit that the robot is not entirely believable.

Eventually, after a few long minutes, the girl’s breathing grows heavier, and she announces, “I am so tired.” Then she bursts into tears.

That night, in a house in the suburbs, her father uploads the footage to his laptop for posterity. His name is Hiroshi Ishi­guro, and he believes this is the first record of a modern-day android.

In the 15 years since, Ishi­guro has produced some 30 androids, most of them female. They have included replicas of a newscaster, an actress, and a fashion model. These androids have made numerous public appearances—in cafés and department stores, singing in malls, performing in a play. Mostly, though, Ishi­guro’s brood of pretty “women” is used for his academic experiments, many of which are conducted at two locations in Japan: the Advanced Telecommuni­cations Research Institute International in Nara and the Intelligent Robotics Laboratory on the campus of Osaka University.

The lab, known as IRL, is embedded within a maze of austere, gray university buildings. In one of these industrial boxes, about 30 students and assistant professors work in a series of near-­silent computer pods and observation rooms. Teams of young men shuffle down the long, linoleum-lined hallways in sweatshirts, pace the research rooms in their socks, or hover over laptops in rows, heads down, subsisting mostly on Red Bull, crackers, and Pocky Sticks. (Women do not seem like a natural fit here. As if to underline this fact, a sign by the rest­rooms reads, “Watch out for male strangers in the ladies toilet.”)

Presiding over this disheveled scene is ­Ishi­guro-sensei. He is immediately recognizable, looking just as he does in promotional photos from recent years: perfectly mod in slim-­fitting black with matching leather backpack and fanny pack. He wears tinted hexagonal glasses and styles his jet-black hair into a mop top that swoops across his forehead. This is his department: Ishi­guro, 54, is a distinguished professor at one of the country’s top universities, with two labs, partnerships with a dozen private companies throughout Japan, a recent $16 million grant from the government (one of its most generous in science and engineering, he says), and seven secretaries to manage it all.

Subscribe to WIRED. November 2017.

Today, the technical ability to produce a robot that truly looks and moves and speaks like a human remains well beyond our reach. Even further beyond our grasp is the capacity to imbue such a machine with humanness—that ineffable presence the Japanese call sonzai-kan. Because to re-create human presence we need to know more about ourselves than we do—about the accumulation of cues and micromovements that trigger our empathy, put us at ease, and earn our trust. Someday we may crack the problem of creating artificial general intelligence—a machine brain that can intuitively perform any human intellectual task—but why would we choose to interact with it?

Ishi­guro believes that since we’re hardwired to interact with and place our faith in humans, the more humanlike we can make a robot appear, the more open we’ll be to sharing our lives with it. Toward this end, his teams are pioneering a young field of research called human-robot interaction.

HRI is a hybrid discipline: part engineering, part AI, part social psychology and cognitive science. The aim is to analyze and cultivate our evolving relationship with robots. HRI seeks to understand why and when we’re willing to interact with, and maybe even feel affection for, a machine. And with each android he produces, Ishiguro believes he is moving closer to building that trust.

In a secluded room at IRL, a collection of androids is stored and maintained: his hardest workers. Arranged in this space today, with its blackout curtains, thin corporate carpeting, and shelves cluttered with cables and monitors and an array of wigs, is a pair of his replicas of grown women. They are models of the Geminoid F series. The name is a play off geminus (Latin for “twin”), a reminder that their human counterparts exist somewhere in the world.

At any given time, students and staff may be testing, measuring, and recording the responses of dozens of volunteers to the androids at their disposal. What about its behavior or appearance, its specific facial expressions and minute body movements, do they find alienating? What draws them closer? These androids are used to find answers to an ever-growing list of research questions: How important is nonverbal communication to establishing trust between humans (and, therefore, between human and android)? Under what circumstances might we treat an android like a human? In this way, Ishi­guro’s collective of labs is dedicated to the engineering of human intimacy.


SCROLL DOWN

“A conversation is a kind of illusion,” Ishiguro says. “I don’t know what is going on in your brain. All I can know is what I’m thinking.”

Over the several months we are in contact, Ishi­guro will share information that strikes me as deeply personal: He has contemplated suicide twice in his life; though he has a family, he considers himself a lonely man. I will hear him use that word to describe himself—lonely—about half a dozen times.

As for me, when I first visit Ishi­guro, my situation is this:

I am 23 months away from what had seemed like the start of a serious relationship but was not. I am 15 months away from a rebound relationship that lingered too long. I am 13 months into a period of spending long stints in a small town in upstate New York for the sake of productive quiet. I’m readying a book to go to the printers—work that, for me, is all-consuming and necessary. And lately, when I step back from the manuscript for an afternoon or at night, I feel it: isolation. This isolation is not complete—I have my close friends, a wider circle of less-close friends, my family—but it is the absence of intimacy. Nothing romantic, no sexual life.

This absence has been, in part, a choice; certain men have always been curious about me. But what I miss more than sex is the feeling of closeness with another person, something I’ve never believed could be conjured up. And though the sensory deprivation has become a little extreme, most of the time—can I put a percentage on it? Is it as high as 80 percent?—I do not think about it. I am semi-radically independent and some kind of artist and in many ways an unconventional liberal woman. However alienating, for me this is a time of deep creativity. It’s that additional 20 percent of the time—that’s when I feel dizzy.

This is where I’m at when I fly 17 hours to meet Ishi­guro. And as a result, if I am honest with myself, my time abroad feels particularly fraught. The very concept of “human connection” has never felt so enigmatic to me. It makes sense that someone would be trying to measure it, to weigh it, to calculate its dimensions. To be able to replicate the sensation of human intimacy would be to control the very thing that confuses us most and eludes so many.

This is how Ishi­guro remembers his childhood:

His family lives in the town of Adogawa, on the western shore of Lake Biwa, from which a river flows through Kyoto into Osaka Bay. At school, in a classroom of disciplined children, Hiroshi doesn’t listen to the instructor. It’s as if he doesn’t notice she is speaking at all. He spends the day making drawings that have nothing to do with the lesson. His mother worries that there may be something wrong with him.

Hiroshi rarely sees his mother or father—as schoolteachers they are as consumed by their work as their son will one day become. Instead, his grandparents are raising him. His mother’s father is a farmer, a devout Buddhist with fixed, traditional ideas about “how to behave like a Japanese man.” He shows the boy the proper way to use chopsticks, to pray, to prepare the house for the New Year’s celebration. Unlike at school, Hiroshi has the patience for these lessons: His grandfather is not telling him how he should think; he is teaching him to aspire to perfection.

