Over the last couple of years, artificial intelligence has been depicted in a surprising number of films—including Autómata (2014), Avengers: Age of Ultron (2015), Big Hero 6 (2014), Chappie (2015), Ex Machina (2015, pictured above), キカイダー REBOOT (2014), Robocop (2014), Terminator Genisys [sic.] (2015), and Transcendence (2014).[i] Although not uniformly apocalyptic in tone, these movies collectively embody a broader current of disquiet about the ethical dilemmas and potential dangers of machine sentience. Nor are these anxieties confined to the typical movie-going public. July 28, 2015, Stephen Hawking, Noam Chomsky, and a host of other luminaries signed an open letter cautioning against a “global AI arms race” and calling for a ban on “autonomous weapons.” In doing so they lent their weight to the idea that in some respects these cinematic fantasies might be well founded.
But for many social theorists, the graver contemporary threat is not killer robots, but worker robots stripping us of employment and purpose.
In this rather long post, I place contemporary robot-movies in dialogue with an essay by John Maynard Keynes (“Economic Possibilities for our Grandchildren“) in order to explore a particular contemporary capitalist nightmare—that the very technological innovation that has undergirded modern productivity might in fact be its undoing. More specifically, I want to see how dystopian readings of automation (in particular the automation of mind or machine sentience) contribute to fears of the loss of sleep and the loss of work. Technological optimists should fret not as in a later post I’ll look at acceleration as utopian liberation. But if you follow me across the fold I’ll examine acceleration as despair or we might say as the death drive of the robot apocalypse.
Killer Robots versus Buddy Robots
The recent film that best illustrates the issue I want to address in this blog post is the Spanish science fiction/thriller Autómata directed by Gabe Ibáñez and staring Antonio Banderas. By way of summary: Autómata is set in 2044 after an environmental catastrophe has wiped out most of the world’s population. The remaining humans largely manage a robot proletariat that makes up for the lost labor. In a transparent homage to Asimov’s Three Laws of Robotics, these humanoid robots have two unbreakable protocols—they are incapable of harming living things and they are incapable of modifying themselves. The protagonist is a robotics insurance agent who investigates what appear to be violations of these rules. Eventually he gets caught up in a robot uprising (or more properly exodus).
Despite its fascinating premise, on aesthetic grounds Autómata is a mess. The pacing is terrible, the dialogue is hackneyed and overly expository, and there is an offensive borderline romance between the protagonist and a fembot prostitute who resembles a piece of Ikea furniture with limited conversational skills (far be it from me to judge fetishes but the assumption that the viewer is supposed to be remotely titillated by this robot suggests that the filmmaker’s erotic fantasy is basically a Klippen-Lalle that can dance, LINK for the brave or the bored).
In most respects Autómata is not a successful film. But it expresses an intriguing moral muddle that cuts to the historical roots of the whole subgenre. To explain, AI films can
usually be divided into killer-robot movies that portray sentient machines as
fundamentally hostile to humanity (e.g. the Terminator franchise); and little buddy-robot movies that imagine sentient machines as sidekicks (e.g. Big Hero 6). The difference often comes down to the moment in the film in which the artificial intelligence achieves consciousness—does the machine try and take over the world (perhaps shouting, “Kill all humans”)? Or does it act like a child (perhaps making cute little bleeps or babbling in Pidgin English)? (As a parenthetical, it is worth noting that both sorts of films often turn dystopian at the moment the AI is exposed to capitalism: either corporate greed is the origin of the evil robot in the first place (Skynet) or the cute robot needs to be saved from capitalist/military exploitation).
