An unusual cafe recently opened in central Tokyo. White plastic robots took orders from customers and brought them coffee. They glided slowly from table to table and spoke in a human voice.
Those voices belonged to actual humans, because actual humans were operating the robots remotely. The robots served as the eyes, ears, and hands of bedridden individuals with amyotrophic lateral sclerosis (ALS), who piloted the machines from home using a tablet. The operators could move the robot around the room, manipulate its arms and head, and speak to customers through an on-board speaker.
While the cafe was only open for a few weeks on an experimental basis, the robots’ creator, Kentarō Yoshifuji, wants to launch a permanent “telerobot” cafe in time for the 2020 Tokyo Olympics and Paralympics. Yoshifuji aims to provide job opportunities for people with immobilizing disabilities who may not otherwise be able to work outside the home. He sees the model eventually migrating from the cafe setting into other service positions like airline check-in attendants.
More broadly, he envisions using the robots to revolutionize telework. Currently, most telework is disembodied: it involves workers in call centers or gig platforms like Amazon’s Mechanical Turk. But Yoshifuji is representative of a broader Japanese trend that aims to use telerobotics to facilitate embodied forms of remote labor. People could work a much wider range of jobs without leaving their home. In this new world, “working from home” could involve everything from serving customers in a cafe to checking in passengers for their flight.
While telerobotics is a global industry, this emphasis is unique to Japan. Most telerobotics companies in North America, China, and Europe are focused on highly skilled positions. They are building telerobots for the healthcare industry (“telemedicine”) and for various white-collar corporate contexts that let workers with specialized expertise be remotely present in a workplace. (Such experiments are not without controversy: a hospital in Fremont, California recently made headlines after a patient’s family objected to receiving a terminal cancer diagnosis via the video feed of a telerobot rather than by an in-person doctor.)
By contrast, Yoshifuji and other Japanese telerobotics innovators are focused on the opposite end of the labor market. They want to use telerobotics for the much larger field of lower-skilled service jobs. With telerobotics, the pool of people who could fill such jobs could grow to include individuals with illnesses like ALS, people with disabilities, the elderly, and those with childcare responsibilities that keep them at home. It could also include foreign workers prevented from entering the country by restrictive immigration laws.
The goal of Japanese telerobotics isn’t to provide better access to remote expertise in the form of higher-skilled workers, in other words, but to technologically recuperate lower-skilled workers who might otherwise be excluded from the workforce entirely. What is envisioned is nothing less than a digital platform for physical work, one that could utilize previously untapped labor reserves to power global networks of on-demand robot avatars.
This may open up potentially meaningful opportunities for those unable to physically travel to a work site. At the same time, it risks further isolating the already immobilized, enabling remote access to their physical labor while fixing barriers to their social mobility ever more firmly in place. While the stated promise of teleworker robots is one of freedom for those employed, in practice it is employers who are most liberated by the arrangement.
Working from Home
Japan has been at the forefront of experiments with embodied telework since the 1980s, but it wasn’t the first in the field. Japanese researchers were building on initial teleoperator experiments in the United States that took place in the 1940s in the aftermath of the Manhattan Project, as military scientists tried to find a way to allow workers to physically manipulate radioactive materials while maintaining a safe physical distance. Subsequent American teleoperator research has continued to have a strong focus on operation in hostile environments, such as systems for military drone pilots to carry out missions remotely.
By contrast, teleoperator research in Japan took shape against the background of the country’s pacifist postwar constitution — itself a product of the American occupation — which for a long time amounted to both a formal and informal taboo around research with direct military applications. As a result, Japanese researchers of the 1980s and 1990s worked on more everyday telerobotics uses, from caring for the elderly to housekeeping to helping the visually impaired navigate city streets.
