The 4th IR, Humanoid Robots, and the Politics of Labor Concealment in Japan
This post is by Yulia Frumer, who holds the Bo Jung and Soon Young Kim Professorship of East Asian Science in the Department of History of Science and Technology at Johns Hopkins University.
What role might humanoid robots play in the world envisioned by 4th Industrial Revolution thinkers? Would robots free humans from work and allow them to engage in creative, satisfying activities? Would robots act as companions? Or would they increase unemployment and deepen inequalities? Numerous Japanese roboticists like to repeat that technology itself isn’t good or bad, but could become beneficial or harmful in human hands, depending on its use.
To see how humanoid robotics could be exploited to mask and deepen inequality, it is worth looking at the role humanoid robots already play in Japanese labor politics. It is perhaps not surprising to see Japanese policymakers proposing to use a technology, in which Japan is considered to be a world leader. And yet, given the poor results, Japanese policymakers’ continuous insistence on humanoid robots as a solution to labor shortages — resulting from an aging Japanese population — seems puzzling. One would think that policy makers would have learned from the colossal failure of the Innovation 25 policy draft, which promised to “revitalize” Japanese society by the year 2025 by means of technological innovation.[1] Or, one may assume that cumbersome and expensive technologies — like the room-size laundry folding machine that takes 10 minutes to fold one T-shirt, one at a time[2] — would make people more cautious about setting deadlines for life-changing technological innovation. One would also expect that 4IR visions that prioritize disembodied AI would shift policymakers’ foci away from embodied, human-shaped technology. But this is not the case. Despite all, in 2016, the Ministry of Economy, Trade and Industry (METI) introduced the Robots for Everyone plan, which “call[ed] for new ideas for utilizing robots in the service field, including department stores, beauty salons, hotels, and restaurants,” and offered about $2.2 million for further ideas for the possible use of robots in the service sector. The METI website announces that “The New Robot Strategy (approved by the Headquarters for Japan’s Economic Revitalization on February 10, 2015) states that increases in labor productivity and added value in the service industry shall be promoted by utilizing robots in the service field.”[3]
It is, therefore, worth dwelling on the probable motivations behind this insistence. I argue that despite their beneficial potential, technologies associated with humanoid robots tend to disguise the human labor required to operate them. As we shall see below, the approach to humanoid technology by conservative parties in Japan reveals that humanoid robots could be used to present a façade of social progress in accord with 4IR visions, all while obfuscating human labor in service, caretaking, and maintenance, and consequently deepening inequality.
The use of humanoid robots to obscure human labor was already evident in the Innovation 25 policy. Unveiled in 2007 by the first Abe administration, the policy envisioned in-ear automatic translation devices to remedy the increasing lack of English-language ability among Japanese youth; suggested investing in virtual reality technology to overcome Japan’s cultural and geographical distance from the Western world; implied that societal issues related to an aging population could be solved using humanoid robots (which, if successful, would have killed several birds with one stone). In addition to supplying manual labor, humanoid robots would replace nurses in geriatric care and free women from household chores so they could join the workforce as telecommuters or, more importantly, have more children.[4]
The draft is indicative of what conservative policymakers hope to achieve by introducing humanoid robots into society. Beyond the typical desire for a magic bullet, or the hype surrounding humanoid robots, or the technological optimism, it is worth paying attention to several details. We should note that the administrations of Shinzō Abe — which promoted both the 2007 Innovation 25, and the 2016 Robots for Everybody — are the most conservative administrations Japan has had in decades. This shows in the contents of the draft. The future society that Innovation 25 portrayed in 2007 preserved traditional gender roles, and sought to reduce the necessity of actually interacting with “foreigners” while preserving Japan’s cultural appeal to the world. Women (and only women) were “liberated” from housekeeping, childrearing, and caring for elders, so that they could have more children. True, according to this vision, by 2025 telecommuting would allow women to work, but only from home. (Innovation 25 represented men as absent from the home, presumably because they were working long hours in the office.) Under Innovation 25, the Japanese would not need to encounter foreigners because tourism would become purely virtual, thanks to virtual reality. In a particularly xenophobic and chauvinist image depicting the wonders of virtual reality, an American man is shown wearing virtual reality headgear that transforms an image of his blond wife into a Japanese geisha. The Japanese wouldn’t need to learn foreign languages either, because an in-ear translation device would translate everything for them (on the off chance they actually encounter a foreigner.)
Had anyone tried to warn the government that this vision was technologically naïve, their voice would have been drowned out by the overall optimism. Actual advances in robotics technology — the humanoid robots seated in the “greeter” booths at shopping malls, robots that can lift geriatric patients, the charming Pepper who chats with customers of the communication giant Softbank — have all been interpreted as sure signs that technology is indeed on its way to the futuristic vision of Innovation 25. People tend to extrapolate from human functions and assume that, if human ability to do X implies the ability to also do Y and Z, then a machine that is capable of X is surely capable of Y and Z.[5] But this is not the case — machines are made to perform specific tasks. A robot who greets you at the entrance to the mall would not be able to respond to an emergency, and one who can lift a heavy weight won’t be able to make a geriatric patient feel comfortable. Attempts to implement these wonders of robotic technology IRL have made it clear that existing technology is nowhere near being ready to entirely take over the service sector. What was at first a nice robotic gimmick soon lost its appeal and did not live up to its labor potential. Within a few years of its announcement, Innovation 25 was forgotten — that is, until 2016.
