Book review by Teresa Heffernan (forthcoming in Novel)
Debates about whether robots will take over jobs or open up as yet unimagined career possibilities dominate the headlines. Silicon Valley and the techno optimists promise us that robots will automate boring jobs and create new ones, leaving humans free to pursue their interests in the arts and sciences and ushering in a great era of equality, creativity, and freedom.
Others warn that robots will take over close to half of all human jobs dramatically increasing unemployment. Those owning the machines and platforms will throw workers into poverty, increasing the already unconscionable gap between rich and poor and further ripping apart the social fabric of democracy. These competing scenarios typically frame questions about the impact of robots on labour in world economic forums and in the media.
Jennifer Rhee’s The Robotic Imaginary: The Human & the Price of Dehumanized Labor (University of Minnesota Press, 2018) interrupts this debate to ask more basic questions about how robot labor is imagined by research labs, by the artificial intelligence industry, and in film, art and literature. Bringing this technology into conversation with cultural and literary studies and the humanities, Rhee considers the ways in which it envisions the historical and current understanding of what it means to be human. Organized around chapters on caring and care labor, thinking and domestic labor, feeling and emotional labor, and dying and drone labor; Rhee’s book is concerned with how the contested terrain of the human is constituted and reconstituted by these new anthropomorphic technologies. This labor imagined in robotic form renders the human knowable, calculable, and recognizable while exposing the dehumanized others that exist outside of the boundary of what is considered familiar and normal. Each chapter concludes with a short review of robotic art that offers an alternative imagining, a reconfiguring of the human as unknowable, particular, and irreducible.
The introduction offers an overview of the origins of robotics, which found its first expression in literature, was developed by scientists, and grew with military funding. The term artificial intelligence emerged out of the Dartmouth Project, which brought together a small group of men in 1956 to debate the hypothesis that machines could be made to simulate human intelligence. The collapse of the human and the machine, the anthropomorphic metaphor underpinning the field, expands and continues to expand the boundary of the human beyond this initial metaphoric union, Rhee argues, invoking Paul Ricoeur’s description of the workings of metaphor. The other critical factor shaping robotics has been DARPA (a branch of the American Department of Defence devoted to technological and military superiority), which has funded most of the research in the field since its creation in 1958.
Two of the founding texts in the field, Alan Turing’s test for machine intelligence and Masahiro Mori’s theory of the uncanny valley, illustrate Rhee’s central argument. In the first example, the imitation game begins with a man and a woman who are both trying to convince a judge via a teleprinter that they are female while the judge, who is in a separate room asking questions, tries to correctly identify the woman. Turing then suggests replacing one of the humans with a computer. The game is famously set up to police the boundaries between the human and the machine, but, as Rhee points out, the judge needs to conceptualize the human before s/he can possibly assess human likeness. Hence the game also opens up the possibility for the judge to misrecognize the human rendering the very category “human” unstable and open while exposing the biases and normative assumptions at the heart of this policing exercise. In contrast, Mori’s theory of the uncanny valley, which sets out to determine the robot design that people would best relate to, enforces narrow normative versions of the human, Rhee observes, that are measured against disability and illness. In several graphs, Mori charts the point at which human-like replicas evoke positive affinity as opposed to eeriness. The “healthy” human occupies the highest point on the graph, while the corpse falls at the bottom of the stillness scale and the zombie at the bottom of the movement scale and the ill person gets slotted below the healthy one. In another of his graphs a prosthetic hand occupies the point of negative affinity. As Mori’s theory is in wide circulation and impacts the development of humanoid technologies and social robots, it is important to expose the biases informing his design model Rhee insists.
Karel Čapek’s play R.U.R (1920) first uses the term robot, derived from Czech words for serfdom and forced labor, long before the development of the field. Driven by the capitalist goals of profit, productivity, and efficiency, designers of organic humanoid robots promise to liberate humans from labor and usher in a new era of freedom and leisure. The question of the robots’ “humanness” drives the play as Helena Glory hopes to liberate them from exploitation while their creators argue they are nothing but soulless machines. The play draws on the cultural memory of slavery and fears of slave rebellions to explore the dehumanization of workers under factory capitalism that promises freedom for some at the expense of others. Alienated from their labor, however, the humans in the play fail to thrive and stop reproducing while the robots, claiming their “humanity” by mimicking human’s capacity for domination and violence, revolt and kill all the humans.