Autómata is interesting because it is stuck between the killer robot and little robot buddy genres. It begins little buddy-style with a pidgin-speaking robot being executed by a police officer at the moment of self-reference, and indeed it spends most of its runtime portraying robots as childlike exploited labor. But this is where it gets into its mess because having identified the robots as slaves it must celebrate the slave revolt, which is depicted climactically in the film when a robot throws off its second protocol and kills a human slave driver. Hence the film is asking us to sympathize with robots as they kill us.[ii] To be sure, Autómata dodges the issue by having the robots immediately set off on an exodus into the desert (there is even a robotic version of Moses), but an inevitable confrontation between humans and robots is still implied if differed.[iii] In effect, this is a sympathetic or even cute robot apocalypse.[iv]
This sounds like a contradiction, but it is actually closer to the origin of the subgenre, such that one can see the whole field of films as suspended within this seeming paradox. To explain, “robot” entered the English language by way of the 1920 play Rosumovi
Univerzální Roboti (Rossum’s Universal Robots) by the Czech playwright Karel Čapek. The term comes from the Czech robota “forced labor, compulsory service, drudgery” and hence was intended to mean “slave” or “serf.” As you might guess from this etymology, Čapek wanted the audience to identify with his robots. The play also ends with a sympathetic revolt in which the robots kill all humans. But it should be noted that Čapek’s robots are not mechanical automata but synthetic biological beings. They are indisputably imagined as alive. So the robot uprising is not the culmination of a war between man and machine, but the liberation of the exploited class or perhaps the arrival of humanity’s children. Autómata (like other contemporary analogues) by contrast depicts mechanical robots. By triumphing the robot apocalypse, it seemingly suggests that it would be moral for us to be executed by a rebellious microwave or more precisely it longs for the day when we liberate our self-driving cars.
In summary, Autómata tries to glorify a flight of the technological, imagining a semi-incoherent exodus of machines from the human life-world. But it has already foreclosed all the potentially utopian aspects of this fantasy—the film is incapable of investing in even a naïve longing for the primitive past, it seemingly rejects or overlooks the possibility of singularity or transcendence of the human common to many AI apocalypses, and it also demonizes the possibility of resistance to the robot revolt. [v] In that sense Autómata is a more chilling expression of the Freudian death-drive than any Terminator movie and it can’t help but suggest a submerged longing for the apocalyptic end of humankind—inadvertently evoking voluntary extinction or species-wide euthanasia by way of cute robots.
It sounds farfetched and nobody is seriously suggesting death by AIBO. But what Autómata and company have captured is a current ambivalence about our relationship to technology. As I alluded above, however, for many people the real anxiety is not that we will be overwhelmed by wrathful terminators, but that the buddy-robots we have made to aid our labors will come to replace us. By way of foreshadowing, the problem is that capitalism has hitherto built its claims to preeminence on claims that it embodies the most efficient engine of productivity, but as we push beyond the old limits of automation it seems that capitalism is in danger of implosion on the same terms. It may be accelerating toward its own destruction.
* * *
In 1930 near the start of the Great Depression, the British economist John Maynard Keynes published the two-part essay “Economic Possibilities for our Grandchildren.”[vi] Intended as an attack on economic pessimism (from both Right and Left), Keynes optimistically predicted future living standards and working habits in about the year 2030. His basic point was that increasingly technological progress would lead toward greater productive efficiency and general improvement in the quality of life. We haven’t quite hit the date yet, but in many ways his essay might seem to be prophetic and contemporary economists have praised its accuracy in projecting average income levels among other things.[vii] But one of the predictions that seems most off-base is his assertion that technological innovation would decrease working hours, such that in the near future the average first-world citizen would only need to work 3-hours per day 5 days per week, spending the rest of their time in leisure.
The idea of a 15-hour workweek seems laughable. But Keynes had good reason to make the prediction in his lifetime, as contemporary economists have noted,“the average annual number of hours worked per worker fell by almost 30 percent between 1870 and 1930, both in Europe and the United States,” and should this pattern have held we would be close to Keynes’ target.[viii] But clearly that isn’t what happened.
As the economist Juliet Schor observed in the now classic work The Overworked American: The Unexpected Decline of Leisure (1992) since 1948, productivity has soared, such that twenty-years ago, we could “produce our 1948 standard of living (measured in terms of marketed goods and services) in less than half the time it took in that year.”[ix] But that productivity has not translated into reduced work hours. Indeed according to some measures, Schor suggests Americans may work as much on average as we did in the 1920s.[x] Moreover, domestic technological improvement (such as the washing machine and the vacuum cleaner) have if anything increased the standards of domestic labor, while the movement of women into the labor force has led to little compensatory reduction in unpaid domestic chores. Restated, not only are we working more hours in the office, but we are also spending at least as much time in unpaid work at home (e.g. the washing machine means not less work, but that we more regularly clean our clothing). Hence, we have in many senses been retrogressing.