They live at the foot of the Hira Mountains, and Hiroshi likes to comb the mountainside for snakes and insects. Maybe a stag beetle, glossy black and segmented, nearly 3 inches long, with a pair of antler-shaped mandibles emerging straight from its head. He fixes new parts to its body: razor blades, found pieces of metal. It is an improvement. The insect may continue living like this, if the glue doesn’t kill it. These are his earliest cyborgs.

One of Hiroshi’s close friends is a boy who lives in a poorer community, down by the water, and his parents collect and prepare the bodies of the recently dead for burial. Hiroshi does not yet understand that these people are considered to be lesser than his family, because they have a job that, according to local prejudice, is tainted. For this reason, when Hiroshi’s mother discovers the friendship, she asks that her son break it off. He will remember this moment for the next 40 years.

Hiroshi is a delicate child. He has suffered from extreme skin allergies from the time he was born; his back and chest and arms are covered in itchy, ugly rashes. His only comfort comes from constant touch: Every night, his grandparents take turns sitting beside him and scratching his back until he is able to nod off. Every week his doctor gives him three painful injections to try to cure the condition—to no effect. (When he is about 12, steroids will finally help, requiring him to keep the drug on hand to this day.) His own body will always be alien to him.

Human emotions, for Ishiguro, are nothing more than responses to stimuli and are thus subject to manipulation.

When it comes time for Ishi­guro to go to college, he chooses a school using three criteria: It will accept an eccentric, sometimes indifferent student like himself; it’s somewhere he can pursue his drawing and painting; and it’s not very close to home. In the fall of 1981, he lands at the University of Yamanashi, near Mount Fuji.

Once there, Ishi­guro continues his careless approach to his studies, finding more pleasure in the string of odd jobs he takes to pay the bills—he works as a cook, the supervisor of a children’s after-school program, a door-to-door textbook salesman (that one lasts a week), and, most lucrative of all, a professional pachinko player. He finds himself on the fringes of student life, rejecting any semblance of mainstream Japanese ambition.

At the same time, he is fashioning himself into that most romantic of outsiders: an artist. Always in a black leather jacket, he skips classes, packs his pads and pencils, and rides his Yamaha chopper into the nearby countryside to sketch the landscape. This is his focus: the strange, organic shapes of the trees, the peach blossoms that appear in the springtime. He produces drawings and oil paintings, and manages to sell a few.

“Risa, you’re totally fine … If you’re feeling sleepy, if your head feels heavy, you can just lean back. Just like sleeping …”

But in his third year, Ishi­guro abruptly gives up painting. Unless he can become a great artist and a tremendous public success, he sees no point in it. (He blames, in part, his color blindness: He is drawn to landscapes, but the entire spectrum of green eludes him.) He has lost what little direction he had. On his darker days, when he takes his motorcycle on a steep and winding road, Ishi­guro imagines giving in to the impulse to not make the turn. To drive straight ahead, fly right off the edge—what would that feel like?

Then a path presents itself. Yamanashi offers courses in the new field of computer science, and Ishi­guro begins to wonder what relationship computer graphics and computer vision might have to the visual arts. These are the early days of the PC, and programming seems wildly creative. Feeling he has little to lose, he switches majors.

Almost immediately, certain elements in his brain click into place: Ishi­guro realizes he can continue to think like a painter in this unpoliced field, but with different tools. He falls in love with the new vocabulary: Assembler, Pascal. The students are relegated to working in a ­single room kept bitterly cold, loud with the hum of the huge computers—conditions designed for the comfort of machines, not humans. He works alone, on software development, but he is learning to communicate with a system—a system that responds to his commands. They have entered into a dialog.

Ishi­guro soon gives up his rides through the country for entire days spent in the lab. And as he becomes more fluent in this new language, more immersed in a conversation with the large machines, a fantasy takes shape: Could there be a way to make this language more humanlike, so that someday computers might understand us intuitively, on our own terms? So that this dialog might become a relationship?

This relationship becomes his singular pursuit, his dream.

Its hands, at rest on its lap, are rubbery to the touch. Lean in close and you can hear the soft hum of a hidden motor; a gentle click is audible each time it blinks.

In 2000, Ishi­guro, as an associate professor at Kyoto University, produces his first humanoid robot: a mechanical-looking contraption that moves on a wheeled platform, waving its jointed steel arms. But he has started to think that a relatable, humanlike appearance is essential if people are going to form real attachments to robots.

It’s about a decade into his marriage (to a pianist he met through a university friend), and he asks his wife if he can make videotapes of her—sitting, breathing, responding to random stimuli. He is trying to determine the nuances of human behavior, to isolate the physical signs that read to us, consciously or unconsciously, as “human.” One minor revelation: Humans never truly sit still.

Ishi­guro is aware of resistance to the concept of an android—at least in the West, from which many Japanese researchers take their cue. Some are worried that consumer revulsion to a humanlike robot (the so-called uncanny valley effect) would be too great to overcome and that a failed android project could undercut public support of robotics. Ishi­guro, too, is worried that pushing ahead with an untraditional approach might cost him his academic career. But he can’t resist. And so when the company he has partnered with on a new robot insists on hiring a respected designer that makes it look, in Ishiguro’s opinion, “like an insect,” he loses his patience. With his next project, he decides to go rogue. He will create an android “to convince them.”

Ishi­guro believes that his first android should be the same height as the insect (about 3½ feet tall), for purposes of comparison. In other words, it will have to be modeled after a human child. And given the painstaking production process—a model must spend hours encased in plaster to cast an accurate replica—there is only one child he can possibly get permission to use: his own.

A few years earlier, Ishi­guro became a father to a daughter, named Risa, and he now turns to his wife to explain his plan. She agrees—she is in charge of raising the girl, and the experiment would be difficult without her help. And so, in early 2002, the entire family, along with makeup and special effects artists, gathers in his lab on campus and begins the two-day process of creating a replica of Risa.