Rhee returns to these founding literary and scientific texts in order to open up the entwined questions of anthropomorphization and dehumanization that frame the next four chapters of her book. Chapter one considers Turing’s model of AI as a child that needs to be educated and Weizenbaum’s early “therapist” ELIZA, demonstrating how care labor has been integral to AI. Gendered female, these often humanized AIs serve as emotional interlocutors, child educators, and romantic partners or spouses that perform both domestic and affective work. In contrast, “male” AIs, like Watson, are machines that are positioned as universal experts that disseminate knowledge in fields like medicine and law. Analyzing Richard Powers’ Galatea 2.2 and Spike Jonze’s Her, Rhee maintains that the gendering of AI thus replicates the historically devalued and underpaid reproductive labor of women that has sustained capitalism. Countering this devaluation, however, Rhee points to robotic art “that highlights affect’s constitutive role in cybernetics, transforming cybernetic circuits of communication and control into those of affect and care” (57). Nam June Paik’s Robot K-456, Norman White’s The Helpless Robot, Momoyo Torimitsu’s Miyata Jiro, and Simon Penny’s Stupid Robot and Petit Mal are presented as examples of robot artwork provoking affective responses from their audiences, demanding that cybernetics be grounded in an ethics of care and interdependence, and foregrounding these traits as critical components of being human.
The second chapter on “thinking” further builds on the marginalization of reproductive labor in the field of AI. Early closed-world versions of AI that relied on highly schematic and simplified models of reality were followed by the hope that the combining of multiple “micro-worlds” would lead to greater complexity in AI systems. Rhee argues that the micro-world approach of AI, which is built on stereotypes and familiar norms and erases the unruliness of the real world, finds it parallel in The Stepford Wives. Ira Levin’s 1972 novel, inspired by Betty Friedan’s The Feminine Mystique, famously recounts the murder of women and their replacement with docile immaculate generic robots that are programmed to do housework and serve their husbands. Like the closed-world AI models, the female robots remain sealed off from the public world of wages, politics, and intellectual work while real women with their complicated desires, politics, and aspirations must be killed off in order to sustain the unchanging ahistorical gendered hierarchy of Stepford.
Yet Rhee also argues that the fate of the real women in Stepford is sealed in part because of their refusal to acknowledge the working-class women, who as “outsiders” of the suburban enclave, are able to document the crimes committed in the area. Concerned with the fate of middle and upper class white housewives, Friedan’s work also ignores the many white working class women, single women and women of colour who were working outside the home in jobs that offered neither economic self-sufficiency nor independence from men, as bell hooks has noted. Moreover the presentation of domestic labor and child rearing, the task of raising another human being, as unskilled and “mindless” perpetuates the devaluation of “women’s” work. The Stepford Wives and its contemporary adaptation, Ex Machina (2015), highlight the exclusionary and at times exploitive narrative of white middle-class feminism that finds racialized and classed women aiding white women’s liberation even as they are excluded from it.
Rejecting the symbolic micro-world models, Rodney Brooks developed an embodied approach to robotics in the 1980s that encouraged robots to interact with messy dynamic environments to develop machine “intelligence” with the hope that they would “evolve” upward to humanoid AI. Yet while Brooks’ robots are physically situated in the world, they, as several critics have pointed out, are culturally and historically “dumb,” perpetuating the closed world approach to AI. In addition to military robots, Brooks’ company iRobot designs autonomous robots, like the Stepford robots, as mindless domestic laborers. In contrast to closed-world AI, Rhee concludes this chapter with several examples of robotic art—including Stelarc’s Fractal Flesh and Ping Body—that stress interdependence, open worlds and the vulnerability of the body.