Initially, with the failure of the Innovation 25 policy draft, it seemed that Japan didn’t have a choice but to welcome societal changes by allowing more immigration, encouraging gender equality, and assigning more value to the work of caregivers. Companies began promoting — albeit, still only token — women, changing policies regarding parental leave, and encouraging both women and men to leave work early enough to spend time with family. This seemed to bear fruit, and over the last few years the downturn in the demographic trend seemed to slow.
So why, then, the renewed push for more robots? The clue may be hiding in the fact that the 2016 Robots for Everyone accompanied METI’s 4th Industrial Revolution Re-Skills Program, which is part of the larger “Future Vision towards 2030s” campaign.[6] METI’s website states that “the Certification Program for Fourth Industrial Revolution Skills Courses […] certifies practical educational courses targeting workers mainly in the IT and data fields in which future growth is highly anticipated and the creation of new employment is forecasted, thereby providing opportunities for such workers to acquire advanced expertise and knowledge, and to advance their careers.”
The structure of the 4th IR policy betrays the societal motivations behind it, and also explain the repeated push for humanoid robots. In Japan, the 4th IR vision made it possible to try to preserve a conservative vision of society in which Japanese salarymen are producers of value and women are caretakers. It is very clear that METI envisions the people retraining to fill 4th IR jobs (whatever they may be) to be men. The program explicitly excludes “fundamental and basic IT skill courses” from their Re-Skills curriculum, which means that only those who are already in the IT field have access to new, desirable employment options. Given that the IT sector in Japan is predominantly male, and predominately from privileged strata of society, this exclusion means that the 4th IR will become a playground for young, upper-class men. The new 4th IR policies work to increase polarization between high- and low-paying jobs, thereby further diminishing social mobility and reversing the increased value of maintenance and caretaking, which was beginning to facilitate diversity and gender equality.
So why robots again? The answer lies in the envisioned division of labor. One can easily spot a striking difference between the jobs METI envisions to hand to humanoids and the ones for which it intends to retrain humans. The former are undervalued jobs in the service sector, usually done by marginalized populations such as women, minorities, immigrants, and retirees; the latter are coveted IT jobs, open only to those who already work in this prestigious sector. The former are about carrying out menial tasks, while the latter are about charting a course towards growth, social and economic “revitalization,” and overall progress. The former are mostly related to maintenance. The latter are seen as the creators of value.[7]
One may ask, of course, what’s wrong with robots doing all the dirty jobs? And the answer is that such a scenario — one of robots doing all the work — would, in fact, be very appealing. But our technology is nowhere near the place where robots are capable of doing all the work. Robots require maintenance, adjustment, oversight, and operation that relies on human labor. And that labor, surrounding the robots, gets concealed behind the illusion that robots do everything on their own. The fact is that Japan would still suffer from the same social ills, and that somebody would still need to take care of the maintenance, service, and care-taking associated with those ills. Japanese society is still aging, the population is still decreasing (even if at a slower rate) the rural areas are still being depopulated, there is still a shortage of unskilled labor, and the Japanese government is still reluctant allow a higher immigration quota.
The new Robots for All offers the appearance of a solution — or, one might say, the solution of appearance. Humanoid robots provide 4IR policies with an adorable and admirable mask to conceal the undervalued yet much-needed maintenance labor. Their humanoid shape suggests that they are capable of more functions than the technology actually allows, misleading humans into thinking that there is no human labor involved. Humanoid robots would make it appear as if all the work is done magically, automatically, without any need to think about who is doing the job, and what price society pays for it.
The fact that humanoid robots conceal human labor becomes clear when we consider remotely operated robots, in which all the functions that robots still cannot perform on their own are controlled remotely by a human. According to widespread advertisements, humanoid robots would be remotely operated by people with Amyotrophic Lateral Sclerosis to serve people at a distant coffee shop.[8] The robots would offer those confined to their beds much-needed social interaction, a sense of place in a society, and even some cash. In the promotional video by Dawn, the coffee shop that employs remotely operated humanoids recently designed by Nagoya Institute of Technology, a beautiful young woman remotely operates a robot by moving her eyes alone, and is visibly touched by customers’ gratitude seen through robot’s sensors. If, at any point, the viewer begins to wonder how so much effort and money was invested in helping the 9,557 Japanese ALS sufferers (out of almost 127 million total population), they can be reassured that this technology can be helpful to any of the 34,944,000 people “who experience difficulties getting out of the house.”
We don’t have to doubt the sincerity of engineers’ motivation to ask how this noble technology might be used elsewhere. The fact is that this technology relies on the labor of a hidden human. It is a 21st century incarnation of the 18th century Mechanical Turk, an automaton which was purported to play chess but was secretly operated by a human hidden inside the pedestal on which it sat.[9] Could this technology of concealment possibly function to keep in place darker social dynamics, to sweep under the rug an array of social ills that require a social solution? Could it be used to reinforce those ills by keeping specific bodies hidden and sequestered from society?