In recent years, this emphasis on the everyday has come to center on a more pressing social challenge: the intensifying labor shortage triggered by the country’s aging population. As of December 2018, there were an average of 168 jobs for every 100 job seekers in Japan, with a shortage of 6.44 million workers expected by 2030. Some of the most acute shortages are in physical labor and service sector work. Convenience stores, for example, are increasingly struggling to find staff, even as much of the country has come to rely on them for access to basic goods and services.
One approach calls for putting bedridden seniors back to work using telerobotics. This is what Tokyo University professor Michitaka Hirose proposes in a 2016 book, arguing that VR-controlled telerobots could employ elderly Japanese well into their nineties and beyond. For seniors who no longer have the physical or mental resources to hold down a full-time telework position, he proposes a “senior cloud” job-share platform where several teleworking seniors could pool their skills to do the work of one younger adult. In recent talks, pioneering telerobotics and virtual reality researcher Susumu Tachi expands the potential labor pool even further, arguing that stay-at-home parents and overseas workers could also work as VR teleoperators.
The inclusion of overseas workers suggests that telerobotics offers a way to extract more labor power not only from those at home, but also from those abroad. This would extend the outsourcing trends that began in the 1980s and 1990s, when information technology made it possible to offshore certain kinds of service work. By the turn of the century, countries like India, Indonesia, and Mexico had large portions of their workforce telecommuting via computer on a regular basis, with nearly 10 percent of workers worldwide working from home every day according to a 2012 Ipsos/Reuters poll. Telerobotics has the potential to vastly accelerate this trend, because embodied remote workers can do many more kinds of jobs — including, potentially, those in manufacturing.
Tachi imagines using a global team of teleworkers to keep factories productive around the clock. He imagines three daily shifts split between workers in Nigeria, Japan, and Mexico, enabling a Japanese plant to stay in operation twenty-four hours a day. Along with reducing labor costs, telerobotics will also keep workers in their home countries. Tachi believes this would prevent the “problems” that an increase in physical immigration might cause.
This vague reference to “problems” echoes a recurring theme in Japanese robotics, as identified by the anthropologist Jennifer Robertson: a preference for technological solutions to the labor shortage as a less culturally threatening (and more politically palatable) alternative to increasing the number of foreign workers by relaxing Japan’s immigration laws. While the ruling Liberal Democratic Party recently pushed a controversial immigration reform bill through the Japanese legislature — set to increase the number of lower and medium-skilled work visas by 345,000 over the next five years — the changes still fall far short of fully addressing the labor crunch. Existing programs to provide foreign labor on a more short-term basis, such as the Technical Intern Training Program, have been notoriously vulnerable to employee abuse and exploitation.
Whereas autonomous robots would seek to replace these human workers entirely, telerobotics instead seeks to import only those aspects of human embodiment deemed economically useful. Somewhat ironically, Hirose, Tachi, and Yoshifuji each present their telerobot systems as a more human alternative to full-scale automation, even as they play up the technology’s advantage over on-site human labor and increased immigration.
Mask of the Robot
This brings us back to Yoshifuji’s cafe robots. Telework systems may offer meaningful opportunities to those who otherwise have limited options for traveling beyond their home or hospital room. Yet the power relationships inherent to these platforms ensure that remote workers remain highly dependent not only on those physically present at the worksite, but also on whoever controls the platform itself. The virtual mobility of the teleworker extends only as far as the labor market demands, and can be unilaterally revoked at any time.
For a preview of what this might look like, we can turn to existing digital “microwork” platforms like Amazon’s Mechanical Turk. The median hourly wage on Mechanical Turk is around $2 an hour. This comes with few opportunities for promotion or skill development, a risk of repetitive strain injuries and eye problems from long working hours, and limited recourse when it comes to conflict resolution or negotiating better working conditions. Many workers on these platforms come from the world’s poorest countries, while their labor helps sustain the economic productivity of the richest ones. Robot telework would extend these precarious and often neocolonial labor relations far beyond the screen, incorporating and exposing workers’ bodies more directly.