In the third chapter, one of the most fascinating, Rhee explores social robots and emotional labor as another aspect of devalued reproductive labor and its ties to the military. In the 1990s, with new research on the importance of emotions in intelligence, robots, funded by DARPA, were developed based on the contested theory of “universal” emotions. Rhee argues that both the myth of universal emotions and the work of producing legible emotions are ways of policing the boundary of the human. Technologies developed from this theory that assume the external body reveals the truth of the individual, such as SPOT (screening of passengers by observation techniques) adopted by US Homeland Security, not only have had little success but expose the power relations embedded in them. Rhee explores the gendering and racializing of emotional labor and the dehumanization that is perpetuated by these technologies in her reading of Philip K Dick’s We Can Build You and his later novel Do Androids Dream of Electric Sheep. The Voigt-Kampff test, at the heart of this latter novel, imposed by those in control, measures emotional responses to scenarios or images to determine the “human” status of the responder. The test of course is never used on its android hunters and Deckard’s sense of shame in brutally eliminating the androids at the command of the state remains inside him in any case and is never visible on the surface. The chapter concludes with two feminist robotic works, Omo by Kelly Dobson and Swarming Emotional Pianos by Erin Gee, which challenge the theory of the universality of emotions and its use in developing dehumanizing policing technologies.
The final chapter on dying considers the entanglement of reproductive labor and drone warfare. Targeting victims and perpetrators outside of any judicial system and under a veil of secrecy, drone warfare perpetuates the colonial and racial legacy of determining who gets included and who gets excluded from the category of human, which has been part of both post Enlightenment subjectivity and US labor history. Rhee reviews American drone policy that identifies any military-aged man in certain areas as the enemy and that refuses to investigate those killed in the strikes or accurately document civilian deaths. She also reviews the history of cybernetics as a “war science” and Norbert Wiener’s early work on defense systems, which encouraged fighter pilots to identify with cybernetic German pilots to better understand the enemy other. The racialized Japanese enemy, however, were characterized as insects and vermin rather than as cyborgs, so no identification was encouraged. This dehumanizing racialization continues not only in drone policy but also in the asymmetry of drone targeting fueled by the massive gulf between operators and their targets, viewed as “ants.” From the high accident rate of the machines to the “ambiguous” information that ends with dead civilians, the technology also reveals itself as highly fallible exposing the misguided faith in technological omnipotence and quantifiable information that drives this form of warfare.
Reviewing drone art, Rhee provides a provocative analysis as she unpacks the differences between works that invite their western audiences to identify with racialized targets and those that challenge that identification in order to underscore the legacy of racial violence in America. She points to the limits of art works that promote identification with those “over there” by invoking Judith Butler and her questions about whose lives count as grievable. Positioning America as a place of safety and justice, works such as Home Drone and Drone Shadow fail to acknowledge the continuity between drone strikes overseas and the violence and injustices inflicted on marginalized communities at home, a point driven home by the adoption of militarized robots by some local US police forces. In contrast works such as Teju Cole’s Seven Short Stories about Drones refuse to ground ethics in familiarity and identification and instead insist on mourning lives that are unknowable. The artistic collective behind #NotaBugSplat, James Bridle’s Dronestagram, and Omer Fast’s film 5,000 Feet is the Best also suggest, Rhee argues, “an ethical relationship that foregrounds disorientation, uncertainty, and the unknown rather than the familiar, the known, the predictable ” that direct cybernetic technology and drone warfare (172).
The Robotic Imaginary exposes the ways in which robot technologies perpetuate existing racial and gender hierarchies by devaluing certain labour and certain humans and valuing others while exploring robotic art as way of opening imaginings that challenge the colonial, patriarchal, class and racial histories. As robots invade work spaces and as privatization erodes social responsibility, Rhee rightly insists we should ask of every robot figure “who is being dehumanized?” And what version of human is considered “sacrosanct and familiar”? While automation and the restructuring of the labor force by multinationals like Google and Facebook that are buying up AI and robotic technology lies outside the argument of Rhee’s book, I did wonder about the very limits of the metaphor of the human as machine and whether dehumanization doesn’t begin with industry leaders in Silicon Valley who have so successfully propagated the view that there is no difference between the two. Rhee’s otherwise excellent reading also falls a little short in its American-centric focus. What, for instance, would she have to say about Japan’s embrace of the “robot revolution,” in lieu of immigration, that is trumpeted in the face of a shrinking labor force? Or about the global fight for control of AI.
A vital contribution to the field, Rhee’s book does not argue fiction is “coming true” as is so often the case in scientific and media reports on robots, but instead it turns to literature and art as providing some insight into the always shifting ground of what it means to be human. Rhee’s book is essential reading for anyone negotiating the intersections of literary studies, anthropomorphized robotics and the impact of these technologies on society.