In the lab, Risa’s mother helps her to undress. She takes off the girl’s clothes and stands her up on a small wooden platform. Together her father and an artist smooth a layer of pale-green paste over her torso and upper thighs; over that, they apply wide swatches of fabric dipped in plaster, asking her to hold very still as it dries. Then the 5-year-old girl, wrapped in a pink towel, her scalp covered in a rubber cap and her ears plugged with cotton, is laid down on a tabletop, her head fenced in with Styrofoam and packing tape. An artist lifts a plastic bucket and pours the paste in until it rises to cover her ears, as father and mother try to reassure her: “Don’t worry!” and “You’re fine!” At last they prepare the girl for the final part of the process: her face.

Through a video camera’s viewfinder, Ishi­guro watches the rigid expression on his small daughter’s face as her mother and an artist slowly cover it in thick paste. “Once we’re done,” her father says, “you can eat anything you like!” They slather it across her forehead, around her chin, and down the front of her neck; they apply it thickly on her cheeks and across her nose, then subsume her entire mouth, her mother laughing, keeping the mood light. “Keep your eyes closed. Like you’re going to bed … Good night!” The whole time, remarkable for a child her age, she does not move or make a sound. And then the paste closes in on her as they smooth it over her eyelids, and within moments her face is layered in the creamy stuff, which has already begun to harden. Her entire face is under—save her nostrils: a single hole left clear for breathing.

“You’re OK,” the artist says. “Just a little bit longer …”

Then Ishi­guro, from behind the camera: “Risa, you’re totally fine … If you’re feeling sleepy, if your head feels heavy, you can just lean back. Just like sleeping …”

They press a square of plaster-soaked fabric over her face (again, a hole for breathing) and it begins to stiffen. And perhaps the professor is now concerned, because he loses the shot, tilting the camera up to point at the wall. “Risa, if you can breathe properly through your nose, please squeeze my hand …”

“Risa,” her mother says, “make sure you don’t cry, because it’ll block your nose. Anyway, there’s no need to cry! Be patient … It’s OK to sleep. Go to sleep …”

When, months later, the package arrives at the lab, Ishi­guro and his team open the crate to reveal the full-body silicone-skin casing of his daughter: Risa, bald, naked, made of rubber. They stretch the skin around foam-padded machinery and prop it up in the lab. His wife has donated one of their daughter’s sundresses so it has something to wear. Ishi­guro names it Repliee R1—R for Risa.

The results of the experiment are mixed. Ishi­guro has to admit that the low-budget android, with its limited, stuttering movements, is more zombie than human. And though he shows the project only to a trusted inner circle, word of the “daughter android” spreads, becoming a weird legend. (In describing it, one roboticist I speak with uses the word “crazy,” another “strange” and “a little bit scary.”) But Repliee R1 gives Ishi­guro the confidence to move forward.

As for his daughter, Ishi­guro rewards her with several Hello Kitty dolls. “But still,” he says, “she cried.” To this day, they’ve never spoken about the incident.

Geminoid F traveled the world performing in a stage play conceived with it in mind. It also played a companion robot in the 2015 film Sayonara.

Three years later, in 2005, Ishiguro unveils Repliee Q1 Expo to the public. Modeled on a grown woman (a popular Tokyo newscaster) and produced with better funding, this version can move its upper body fluidly and lip-synch to recorded speech. Ishi­guro’s lab conducts several studies with it; the results are featured in a major Japanese robotics journal; the lab is filmed for television; he hears about a copycat android in South Korea. As a growing audience is drawn to Ishiguro’s simulated human, his instincts are validated.

But he now wants something more. Twice he has witnessed others have the opportunity, however confusing, to encounter their robot self, and he covets that experience. Besides, his daughter was too young, and the newscaster, though an adult, was, in his words, merely an “ordinary” person: Neither was able to analyze their android encounter like a trained scientist. A true researcher should have his own double. Flashing back to his previous life as a painter, Ishi­guro thinks: This will be another form of self-portrait. He gives the project his initials: Geminoid HI. His mechanical twin.

Ishi­guro has hundreds of photos of the Geminoid’s assembly. Here is his assistant wrapping the facsimile of his then-43-year-old face around the machine head and zipping it up the back, its bald scalp studded with sensors. Here is the Geminoid seated upright, a padded vest in place of its torso, its mechanical biceps visible, its arms only “flesh” below the elbows, as if it were wearing elegant gloves. The hands have veins and sunspots and the faint wrinkles that gather around the wrists; the nails have cuticles, pale and precise. Here it is dressed, in a fitted black shirt identical to Ishi­guro’s. His assistant raises its arms, one by one, to tug down the sleeves, as if dressing a complicated child.

It also wears fitted black slacks, like Ishi­guro’s, and black sneakers stuffed with prosthetic feet in matching socks; a black wig, styled like the hair of its maker, is fastened onto the android’s scalp with snaps. Here is the machine that pumps air into its chest—a series of cables runs from its tailbone into a metal box—as the professor’s double sits at attention and speaks for the first time.

This android is a step forward, but it still falls well short of verisimilitude. Its hands, at rest on its lap, are rubbery to the touch; its eyes have a surprising intensity, not unlike Ishi­guro’s, but they are clearly made of a hard, bright plastic. Lean in close and you can hear the soft hum of a hidden motor; a gentle click is audible each time it blinks. At times, its overall effect, and that of its sisters, is of a human-sized puppet—like the animatronics in a Disney World display. But the Geminoid is also unsettling. Because, somehow, all these elements work in concert to simulate a sympathetic interaction with a human. The viewer cannot help but assign an entire range of emotions to its face: melancholic (mouth downturned), upset (eyes squinted shut), skeptical (a sideways glance), pensive (the tilt of its head to the left). When its eyes meet yours, motion sensors detecting your position, just for a moment you feel that it—this “he,” this “Ishi­guro”—is aware of you.

“Android has my identity,” Ishiguro says. “I need to be identical with my android, otherwise I’m going to lose my identity.”

This replica, Geminoid HI, brings Ishi­guro the recognition he has longed for. Using his double, he and his team publish dozens of studies, analyzing the participants’ range of reactions to him and his doppelgänger. (The studies involve operating the android remotely and wirelessly: tele­operation.) Side by side, he and his Geminoid make appearances on TV shows across Asia and Europe. Ishi­guro also begins giving lectures around the world without leaving his lab in Osaka, teleoperating and speaking through the android, which is carefully transported abroad by an assistant. (Its legs and torso are checked with the luggage; its head is carry-on.) Ishi­guro-sensei becomes a source of fascination; he is transformed from a researcher to the man who made his copy. Invitations for conferences and festivals stream in.