Nor is this shift a historical blip. As Schor argues, there is good evidence that the average person under capitalism works harder and for more hours than in pre-capitalist societies:
“One of capitalism’s most durable myths is that it has reduced human toil. This myth is typically defended by a comparison of the modern forty-hour week with its seventy- or eighty-hour counterpart in the nineteenth century. The implicit — but rarely articulated — assumption is that the eighty-hour standard has prevailed for centuries. The comparison conjures up the dreary life of medieval peasants, toiling steadily from dawn to dusk. We are asked to imagine the journeyman artisan in a cold, damp garret, rising even before the sun, laboring by candlelight late into the night. These images are backward projections of modern work patterns. And they are false. Before capitalism, most people did not work very long hours at all.”
I don’t share Schor’s implicit nostalgia for the life of the medieval peasant. But I think she makes a cogent point when she argues that that despite radical advances in technological productivity (from the personal computer to the microwave to space satellites) average leisure time in America has generally decreased since the late 1940s.[xi] At the very least, we work much longer hours with fewer vacations than Keynes anticipated. Furthermore, this increase in labor has significant human costs.
In a more recent study, Jonathan Crary (2014), 24/7: Late Capitalism and the Ends of Sleep, observes that:
“Over the course of the twentieth century there were steady inroads made against the time of sleep-the average North American adult now sleeps approximately six and a half hours a night, an erosion from eight hours a generation ago, and (hard as it is to believe) down from ten hours in the early twentieth century.”[xii]
Crary goes on to describe the new time-world of contemporary capitalism, noting that instead of leading to a promised era of ease, contemporary technological culture has given birth to a new age of sleeplessness. He argues that we live in a 24/7 temporality of barely differentiated seasons and blurred nocturnal-diurnal rhythms. Moreover, instead
of alternating between periods of work and relaxing calm most of us are in some sense always at the beck and call of work, our supposedly off hours chiming with new messages and cellphone calls, our laptops facilitating a greater muddling of office and home, our vacations disrupted by Skype meetings, and so on. So it might seem that rather than leading toward Keynes’ idealized relaxation, technological progress has if anything robbed us of time or at the very least robbed us of true autonomy over our mental state.
But Keynes was prescient in a different way because in his 1930 essay, he also puts his finger on a very contemporary anxiety:
“We are being afflicted with a new disease of which some readers may not yet have heard the name, but of which they will hear a great deal in the years to come–namely, technological unemployment. This means unemployment due to our discovery of means of economising the use of labour outrunning the pace at which we can find new uses for labour.”
Few historians today would agree with Keynes in seeing the Great Depression as resulting from technological unemployment, but in recent years the issue of automation and its impact on labor has appeared as a matter of contemporary concern. There have been a host of significant monographs on the subject of which Martin Ford Rise of the Robots: Technology and the Threat of a Jobless Future (2015) has probably received the most press (for more see note).[xiii] In August 2015, the BBC ran a story titled “Will machines eventually take on every job?” citing an article that suggests that, “the coming wave of technological breakthroughs endangers up to 47% of total employment in the US.”
