Get Access to Print and Digital for $23.99 per year.
Subscribe for Full Access
May 2023 Issue [Easy Chair]

Where Tomorrow Meets Today

“Let me persuade you to come to the place where tomorrow meets today,” a voice-over invites near the beginning of Design for Dreaming, a General Motors promotional film from 1956. A masked Lothario whisks a young woman out of her bed to a car show at the Waldorf Astoria. It’s a magical setting, and there are many wonderful automobiles to choose from, but, just as she’s stepping from a Cadillac, an apron appears over her gown. “Better get her into the kitchen quick!” the narrator quips.

Sure enough, our heroine finds herself baking a cake, but she’s doing so in the Frigidaire Kitchen of the Future, where “push-button magic” makes light work of any domestic task. A screen displays a lurid confection, complete with a list of ingredients. The woman pops her cake tin under a transparent oven dome and reappears in tennis gear, ready to play: “Ticktock, ticktock! I’m free to have fun around the clock!” The cake duly appears, perfectly cooked and replete with icing and candles. She gets into a silver sports car and is driven off by her beau on the “highway of tomorrow.”

Midcentury visions of lives freed from drudgery by beneficent machines now seem somewhere between quaint and sinister. I can’t remember the last time I heard the phrase “labor-saving devices” used unironically. We are far enough into the age of automation to know that no leisure awaits us down that particular highway. Instead, we face an arms race in which each advance seems to lead to more acceleration, more intensification: more labor, not less.

Historically, what freed upper-middle-class women from domestic work wasn’t push-button magic but their entry into the workforce. As the twentieth century gave way to the twenty-first, double-income families were employing domestic workers to use the shiny machines in their kitchens, and the idea that automation might put an end to work came to seem less like a promise than a threat. What is to be done about those whose jobs are better performed by machines? Should they be supported? Should we hope this “surplus population” just withers away? Should measures be taken? And here right-thinking technocrats begin to whistle and look at their hands.

Even the most sanguine champions of technological advancement have felt compelled to reckon with the problem. In 2014, the digital gurus Erik Brynjolfsson and Andrew McAfee published The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies. Blurbed by an array of Silicon Valley luminaries, from the venture capitalist Marc Andreessen to Google’s chief economist Hal Varian, the best-selling book tells an exciting story about automation, but it also sets out a now-familiar dualistic picture of the future. First there is the upside: “We’re heading into an era that won’t just be different; it will be better, because we’ll be able to increase both the variety and the volume of our consumption.” Then there’s the downside: “Technological progress is going to leave behind some people, perhaps even a lot of people, as it races ahead.”

Books of this ilk often quote a figure derived from a 2013 Oxford University study, which suggests that 47 percent of jobs in the United States are “at risk” from automation. This picture has become received wisdom. Few would doubt that we’re in a time of profound transformation, and that there are winners and losers—video-rental store owners are the handloom weavers of the early twenty-first century, figures of nostalgic pity. Are we approaching a profound singularity, an “inflection point,” as Brynjolfsson and McAfee term it, “where many technologies that used to be found only in science fiction are becoming everyday reality”? Or are we experiencing something less grand and eschatological, something more like the introduction of dishwashers to the home kitchen? That seems open for debate. However, there’s an underlying assumption that isn’t addressed: that automation is accelerating economic productivity. The dominant automation story takes as fact the idea that machines are helping us make more valuable stuff and perform more valuable services—and why wouldn’t we believe it, when the culture bombards us with the extraordinary spectacle of “brilliant technologies” succeeding brilliantly and creating brilliant fortunes as if out of thin air.

We are experiencing real material changes brought about by foundational technologies such as GPS and the TCP/IP stack. Our social lives and habits of consumption are in flux. Yet despite all this, there’s a ghost at the feast. The tale of a second industrial revolution driven by digital automation ignores an inconvenient economic reality. Fifty or more years into the information age, we ought to be seeing the exponential takeoff in productivity that we were long promised. Instead we see the opposite. As critics such as Jason E. Smith (Smart Machines and Service Work) and Aaron Benanav (Automation and the Future of Work) demonstrate, there are signs of stagnation across the developed world, where real wages for most workers have remained at the same level since the Seventies, the end of a period of growth and reconstruction that followed World War II. We may be in a period of change, but it’s not a given that it’s a period of growth.