The success of this particular android is due, in part, to how it seems to operate on several levels. It is, like its predecessors, a circus trick: Look at the human, look at his copy! Try to tell them apart! It is also Ishi­guro’s bid at solving an existential dilemma—a striking attempt by the maker to master himself, to make of himself something more enduring.

At the same time, it has created a new predicament. Ishi­guro has discovered unexpected consequences of living alongside his own replica. He’s been dressing in black since his grad-school years, and now this has become both his and the HI’s official uniform; he was thrilled to realize this clearer vision of himself. But now he must keep his (naturally shifting, aging) human body corralled within the android’s static limits. He finds himself accommodating his android, measuring himself against it, being defined by it, his worth determined by it. In this way, his android makes him both painfully conscious of his aging body and more physically confident than he’s ever been.

Ishi­guro is multiple myths simultaneously. With his female androids, he is Pygmalion, bringing his Galatea to life. But with his own replica, he is Narcissus, staring into his reflection for hours. Unlike Narcissus, of course, Ishi­guro is conscious of the situation he has created, but he’s set an unexpected trap for himself through his image. He poses beside his android, in press photos and TV appearances, in ways that accommodate the Geminoid, setting his face to mirror its expression. (At one point at the research institute, Ishi­guro notices me photographing him in front of his android and reflexively drops his smile to match the robot at rest.)

Soon his students begin comparing him to the Geminoid—“Oh, professor, you are getting old,” they tease—and Ishi­guro finds little humor in it. A few years later, at 46, he has another cast of his face made, to reflect his aging, producing a second version of HI. But to repeat this process every few years would be costly and hard on his vanity. Instead, Ishi­guro embraces the logi­cal alternative: to alter his human form to match that of his copy. He opts for a range of cosmetic procedures—laser treatments and the injection of his own blood cells into his face. He also begins watching his diet and lifting weights; he loses about 20 pounds. “I decided not to get old anymore,” says Ishi­guro, whose English is excellent but syntactically imperfect. “Always I am getting younger.”

Remaining twinned with his creation has become a compulsion. “Android has my identity,” he says. “I need to be identical with my android, otherwise I’m going to lose my identity.” I think back to another photo of his first double’s construction: Its robot skull, exposed, is a sickly yellow plastic shell with openings for glassy teeth and eyeballs. When I ask what he was thinking as he watched this replica of his own head being assembled, Ishi­guro says, perhaps only half-joking, “I thought I might have this kind of skull if I removed my face.”

Now he points at me. “Why are you coming here? Because I have created my copy. The work is important; android is important. But you are not interested in myself.”

“A beautiful woman you don’t picture going to the restroom or getting tired,” Ishiguro says. “So I think beauty is better represented by android.”

On a winter day in 2012, a crowd clusters around a large glass case in Tokyo’s Takashimaya department store. Perched inside is a Geminoid F in an elegant silk day dress, long brown bangs parted like curtains around its face. Valentine’s Day is coming soon, and “she” sits, as if waiting for someone, before a backdrop of gift boxes wrapped in rose-patterned paper and large red bows.

She spends her days staring at her smartphone and mostly ignoring the thousands of visitors who press close to the glass. All the while, she goes through a range of facial expressions, a spectrum of subtle emotions, as if reacting to some text she has just received. It’s a clever ploy: By not interacting much with her onlookers, the simulation maintains the appearance of a human likeness—after all, real people spend a lot of time willfully ignoring their surroundings. But occasionally, when you approach, she looks up at you and smiles, and for a moment this feels like an encounter with a pretty stranger.

Some days, Ishi­guro stands across the aisle, by the main entrance, and watches the people who stop in front of her. He likes to imagine what they believe she is thinking.

As complex as we assume ourselves to be, our bonds with one another are often built on very ­little. Given all the time we now spend living through technology, not many of us would notice, at least at first, if the friend we were messaging were replaced by a bot. And humans do not require much to stir up feelings of empathy with another person or creature—even an object. In 2011 a University of Calgary test found that subjects were quick to assign emotions and intentions to a piece of balsa wood operated with a joystick. In other words, we are so hardwired for empathy that our brains are willing to make the leap to humanizing a piece of wood. It’s a level of animal instinct that’s slapstick-hilarious and a degree of vulnerability that’s terrifying.

But as the object of our attention moves closer in appearance to human, our expectations of them grow far more complex. The uncanny valley effect kicks in—a huge drop-off in the graph of our empathy as we sense we’re encountering something both familiar and not quite right. The same year as the Calgary test, having recently developed his first-generation Geminoid F, Ishi­guro and the University of California at San Diego published a study of neurons associated with empathy. The team used an fMRI machine to scan the brains of 20 people in their twenties and thirties as they watched separate videos of one of Ishi­guro’s female androids, the same android with its machinery revealed, and the living human that the android was modeled after.

The subjects saw each in turn wave its hand, nod, pick up a piece of paper, wipe a table with a cloth. Of the three videos, it was while watching the humanlike android’s motions that the parietal cortex of the subjects’ brains would light up most—in particular, the areas that connect our detection of bodily movement with our so-called empathy neurons. The researchers believe this revealed that the smallest gestures can create perceptual contradictions in the brain, sparking the uncanny valley effect. Ishi­guro returned to the lab and redoubled his focus on the android’s most minute movements: the precise tilt of the chin, the rotation of the head, the restraint of the smile.

Around the same time as the department store display, Ishi­guro managed to use the Geminoid F to generate a bond between two humans. Tettchan, then a game designer based in Tokyo, was recently divorced when he met Ishi­guro in 2012, and he mentioned that he was curious about the possibility of a romance with a longtime friend named Miki. Ishi­guro invited them both to his research institute in Nara, where he’d asked his students to have a female android ready for teleoperation. He placed Tettchan at the teleoperation desk and closed the door; he took Miki into the other room to meet the Geminoid F. Then he invited Tettchan (who was listening in) to talk to him and Miki through the robot. As Tettchan spoke, his voice computer-­altered to sound female, the android’s lips moved in sync with his words, the tilt of her head and her long human hair in rhythm with his own movements. “It’s like a real female,” Ishi­guro told Miki, enjoying himself. “This is not Tettchan, this is a new woman, really cute and beautiful.”