For a stark illustration of the issue, one has only to note that in one of the earlier of these studies Frank Levy and Richard Murnane, The New Division of Labor: How
Computers Are Creating the Next Job Market (2004) observed that technological advances were having a major impact on the labor market and that computers/automation would ultimately replace a host of jobs with simple and procedural rules. Nevertheless, they predicted that there were whole areas automation wouldn’t touch for a long time, including various forms of white-collar work and even certain blue-collar careers. [xiv] Their argument was basically that some fields were too complicated (or dominated by implicit knowledge) to be easily automated. They illustrated the difference between the two by contrasting a student adding a column of numbers and a bakery truck driver:
“The student doing addition is processing a set of numbers by consciously applying rules: 7 + 3 = 10. The rules are what allow the addition to be programmed. The bakery truck driver is processing a constant stream of information from his environment: visual information on traffic light signals, visual and aural information on the trajectories of children, dogs, and other cars, aural information on unseen vehicles (perhaps including sirens), tactile information on the performance of the truck’s engine, transmission, and brakes. To program this behavior, we could begin with a video camera and other sensors to capture the sensory input. But executing a left turn across oncoming traffic involves so many factors that it is hard to imagine discovering the set of rules that can replicate the driver’s behavior.”[xv]
This distinction seems a lot less reassuring a decade later in the face of Google’s self-driving car and Freightliner Inspiration—the first self-driving semi-truck to hit the road. What had once seemed impossible to program is now nearly within our technological
grasp. Truck driving, which according to the US census is the most common job in 29 states, now seems likely to fall under threat in the near future.[xvi] Other common professions have begun to see the effects of automation from janitors to waiters to a host of entry-level service jobs displaced by new technologies (think controversial but widespread self-checkout aisles and the automated attendants/operators that have already replaced many bank tellers and receptionists). We know that in the face of automation and outsourcing manufacturing has fled the first-world, but now it appears service-jobs are already in decline.[xvii]
As the American economist Jeremy Rifkin put it in The End of Work: The Decline of the Global Labor Force and the Dawn of the Post-Market Era (1996):
“We are entering a new phase in world history—one in which fewer and fewer workers will be needed to produce the goods and services for the global population… In the past, when new technologies have replaced workers in a given sector, new sectors have always emerged to absorb the displaced laborers. Today, all three of the traditional sectors of the economy- agriculture, manufacturing, and service-are experiencing technological displacement, forcing millions onto the unemployment rolls.”[xviii]
Already in 1996, Rifkin predicted that technology would displace labor leading to a “Decline of the Global Labor Force” and ultimately perhaps to the end of work as such.
How do we square the “overworked American” with the end of employment? Part of what is going on is the rise of what the Austrian social theorist Ivan Illich called “Shadow Work,” the so-called “unpaid, unseen jobs that fill your day.” [xix] Illich originally focused on the rise of increasing unpaid domestic labor, but one side effect of incomplete automation of the service sector means that customers are meeting machines halfway. For example, self-checkout and similar systems not only mean fewer jobs, but also more work for consumers. The degree to which we do this “shadow work” without renumeration, we lose even more of our precious time.
But a more fundamental explanation of the relationship between technological unemployment and increasing work hours (and pay stagnation) has been put forward by Erik Brynjolfsson and Andrew McAfee Race Against the Machine: How the Digital Revolution is Accelerating Innovation, Driving Productivity, and Irreversibly Transforming Employment and the Economy (2012) who argue that we are being forced to work harder because we are being required to compete against the technology. Automation instead of increasing leisure is putting downward pressure on workers to match the efficiency of the machines and also to work for less. Restated, the role of labor in our post-Fordian Era requires that we be continually flexible and open to retooling as our jobs are at risk of being displaced or lost to automation. There is a lot of evidence that as technology makes labor cheaper, capital makes more relative gains.[xx] Simply put, automation means that human employees have to labor for less and the profits go to the investors who purchased the technology in the first place. Restated, to keep their old jobs, management asks workers to work for less or risk being replaced by cheaper machines.[xxi] Thus, Brynjolfsson & McAfee argue we are in a “race against the machine.” In sum, like the folktale about John Henry and the locomotive, we are being asked to work harder to keep up. If that is the case, then humans seem destined to lose.
Plus, what had once seemed to be a problem mainly for the manufacturing and service sectors has also begun to push into the domain of traditional “knowledge workers.” Things might seem to be worse than even the pessimistic readers of Rifkin imagined. To provide a couple of examples:
Conventional approaches to medicine are already been revolutionized by “computer-aided diagnosis of disease.” At the moment, the technologies aren’t very developed and physicians are largely unthreatened (or even aided) by them, but automation has already begun to push into a host of specialized subfields (radiologists, lab techs) and even robotic surgery has emerged from the realm of science fiction. At some point, Siri or something like it may be the first step in a medical diagnosis as healthcare becomes increasingly automated.
According to some accounts, the legal field too seems at the cusp of a revolution. As business professor Thomas H. Davenport argued in a provocative opinion piece titled “Let’s Automate All the Lawyers”:
“The law is a profession based on rules, procedures, evidence, and precedent. It turns out that intelligent technologies are increasingly able to codify these decision criteria into automated and semi-automated systems.”