The selectively transmitted embodiment of telerobotics also risks the further erasure of already marginalized bodies. Unlike almost all teleworker robots produced outside of Japan, Yoshifuji’s robots do not show the operator’s face. Instead, they have a white plastic head with large mint-colored eyes that glow to signal a teleoperator is present. Standing a diminutive 120 centimeters tall, Yoshifuji’s robots are humanoid in shape but deliberately gender-neutral and vaguely alien in appearance. Yoshifuji argues that this abstracted appearance allows those who encounter the robots to more easily project emotions and personalities onto them as they see fit.
Yet these design decisions also foreclose the possibility of face-to-face human contact with the remote workers. The physical person of the worker is replaced with a cute, infantilized character designed according to the norms of Japanese avatar culture. Without discounting the potential appeal of swapping out one body for another, it will be the employer who decides how a teleworker will appear to the world, and what limits will be placed on their physical agency. While Yoshifuji claims that his goal is to eliminate the loneliness of the bedridden by enabling physical interaction outside the home, it seems likely that wearing the mask of the robot will intensify alienation rather than alleviate it.
Work Without Workers
The telerobot proposals currently circulating in Japan rely on a tacit shift of social expectations surrounding workforce participation. The implicit message is that every last member of Japanese society (and beyond) must contribute their labor to the nation’s productivity over their entire lifetime, using whatever they have left to give.
Yoshifuji goes so far as to envision a day when telerobots could allow the bedridden to look after their own physical needs by controlling the bodies of their robot caretakers. This would address the already severe shortage of care workers by placing the care burden back onto those in need of care, extending neoliberal demands for self-reliance to even the most socially vulnerable. While telerobotics initially emerged as a way to shield workers from hostile environments, here the technology would serve to shield society from the need to take care of people who have ceased to be economically useful.
At the trans-national scale, embodied telework will intersect in complex ways with ongoing debates around the world over immigrant labor and cross-border migration. In parallel with the erasure of the elderly and people with disabilities, embodied telerobots risk the further effacement of the foreign workers keeping the service industries of countries like Japan and the United States afloat. Alex Rivera’s prescient 2008 science fiction film Sleep Dealer provides one dystopian vision for how embodied telework might interact with existing border politics. Rivera imagines a future where Tijuana factories use VR-enabled telework systems to allow Mexican workers to perform construction work and other physical labor in the US remotely. As one Mexican robot teleoperator in the film trenchantly puts it, “We give the United States what they always wanted: all the work, without the workers.”
Yet even as embodied telework seeks to extend the outsourcing projects of earlier decades, it will likely operate in tandem with artificial intelligence and automation projects increasingly targeting these same positions. The most economically plausible scenario is a blend of the two approaches, relying on automated systems for some tasks while falling back on remote human labor for roles where a live human still has the advantage. This is already happening in existing online teleworker platforms, where piecemeal human labor often provides a kind of stopgap intelligence to make up for where full-scale automation falls short. For example, the Bangalore-based company Vernacular.ai provides chatbots to handle customer service calls, but switches over to live human operators when the AI is (as yet) unable to come up with a convincing response.
In a recent book, the economist Richard Baldwin predicts what he calls a “globotics upheaval” driven by this combination of globalized telework platforms and fully automated robots. But while Baldwin promises a richer and more “human” existence for those at the receiving end of these services, his book is conspicuously silent on the question of what it will be like for those already marginalized in the global economy to compete with AI systems for remote work. If prior eras of technologically enabled globalization are any guide, it will most likely widen the existing power divide between those operating the platforms and those with few options except to rely on them for employment.
The more embodied and perceptually immersive quality of the work means this power gap will be all the more viscerally felt by those whose bodies are put to work. As these physical telework platforms take shape, the critical question is how to ensure they do not simply strip the labor from the already vulnerable while shackling their bodies, extracting what is of economic value while leaving the rest behind.