And so they “played,” making small talk, Tettchan trying out his new female incarnation. He made Miki and Ishiguro laugh, and watching Miki’s face through the monitor, he could see a change. That was when Ishi­guro, knowing Tettchan’s complicated feelings for Miki, said to her, “OK, you should kiss her.” And Miki, looking hesitant, leaned in toward the android—the android inhabited by Tettchan—and kissed it on the cheek. The feeling, Tettchan said, was “like thunder.” Any boundary between them suddenly vanished.

Not long afterward, Tettchan and Miki decided to live together. Tettchan is still not exactly sure how Ishi­guro’s machine worked on them, but he remains convinced that it made them into a couple.

Ishi­guro believes that since we’re hardwired to interact with and place our faith in humans, the more humanlike we can make a robot appear, the more open we’ll be to sharing our lives with it.


SCROLL DOWN

Ishi­guro believes that since we’re hardwired to interact with and place our faith in humans, the more humanlike we can make a robot appear, the more open we’ll be to sharing our lives with it.

Over dinner with Hiroshi:

He has spent a lot of time talking to himself through his androids, testing them, imagining their effect on other ­people. Hiroshi (who by now has asked me to call him by his first name) tells me he’d like to record himself saying “I love you” and then program an android to repeat it back to him in a female voice. He is kidding when he says this—but maybe it’s another of his half-jokes. At the very least, he believes the need for such an exchange exists. It would be, he says, “a real conversation.” A conversation with himself.

“A conversation is a kind of illusion,” he says. “I don’t know what is going on in your brain. All I can know is what I’m thinking. Always I am asking questions to myself, but through conversations.” Over the years of operating his androids, communicating through them or with them, he has found that he isn’t really concerned about the other person’s thoughts. “Always I am thinking of myself. I need to understand your intention, but it is not priority. Before that, I want to make clear something in my brain. Otherwise, what is the motivation to talk?”

In other words, he can only imagine using conversation with others as a means to better understand himself—and nothing is more pressing than that. He turns to the conversation the two of us are having. “We don’t know how much information we are sharing,” he tells me. “I am always guessing, and you are always guessing, and through our conversation patterns we can believe that we exchange information. But I cannot access your brain directly.

“What is ‘connection’?” he asks. “Other person is just a mirror.”

On some fundamental level, we understand each other’s immediate intentions and desires—of course, we do; how else would we function? But Hiroshi’s view, though stark, seems sadly right: There are entire planets of intimate information, our most interior level of consciousness, that we will never fully be able to share. Our longing to connect, to bridge this divide, is a driving human desire—one that Hiroshi believes will someday be satisfied through humanlike machines. He is convinced that human emotions, whether empathy or romantic love, are nothing more than responses to stimuli, subject to manipulation. Through the fluid interplay of its pneumatic joints, the arch of its mechanical brow, the tilt of its plastic skull, the many subtle movements achieved through years of research studying the human template, the android becomes more able to span that gap, to form a perfectly engineered bond with us. An elaborate metaphysical trick, perhaps—but what does that matter, if it fills a need? If it feels real?

I think of the gentle look on the face of Geminoid F as she glances down at a smartphone that she cannot read. He wants us to imagine her reading notes we have sent her, to imagine her loneliness, to love her. Every time we project our own feelings onto her—imagine a shared experience, a connection—his work inches forward.

Hiroshi says little about his personal life, but, with his constant travel and self-imposed 16-hour workdays, I understand that he and his wife lead fairly independent lives. “We have some simple rules. She never asks about my job, I never ask about her hobbies.”

Quickly, he brightens up—he has found a way to return, in his mind, to the work. “I want to know the meaning of ‘love.’ Do you know the real meaning? What is ‘love’?”

I think for a moment. “It changes all the time in my mind.”

“That’s good!” he says, surprised. “You are like a scientist. I am always changing too. I am having different hypotheses every year. Before I pass away, I want to have a better understanding about love.”

Hiroshi now tells me of the two times he has seriously considered suicide: first at 36, when one of his top students bested him at a computer-­programming challenge (his focus at the time), and again 10 years later, when another student proved to be a sharper, more prolific writer of technical papers (something Hiroshi took great pride in). Both times, he swerved out of the depression by finding a new angle on his work. But those instances heightened his dread that he might not be able to prevent the slow, natural deterioration of his mind. He is already certain that his concentration is not what it once was. Developing dementia as he ages is his worst fear. Without being able to generate new ideas, “probably I cannot find any reason to survive in this world. I don’t like to imagine that.”

We are quiet for a moment—then he leans in again.

“Do you know what is the soul?” he asks. “Soul is not so personal. In Japan, when we pass away, our soul goes back to the same place, back to the mountain. So now we are living individually, like this”—he motions to the two of us sitting on mats. “We have our own souls. But when we pass away, we’re going to share something. Soul is going back to the place where souls are coming together.

“Soul is not lonely,” he says. “Soul is not alone.”

Ishiguro now must keep his (naturally shifting, aging) human body corralled within the android’s static limits.

On a Saturday night, I meet up with Hiroshi and Rosario Sorbello, a robotics professor from the University of Palermo who makes a couple of pilgrimages to Hiroshi’s lab each year. He often sends his students to study there, and he arranged for Hiroshi’s android play to be performed in Sicily. For a tall man in a well-tailored suit and fine leather shoes, Sorbello is boyish, and he clearly relishes his access to Hiroshi—he reminds me, twice, that Hiroshi is “a very important person.”

We meet in Minami, one of Osaka’s hectic shopping districts, and have an evening of street food: huge bowls of ramen and batter-fried balls of octopus. (Hiroshi used to come here a lot in his days as a poor grad student.) After red-bean dessert soup, served by a woman in a ruffled apron, Hiroshi makes a decision: Rather than head to a bar, he says we should go to “the bar in my office.” En route, we stop at a fluorescent-lit, 24-hour convenience store to pick up drinking food—wasabi peas, octopus jerky, chocolate Pocky Sticks—before boarding the train back to the university.

As Hiroshi scrolls through his phone, Sorbello talks about the desire for intimacy with androids—something he’s clearly thought a lot about. “Can you imagine what it would be like,” he asks, “to want to kiss a robot? To want to kiss that rubber, not-human flesh? There are people who have those kinds of desires. Imagine if you could run heat through its skin so that it feels not like cold rubber but warm to the touch? There are people who want to try things with that.” Human sexual and romantic relationships are unavoidably messy, he says, and many people would like to keep their lives simple—in which case a relationship with an android might be a solution. “I think this is the future,” he says.