Davenport’s primary example is the field of E-discovery, which renders superfluous legal aides whose time was previously spent reading hundred pages of documents for information relevant to a particular legal brief. From my recent experience buying a house, it is easy to imagine a whole subfield of lawyers replaced by programs designed to generate boilerplate legal texts. The business of writing mortgage contracts, wills, terms of service, incorporation documents and whole host of similar tasks seems ripe for automation.
I could continue going through the professions as there is some evidence to suggest that everything from journalism (see also) to software engineering to teaching to middle-management might be threatened by automation in the coming decades.[xxii] There is some hope that new professions may appear to replace the old ones. Keynes’ had already argued that new technologies might merely result in a readjustment of the labor force. But for how long can these readjustments go on?
The real fear (or hope) is that this technological growth is exponential. A scary fable about the implications of this exponential growth can be found in Ray Kurzweil’s Age of Spiritual Machines (1999) and his discussion of “the legend of the inventor of chess.”
“The emperor had so fallen in love with [the Game of Chess] that he offered the inventor a reward of anything he wanted in the kingdom.
‘Just one grain of rice on the first square, Your Majesty’
‘Just one grain of rice?’
‘Yes, Your Majesty, just one grain of rice on the first square, and two grains of rice on the second square.’
‘That’s it – one and two grains of rice?’
‘Well, okay, and four grains on the third square, and so on.’
The emperor immediately granted the inventor’s seemingly humble request…
It was fairly uneventful as the emperor and the inventor went through the first half of the chessboard. After thirty-two squares, the emperor had given the inventor about 4
billion grains of rice. That’s a reasonable quantity – about one large field’s worth – and the emperor did start to take notice…. It was as they headed into the second half of the
chessboard that at least one of them got into trouble…
[This is because]:
…the doubling of grains of rice for each square ultimately equaled 18 million trillion grains of rice. At ten grains of rice per square inch, this requires rice fields covering twice the surface area of the Earth, oceans included.” [xxiii]
This fable probably sounds farfetched, but computer scientists like to remind us about Moore’s law, “the simplified version of this law states that processor speeds, or overall processing power for computers will double every two years.” For the last fifty-years this law has roughly held. Kurzweil’s argument is that we are still on the first half of the chessboard in terms of computing power, but if it continues the effects will amplify exponentially. He assumes a relatively straightforward extrapolation from current trends, something that has often been a dodgy proposition. But regardless a radical transformation of the labor market seems to be in effect.
I’d like to end with another image (although perhaps at this point I’m beating a dead horse). In 1983, another economist, the Russian-American Noble Laureate, Wassily Leontief described a pessimistic version of Keynes’ technological unemployment by comparison to the role of the horse in agriculture:
“Any worker who now performs his task by following specific instructions can, in principle, be replaced by a machine. That means that the role of humans as the most important factor of production is bound to diminish — in the same way that the role of horses in agricultural production was first diminished and then eliminated by the introduction of tractors…. Reduction in the price of labor — that is, in the real wage rates — can and in certain instances did postpone its replacement by machines for the same reason that a reduction of oats rations allocated to horses could delay their replacement by tractors. But this would be only a temporary slowdown in the process; improvements in the efficiency of tractors and other inanimate means of production can be expected to proceed without any limits.” [xxiv]
Basically, Leontief was arguing that just as agriculture had replaced workhorses with tractors, the computer revolution was in the process of replacing human labor, rendering
the worker superfluous. It might seem that without workers, there would be few consumers and the capitalist system would end up devouring itself.
To conclude, if the coming age of automation continues to fulfill this dystopian fantasy, capitalism having deprived us of sleep may be destined to deprive us of work. In so doing, it may prove its undoing.
[Note to readers: after a transition post, I’ll be shifting gears into utopian mode including utopian readings of automation.]
Thanks to Jackie Hidalgo and Jeremy Bellay for their suggestions and for contributing useful sources related to this post. Also, thanks to Dalena Frost for thoughts and proofreading.
[i] an older version of this post also analyzed Ex Machina, I have a bunch of thoughts about fembots too, but will have to save the issue for a different post or perhaps a chapter version of this…..