It is not difficult to envision a positive answer to the above question by looking at a different iteration of the same technology, such as Ugo, the remotely operated laundry-folding robot by Mira Robotics.[10] So far, the robot is quite slow, imprecise, and requires a human to avoid stumbling over its power cord. But let’s assume designers will work those details out. There are also some privacy issues: for instance, the remote human operator can actually see inside somebody’s house. However, issues like this one could be solved by literally keeping the robot in the closet.[11] At first glance, this technology does not have to be malicious — one can imagine, for example, that same woman with ALS from the Dawn video feeling grateful for the opportunity to fold others’ laundry. (And given the fact that the robot in the Dawn video was named OriHime, or Folding Princess, one really does wonder if the original plans for the waiter-robot were perhaps more about laundry-folding than serving coffee.) But there are simply not enough ALS sufferers to staff all the jobs that would require remote operators. Not to mention that confining them to a closet would defeat the noble goal of offering them human interaction. So, most likely, the operators of the laundry-folding robots would come from the ranks of the elderly, forced back to work due to poverty, women denied work opportunities outside the home, and, even more probable, cheap laborers from outside of Japan.
And here we finally see how humanoid robots can serve to preserve a conservative vision of Japanese society by concealing the labor of humans who count for less. Robots would absolve those Japanese citizens who are valued as 4IR participants from doing the labor necessary to maintain a functioning society (or even from acknowledging that such labor exists). There would be no reason to encourage men to share household responsibilities because their share would be done by somebody else — albeit under the guise of a machine. There would not even be a need to interact with the actual humans providing the labor. One would not need to feel guilty about the fact that an elderly person folds their laundry. One would not need to be reminded of the social structures that keep women at home by having to face a mother who does her third shift at somebody else’s home. One would not need to let the poor into one’s home, or even see them. And if there are not enough Japanese elderly and women due to the dwindling demographics, remotely-controlled robots would allow Japan to employ foreigners without the need for foreigners to actually set foot on Japanese soil. Nobody would need to worry about whether they were paid fairly, or whether they received benefits. Nobody would be required speak to foreigners or see their faces. Those faces would be concealed by a cute robotic smile and a perfect — albeit robotic — Japanese accent. People would be able to interact with the robot and forget (if they knew in the first place) that behind this robot there is a living human being who is doing the dirty but essential work.
This scenario is not inevitable. Robots, even humanoid, and even remotely-operated, do not have to become tools through which oppressive social structures are reinforced. As with the robots targeted to ASL sufferers, they can be a blessing. Yet the ease with which such robots could be exploited to conceal and abuse the labor of living, breathing human beings is a likely dark spot on the otherwise rosy visions of 4th IR.
[1] Jennifer Robertson analyzed the problematic assumptions underlying this policy in 2007, even before actual technological developments fell short of the overly optimistic predictions. (See Robertson, Jennifer. “Robo sapiens Japanicus: Humanoid Robots and the Posthuman Family.” Critical Asian Studies 39, no. 3 (2007): 369–398.)
[2] https://www.youtube.com/watch?v=Gews8sQE9LY#action=share
[3] http://www.meti.go.jp/english/press/2016/0210_04.html
[4] http://www.cao.go.jp/innovation/action/conference/minutes/20case.html
[5] This is bias #3 in the series of technology-related biases detailed by Rodney Brooks in “The Seven Deadly Sins of AI Predictions.” MIT Technology Review Vol 120 num. 6 (2017): 79–86. https://www.technologyreview.com/s/609048/the-seven-deadly-sins-of-ai-predictions/
[6] http://www.meti.go.jp/english/press/2017/0915_002.html
[7] This division is strikingly similar to the Victorian views described by Norton Wise in “The Gender of Automata in Victorian Britain,” in Genesis Redux: Essays In the History and Philosophy of Artificial Life. Edited by Jessica Riskin. Chicago: University of Chicago Press, 2007: 163–195. Although the division of labor into “tasks” and “revitalization” does not explicitly defined along the gender lines, like in the case described by Wise, we cannot ignore the de facto gendered nature of many of the tasks in question.
[8] https://www.youtube.com/watch?v=lUuzfplbwwg
[9] See Lilly Irani’s astute analysis of the Amazon Mechanical Turk system in “Hype, Profit, Labor, and Agency in the Shadows of the Fourth Industrial Revolution” on this platform. AMT relies on undercompensated human labor while presenting the fruit of this labor as a result of algorithmic output.
[10] https://www.youtube.com/watch?v=YbdSIsBbV7o
[11] Evan Ackerman, “Remotely Operated Home Robot Can Do Your Laundry,” IEEE SPECTRUM, Feb. 7, 2019. https://spectrum.ieee.org/automaton/robotics/home-robots/remotely-operated-home-robot-can-do-your-laundry?fbclid=IwAR0d2pI_r_ikLV9XY7tWYOWnv2Fz9hl8fwThr4Ua8OqhqLSQl6OVs6fxLkk&utm_medium=facebook&utm_source=dlvr.it