Sex is arguably the ultimate physical act of human connection—but it can also be merely that: an act, a simulation of intimacy. Sex can be thought of as something that transcends the purely physical, but in reality it is often an experience that’s mostly physical, not as intimate as we pretend it can or should be. Looked at in this light, a whole range of sexual experience, at least in theory, could be replicated with an android.

On Sorbello’s recommendation, I later read Love and Sex With Robots, a 2007 book by AI expert David Levy. In it he proposes that we are not far from a time (he suggests roughly the year 2050) when humans will desire robots as friends, sexual partners, even spouses—a premise he seems unnervingly OK with. It all comes down to our willingness to believe in the robot’s emotional life and desires. Designed with the physical proportions that its human owner prefers, the preferred voice timbre and eye color and personality type, and the ability to recall and riff on its owner’s personal stories and little jokes, android will captivate human.

Levy takes Alan Turing’s famous claim that the convincing appearance of intelligence (in AI) is proof of intelligence, and he expands that into the emotional realm: “If a robot behaves as though it has feelings, can we reasonably argue that it does not? If a robot’s artificial emotions prompt it to say things such as ‘I love you,’ surely we should be willing to accept these statements at face value … Why, if a robot that we know to be emotionally intelligent, says, ‘I love you’ or ‘I want to make love to you,’ should we doubt it?” Human emotions, he argues, are no less “programmed” than those of an intelligent machine: “We have hormones, we have neurons, and we are ‘wired’ in a way that creates our emotions.”

In other words, Levy argues, our inner lives are essentially algorithmic, much like an AI’s. A few decades from now, he writes, the differences between human and android may be “no greater than the cultural differences between peoples from different countries or even from different parts of the same country.” As for the actual sex, Levy believes that it will become not only a recourse for the socially isolated but also an accepted outlet for the sexually adventurous or for someone whose partner is sick or traveling.

ishiguro_3.jpg

These are pretty radical ideas about human nature and intimacy, and yet I recognize the desire some might have to turn to an android for closeness, for companionship—for comfort when you’re far from home, maybe on the other side of the planet, on assignment for weeks at a time. And if someone provides you with a salve, why not take it? Most of us already allow technology to mediate what was once simple, direct human interaction—what really is the difference? And is that difference so essential to the experience of being human that it must be preserved?

Back on campus, we pass the few students who are still sequestered in the lab, working late, and hide out in Hiroshi’s office. There he slides back his whiteboard to reveal a hidden liquor cabinet. He pours us some excellent local whiskey, and we sit back and listen to his collection of music, everything from Japanese pop ballads to Simon & Garfunkel. We’ve all had a few. Hiroshi tells us about how, from the moment he started exposing people to his androids, a shift took place: The androids, he says, seemed to unmask the humans around them, to reveal a desire they’d carefully been hiding—for connection, for touch. There was the expected: Men who leered at the female androids during industry showcases, men who had to be watched closely, for fear they’d try to kiss and grope the robots. But something more complicated was also taking place.

Shortly after the android of his daughter was completed in 2002, Hiroshi had his students at Kyoto University use it to test the differences in human response to a mechanical-looking robot and one that was humanlike. When not in use, the android was left in the middle of the lab, and soon a few students complained that they were having trouble working in front of it. They felt it looking at them. (From then on, they made a habit of placing it with its face to the wall.)

Things were further complicated when Hiroshi was informed that one of the students had become attached to his daughter’s replica. During the day, this student would run the experiments, but late at night, when he thought he was alone in the lab, he would serenade the android with his flute and then chat with it, asking what it thought of his playing. It was as if he felt he could only reach out for companionship this way, in secret.

This incident made Hiroshi realize that these androids could have unexpected emotional impact. “That was the first android,” Hiroshi says. “We did not know what would happen.” He moved the android to Osaka University and assigned another student to oversee the work. He also laid some ground rules for how it could be used: not late at night and not alone.

“It feels like—how can I say?—not my daughter, but a special person for me,” Ishiguro says.

When he then created the first replica of a grown woman, he was a little wary of what his students might do with it in the lab. Would they want to sleep holding her in their arms? Hiroshi witnessed how one staff member, who’d been closely involved in the Geminoid’s production, became visibly flustered in front of “her.” Hiroshi’s theory is that a friendly human woman will always be merely a “real person,” never as “elegant” as her android counterpart. “We want to have some ideal partner, and the android can be a very strong mirror to reflect your own idea.” In this way, a relationship with an android is like having a partner who is, literally, an extension of yourself.

The response of so many men to Hiroshi’s female androids unsettles him. But it’s also one he’s been cultivating. In 2014, he embarked on a new project that marries his personal perfectionism with his ideas about female beauty: During my visit, he and his robotics team are at work on what he refers to as “the most beautiful woman.” His not-entirely-­empirical approach to its appearance has included speaking with a popular cosmetic surgeon in Osaka (his own), analyzing images of Miss Universe pageant finalists, and, in the end, trusting his gut. (He has reminded me a few times that he thinks more like an “artist” than other roboticists.) Hiroshi worked for two 12-hour sessions with a technician to create the android’s 3-D rendering. He was thrilled to discover that the slightest change to its eyes or nose transformed the rendering into a completely different person. “It feels like—how can I say?—not my daughter, but a special person for me,” he says.

Now, when I ask Hiroshi why he puts such emphasis on good-looking mechanical women, he reminds me that the larger goal of his field is to have people accept robots into their lives. “And which is more acceptable to many people,” he asks, “beautiful woman or ugly woman?” In a corporate lecture I later hear him give, he sums it up like this: “A beautiful woman you don’t picture going to the restroom or getting tired. So I think beauty is better represented by android.”

At this point, Hiroshi stands up from his ergonomic chair, as if inspiration has struck. He turns his back to me and Sorbello, rummages through his drawers, and produces a black zipper bag. From inside, he pulls out two hand-size foam mock-ups of a humanoid figure and offers me one as a gift. He picks up the other and holds it out to me.

“Let’s make an experiment,” he says. “We bring them together and we make them kiss.”

I’m unsure of where this is going. “OK.”

I bring the face of my tiny figure to meet the face of his, and their motionless mouths touch.

“It feels funny, right?” he asks. And it does. It feels just a little like crossing a line.