[ii] For a more sophisticated version of this muddle as gender politics see Ex Machina (which I could talk about more in a chapter version of this post???).
[iii] Although perhaps Hephaestus is the closer reference.
[iv] For one version of how robot apocalypse would actually go see http://what-if.xkcd.com/5/
[v] For more AI apocalypse and its roots in the Christian imaginary see Robert M. Geraci “Apocalyptic AI: Religion and the Promise of Artificial Intelligence”
[vi] Keynes, John Maynard. 1930. “Economic Possibilities for our Grandchildren,” The Nation and Athenaeum 48.3 (October 18): 96–98, 48.2 (October 11, 1930): 36.
[vii] Lorenzo Pecchi and Gustavo Piga. 2008. Revisiting Keynes: Economic Possibilities for our Grandchildren. MIT Press, Cambridge, MA. See esp. P.2.
[viii] Pecchi and Piga P.6.
[ix] Schor 1992, 2.
[x] Ibid, 1.
[xi] Why it has decreased is an interesting question. Schor argues that it has a number of causes, including the shift on the part of Labor Unions toward arguing for health and retirement benefits instead of less working hours and an increase in overall consumption.
[xii] Crary 2014, p.11.
[xiii] It is discussed in monographs by Erik Brynjolfsson and Andrew McAfee (Race Against the Machine: How the Digital Revolution is Accelerating Innovation, Driving Productivity, and Irreversibly Transforming Employment and the Economy, 2012 [this is the source which inspired the title of this blog post] and The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies, 2014), Nicholas Carr (The Glass Cage: Automation and Us, 2014), Martin Ford (The Lights in the Tunnel: Automation, Accelerating Technology and the Economy of the Future, 2009, and Rise of the Robots: Technology and the Threat of a Jobless Future, 2015), Jerry Kaplan (Humans Need Not Apply: A Guide to Wealth and Work in the Age of Artificial Intelligence, 2015), Frank Levy and Richard Murnane (The New Division of Labor: How Computers Are Creating the Next Job Market, 2004)and Jeremy Rifkin (The End of Work: The Decline of the Global Labor Force and the Dawn of the Post-Market Era, 1996).
[xiv] Example appears in B and M.
[xv] Levy and Murnane p.20, discussing Michael Polyani’s concept of implicit or tacit knowledge. As they went on to argue “Examples include driving a truck, cleaning a building, and setting gems in engagement rings. Computers do not complement human effort in carrying out most such tasks. As a result, computerization should have little effect on the percentage of the work force engaged in these tasks.” Ibid, p.48.
[xvi] See the BBC article above for connecting this observation to the issue of automation.
[xvii] Brynjolfsson & McAfee 2011, 24.
[xviii] Rifkin 1996, p.xvi-xvii.
[xix] See Ivan Illich, Shadow Work. Boston: M. Boyars, 1981. Illich’s argument has been recently updated in Craig Lambert,. Shadow Work: The Unpaid, Unseen Jobs That Fill Your Day. Berkeley, CA: Counterpoint Press, 2015. I’ve been inspired in the juxtaposition of automation and shadow work, by http://www.nytimes.com/2015/05/17/books/review/rise-of-the-robots-and-shadow-work.html
[xx] Brynjolfsson & McAfee 2011, 45-46.
[xxi] This has the effect of producing “employment polarization” in effect automation driving down the wages at the lower end of the economy, hollowing out the middle class, and increasing the profitability for a narrow sector of the top of the economic class.
[xxii] For teaching see also http://gizmodo.com/5167214/japanese-elementary-school-kids-now-being-taught-by-saya-the-robot
[xxiii] Referenced also in Brynjolfsson & McAfee 2011, PAGE?
[xxiv] Wassily Leontief (1983) “National perspective: The definition of problem and opportunity,” in: National Academies, The Long-term Impact of Technology on Employment and Unemployment: A National Academy of Engineering Symposium
Pingback: Capitalist Dystopia Part 3——Fissures in Paradise: First World Problems or Micro-dystopias, Bullshit Jobs, and the Seeds of Bourgeois Alienation | Absolute Disruption: Theory after Postmodernism