The reasons for this stagnation are—unsurprisingly—complex, and economists disagree about what’s to blame. One reason that the second industrial revolution doesn’t appear to have taken off in the manner of its steam-powered predecessor is that developed economies have deindustrialized. When we think about automation, chances are we picture a robot making a car. But in rich countries, few people work in manufacturing, and the sector accounts for a shrinking share of economic activity. Eighty percent of American workers perform “services”—a vague and rather outdated catchall term for everything we do economically that isn’t mining, construction, manufacturing, or agriculture. Ninety-six percent of employment gains since the turn of the century have been in this category. Projections suggest that most new jobs will be in the service sector, and that much of that growth will come from low-paid positions in the field of health and social assistance. Such activities—providing complex forms of physical and emotional care—do not lend themselves easily to mechanization.

The latest turn in automation is toward worker management—that is to say, intensifying the regime of surveillance and control in the workplace. Bathroom breaks can be monitored. Paces walked, calls made, units completed—data on all these things can be collected, so targets can be set. The idea, as old as the Taylorist factory system, is to increase productivity by squeezing more work out of each individual worker. In January, Davos attendees were treated to a presentation on “brain transparency”: neurotechnology for workplace monitoring. A dystopian vision of the boss having twenty-four-hour access to the worker’s inner mental state was followed by the revelation, now a cliché in post–TED Talk business presentations, that “it’s a future that’s already here.”

None of this feels like the kind of innovation characteristic of a boom, let alone an earth-shaking singularity. Shaving a few minutes of extra attention-time from your nano-managed workforce is not exactly a Promethean bending of the world to your will. All this perhaps explains the seams of anxiety and disorientation that run through twenty-first-century automation discourse. The world we’re experiencing doesn’t feel like the world that’s being described.

The figure of the so-called care robot, a recent media preoccupation, exemplifies the knotty complexities of what’s actually at stake. They are often presented through stories about Japan, where the aging demographic has led to a focus on this kind of research—allowing Western journalists to sprinkle a certain pleasing fairy dust of Orientalism over debates about neglect and exploitation. But the phenomenon is global. Last year, New York State contracted an Israeli robotics firm to provide its Office for the Aging with more than eight hundred “empathetic care companions.” Health care is labor-intensive and wages are often the most substantial cost, so there’s a strong incentive to automate. There are clearly ways in which automation can help in health care settings; exoskeletons that can assist staff with lifting and turning bedbound patients and robot porters that can fetch and carry equipment are both likely to appear in well-resourced hospitals in the near term. It’s in the specific domain of care that the anxieties intensify.

Few commentators seem to argue that any kind of robot would exceed the ability of human carers. They’re a “good enough” solution, a fix for health care systems rather than for patients. The people they would displace (or, if you’re feeling kinder, whose work they would augment) are relatively powerless. Home assistance work is often done by immigrants and women of color. It requires little or no formal education, but it’s far from unskilled, and its modalities are, to put it mildly, not easy to automate. It demands kindness, tact, compassion, the ability to assess and improvise, and a talent for intuiting the needs of people who may have difficulty expressing them. We think of care as a foundational ethical principle, and also as a practice, an activity that has an effect on the caregiver as well as the recipient. What does it mean to remove the element of human exchange? The image of a lonely or vulnerable person reliant on a robot is a staple of dystopian fiction for a reason.

Care is one of the pillars of the human, a last bastion of uniqueness in the age of intelligent machines. Another is creativity. With the sudden explosion of generative AI models, we now have systems that can produce images, writing, and music in response to simple natural language prompts. This appears to be a new incursion into the eroding territory of the human. Until recently, we’ve assumed that only low-status tasks would be assigned to machines. The automation of creative work has revealed to a whole new class of people that they are not immune to Schumpeterian creative destruction. Illustrators are now forced to compete with models that have ingested their work as training data. Blurred signatures and even stock watermarks come up in generated images, making it painfully clear that this is a kind of remixing rather than ex nihilo creativity. Of course, all artists, human or artificial, exist within a matrix of influence, but that seems too dignified a term for the process of scraping the internet and reproducing elements of stolen material without attribution or payment.

The new large language model chatbots appear set to displace human workers who answer customer support questions, code, write and edit copy, trade, conduct market research, and perform various kinds of paralegal and financial work. Perhaps an LLM trained on the recorded interactions of home care assistants and their clients could even simulate companionship. If, hypothetically, a machine could be empathetic, or rather, if it could convincingly simulate for a human the experience of being empathized with, then perhaps we could see the automation of some forms of caregiving. Will these models finally set the revolutionary automation boom in motion? In October 2022, a mental health non-profit called Koko, whose model is a peer-to-peer chat service, experimented with allowing users to generate draft messages using an iteration of OpenAI’s GPT-3. As reported by Gizmodo, “the 30,000 AI-assisted messages sent during the test received an overwhelmingly positive response, but the company shut the experiment down after a few days because it ‘felt kind of sterile.’ ”


More from

Close
“An unexpectedly excellent magazine that stands out amid a homogenized media landscape.” —the New York Times
Subscribe now

Debug