I return to Tokyo for a few days to meet more of Hiroshi’s colleagues. And in the midst of this back-and-forth from Osaka, something is starting to happen: I am falling for someone I met on the second night of my trip.

I was introduced to Ethan over email by my literary agent, who knew I’d been searching for helpful contacts in Japan. He’s an American (also in his thirties) who moved to Tokyo for graphic-design work a decade ago and is fluent in Japanese. Ethan (a pseudonym) emailed me the names of fixers and translators and boutique hotels and agreed to join me for dinner before I took the Shinkansen bullet train west to Osaka. When I found him that evening, at our meeting spot in front of a Shibuya-ku Metro station, his eyes reflected back to me the thought just then crossing my mind: This will be a very good night.

I’ve never been particularly drawn to men who are handsome in a conventional way. But Ethan’s looks are so classically handsome, it seems impossible that he walks around with such a face and such a strong jawline and such a finely shaped head. (Have I ever before thought about the shape of a man’s head?) There is also the small hollow at the back of his neck and the width of his shoulders (something about their proportion gives me a startled feeling in my chest) and the smell of his skin and the timbre of his voice (deep and musical).

He becomes my guide in an unfamiliar city. I am led around and am much happier for it. We drink in a white bar with sliding paper screens; a jazz bar in which no one is allowed to laugh out loud; a space with eight seats, covered in Wim Wenders film posters; a hotel lounge with a piano singer and 52nd-floor views of the city. We talk about books; we talk about our families; we talk about the people we’ve thought we loved. We walk down the streets at night with our arms lightly touching; we sit with our knees lightly touching; I lay my palm on that hollow at the back of his neck. And in private, we lie down in his bedroom, on a thin mattress on the floor, and remove all our clothes. It has been ages since either of us was drawn to someone in this way, an attraction that feels like a planetary pull, seemingly outside the realm of reason and predictability—the very thing we spend so much time trying to conjure up but over which we have no control.

It is thrilling. And for me right now, immersed in the world of android design—a heavily mediated world in which soft silicone shells stand in for human skin, in which we search for signs of human kindness or sadness or pity in a mechanical face—it is also a relief that something so simple can still happen. It is a relief because it means that we are animals, not ideas; that our chemistry is not as cool as a set of programmed responses—there’s an immediate magic to it. To know that that instinct is not broken in me, and to be able to answer it, makes me feel like a person again.

Android Evolution

We traced Ishiguro’s android milestones on the road from Uncanny Valley to (near) Total Turing. —Caitlin Harrington

2002

REPLIEE R-1: Ishiguro’s first android was produced in 2002 and modeled on his 4-year-old daughter. It looked human enough, but its movements were all machine: jerky and creaky-gear-sounding

Photo: Osaka University

2005

REPLIEE Q-1: The lab installed pneumatic actuators in this model, allowing for humanlike gestures. To boost authenticity, they added autonomic elements like
breathing and posture shifting. The android debuted at Japan’s 2005 World Expo.

Photo: OSAKA UNIVERSITY AND KOKORO CO., LTD

2006

GEMINOID HI-1: The first replica of its creator, HI-1 was also the lab’s first android to work via teleoperation:
Ishiguro controlled his doppelbot via mics and webcams.

Photo: ATR HIROSHI ISHIGURO LABORATORIES

2010

GEMINOID F: The team simplified its new model, Geminoid F, for experimental use outside the lab,
limiting the mobility of its limbs. It went on to appear in numerous international productions of a play written with the android in mind.

Photo: OSAKA UNIVERSITY / ATR HIROSHI ISHIGURO LABORATORIES

2010

TELENOID: Hiroshi and his team stripped away differentiators like age and gender to make this droid universal, then dispatched several of them to nursing homes as companions.

Photo: ATR HIROSHI ISHIGURO LABORATORIES

2011

ELFOID: The team miniaturized the Telenoid and turned it into a cell phone. Elfoid creates what Ishiguro calls telepresence: the sensation that the person you’re speaking to is physically embodied in the 8-inch-long device.

Photo: ATR HIROSHI ISHIGURO LABORATORIES

2015

ERICA: The lab’s first fully autonomous android, Erica can hold 10-minute conversations. Voice recognition, infrared human tracking, speech synthesis, and
natural motion generation all bolster her humanness.

Photo: ERATO ISHIGURO SYMBIOTIC HUMAN-ROBOT INTERACTION PROJECT

When Hiroshi first considered building an android, he began a search for the right silicone. He turned to Orient Industry, a company that specializes in high-end “love dolls,” sex dolls that can cost thousands of dollars. They collaborated on a trial model—but Hiroshi soon severed the relationship. As his reputation grew, he worried about how such a collaboration might look. The government does not want its money associated with love dolls.

The sex industry, however, does not need government approval to thrive. Back when they briefly worked together, Orient Industry operated out of a single room; now, almost two decades later, it occupies an entire building—and it sells nothing more advanced than poseable dolls. Human-robot sex, Hiroshi believes, will definitely be part of our future, it’s merely a question of when. He knows that his research would be very helpful in that arena, but as a respected academic he would require a noncommercial, for-the-betterment-of-society reason to pursue that line of inquiry. Perhaps for people with disabilities, he suggests. “Once we create a pretty good sex doll, you know, definitely other people want to use it,” he says. “It’s a basic desire.”

We’re heading back to Osaka from Nara as he says this, speeding down the highway in his slim black Mazda—Hiroshi drives the way he walks: irrationally fast—and eventually our talk turns to the 1982 film Blade Runner. He’s stuck on the lead female replicant, whose name he can’t remember. “She looks like you!”

Hiroshi pauses for a moment, and when he speaks again, it’s in a thoughtful voice. “Someday I want to have my own replicant,” he says. “Probably everybody want to have one, right? Don’t you think?”

“Their own attractive robot?”

“Yeah. I think so.” In what must be another of his one-sided conversations, he’s agreeing with himself. “It is not just robot—it’s almost human. It’s ideal.”

“An ideal woman?”

“Probably. No idea.” He laughs. “That is one of the projects”—the “most beautiful” android.

We drive on in silence, then he asks a surprising question: What would people think if he made my copy?

For whatever reason, the possibility—even in the abstract—has never occurred to me, and the idea is unexpectedly intimate.

I try to imagine how this would play out. They would encase my body in plaster, and then my various parts would be molded and manufactured and bolted together. And a silicone replica of my face, a bald and half-smiling not-me, would be stretched around its mechanical skull. And then my parts would be delivered to one of Hiroshi’s labs and unpacked and assembled and dressed in a skirt and blouse and a long black wig; maybe a student would take a pair of heels (patent leather, slipped off an older model) and place them on my feet. Eyes that are not my own, but of a convincing shine and color, would appear to stare back at the gathering of researchers.

Let’s say I’m not used in the lab at first but put out into the world, on display: destined for a new stage play or an android opera. An assistant professor and I will travel together from venue to venue; at the end of each international stop, back at the hotel, maybe he will crack open the valise that holds my head and talk to me about his frustrations. And eventually, when this android theater comes to the end of its run, I will be retired to an observation room, stacked against the wall with my clothes and hair stripped off and my head bowed. And the students will sometimes entertain themselves late at night by making my replica sing karaoke while they drink beer. And for the rest of time—or until my silicone is no longer considered worth replacing—this facsimile of myself will be made to do things, to say things, that are beyond my control, always borrowing my appearance, my face, my expressions, the memory of the living woman who was her model.

I’m not prepared to give away my likeness.


SCROLL DOWN

The Telenoid’s delicate features sometimes appear feminine, sometimes like those of a small boy—but altogether too knowing for someone so young.

I compared Hiroshi to Pygmalion, but that comparison is only partly right. His desire to create, that personal obsession, is driven less by romance and more by ego. In all my time with him, I never get the sense that Hiroshi—unlike some of his robots’ fans and perhaps some of his colleagues—fetishizes his female androids. What excites him is the power of his role as Creator, the notion that he may one day crack the code of our emotional bonds. And he doesn’t care what shape that solution takes. If he could reduce the human form to its barest, most minimal structure, he would. What if so many of these physical details—the precise silicone mold, the perfect eyelashes and cuticles—were a distraction from the true nature of sonzai-kan? One way to know would be to strip the android down to something more essential.

He has done just that. The shape came to him in a dream. When he awoke, he sculpted a model out of clay. The Telenoid is about 1½ feet tall and ghost-white, a toddler with an alien-smooth face. It has stunted arms and, instead of legs, a bulbous stump—as if, in place of genitalia, the halves of the ass had continued all the way around to form two spheres. A stretch of silky white spandex serves as the lower neck, bridging the head and the body, but it is otherwise a continuous, jointless piece of supple plastic, as smooth as a naked child.

In repose, the expression on its face is serene enough to be unsettling—perhaps because of its deep-set black eyes; its thin, pursed lips, ever so slightly upturned at the corners; and its gentle, barely perceptible brow. Its delicate, thin features sometimes appear feminine, sometimes like those of a small boy—but altogether too knowing, too serene, for someone so young.

Its face conveys a calm authority that a human toddler would not possess, but its body and small gestures transmit the vulnerability and neediness of a child.

In a research room at Hiroshi’s institute, his team shows a group of Danish visitors the latest model. Propped up on a tripod, low to the floor, the Telenoid squirms to life once activated. It looks up at us and begins to make a play for our attention, glancing around, wiggling its short arms. Its little movements are completely fluid and easy, giving it a sweet demeanor. It starts speaking to us in Japanese in a feminine voice, drawing a grad student named Miriam into an animated conversation. For now the Telenoid is teleoperated, but Hiroshi hopes to make it autonomous within the next few years. Its face conveys a calm authority that a human toddler would not possess, but its body and small gestures transmit the vulnerability and neediness of a child.

Miriam lifts the toddler-thing to cradle its stump in the crooks of her arms, and the two continue chatting in cooing, affectionate tones. And at this point, after a few minutes of observation, the words that come to mind are no longer repulsive and nightmare, but small and dear and friend. It becomes easy to feel protective of the little alien.

As it turns out, the visitors at the lab when I visited were there because Hiroshi was hoping to partner with a venture capital firm to install Telenoids in senior care facilities throughout Denmark. For a couple of years, he had been traveling there every few months. Hiroshi’s team and their Danish partners were in the final stage of their field tests; they hoped to have a viable business plan in place soon. Everyone was optimistic: Test subjects have been quick to connect with the strange humanoid. Media events for the Telenoid in Denmark were attended by Japanese ambassadors and the Danish crown prince, who embraced the humanoid on-camera. He said the experience reminded him of holding his own child.

And the video footage—of elderly people in nursing homes, each supposedly with some degree of dementia—is compelling. In one, an older woman in a colorful turtleneck sits on a sofa in a facility in Kyoto with a Telenoid on her lap. Though her caretakers have explained that she rarely speaks with them, she is shown here in excited conversation with the humanoid (which she may or may not understand is being teleoperated by a volunteer in Osaka). In another clip, a far frailer-looking woman, more than 100 years old, sits slumped at a desk, her arms wrapped around herself. “She’s depressed and does not talk with others,” one of Hiroshi’s researchers says. When a caretaker sits beside her and hands her a Telenoid, however, she lights up, grinning and laughing. Out of sheer pleasure, she begins making short, clipped babylike sounds: “Ah-ah-ah-ah!” She grips the mechanical toddler to her chest, a blissful expression on her face, and starts to rock it slowly back and forth.

This clip is powerful evidence that a machine can conjure up an emotional connection—but a connection to what? Is it a look of recognition that flashes in this century-old woman’s face, the resurrecting of some long-ago happiness? “We don’t know yet, exactly,” the researcher says. “But those who love Telenoid tend to be someone who used to have a baby.” It takes a moment for the horror of this statement to sink in—that someone would be left alone in their advanced age to relive the joy of having a child through the cradling of a robot with stunted limbs.

More than a dozen years of progress have brought Hiroshi full circle: from his young daughter’s android to another child robot—one that is blank, one that can be anyone’s small child. A humanlike robot that is both terrifying in appearance on the most primal, gut level and undeniably effective: Once it’s in operation, you are drawn to it, in sync with it, cannot help but feel empathy with it. The countless ways in which we judge someone based on their appearance all evaporate in the face of this “neutral appearance,” as Hiroshi calls the Telenoid’s blank, abstract body. And what is left in its place is that ineffable thing he has been trying to define: a distinctly human presence, free of the uncanny. It is an outsider, like its maker—but one who manages to trigger our affection. While holding the android, it hardly matters that this humanness is emitting from something that barely resembles a human at all.

In repose, the expression on the Telenoid’s face is serene enough to be unsettling.