Allison Pugh is a professor of sociology at the University of Virginia and a 2019-2020 Berggruen Institute fellow.
The coronavirus crisis has transformed many jobs, but some of the most dramatic changes have been experienced by teachers, therapists and other face-to-face workers. Even before the pandemic hit, however, their work was being altered by technology and targeted by artificial intelligence researchers. It seems we didn’t need a virus for people to call for nonhuman providers, and in the post-COVID-19 future, that call might become ever louder. But these calls ignore the unique presence that humans have to offer, a presence that nonetheless has been unequally distributed.
Many jobs — primary care physicians, therapists, teachers, coaches, ministers, even high-end sales staff and business managers — rely on relationships, on an emotional understanding between the worker and other people. These jobs depend on the ability to connect with others: clients healing, students learning, employees motivated and engaged, congregants inspired or comforted. I call this work “connective labor,” and I have been studying it for the past five years, interviewing more than 100 people and observing them at work for hundreds of hours.
Practitioners tell me that the “magic” of connective labor — what makes it work — lies not in feelings of affection, care or even trust, but rather in witnessing or reflecting the other person so that they (the client or student or patient) feel “seen.” Computers can do the seeing — paradoxically, we often call that “personalization.” But it turns out that being seen by a human can confer a unique dignity.
At a cocktail party a few years ago, standing in a group of academics under a tent while a thunderstorm rumbled overhead, I met a radiologist. Writing this today, it seems surreal that we were able to chat casually in groups, but at the time, I remember thinking that meeting him was like meeting a polar bear floating on an ice floe dislodged by global warming. Radiology is one of those professions that futurists and pundits say is likely to shed a lot of jobs, because artificial intelligence has already demonstrated impressive pattern recognition in reading X-rays for anomalies.
The doctor told me he was working at the national level to try to change the way radiologists do their work, so that they would no longer simply discuss results with fellow physicians. The only way out of the path of the oncoming AI train, he said, was to make radiology more “patient-facing,” to bring radiologists into the examination room with the patient and the primary care physician. Radiologists need to become Sherpas for the images they read, helping interpret and explain results to patients directly, the radiologist told me as we waited out the storm. Otherwise, the field is in real danger, he said. “We’re trying not to be the low-hanging fruit.”
Most public writing about AI coalesces around job loss. The hype of how AI will diagnose, organize and process information faster and more accurately than humans feeds a hysteria about which jobs will it “disrupt.” But analysts at McKinsey & Company argue that a focus on occupations is misguided, that it is far more likely that AI will change jobs rather than eradicate them. “Capabilities such as creativity and sensing emotions are core to the human experience and also difficult to automate,” the authors of a 2015 report note. “The amount of time that workers spend on activities requiring these capabilities, though, appears to be surprisingly low.” A mere 4% of U.S. work requires creativity, apparently, and only 29% requires emotion sensitivity.
By taking away other tasks, the report claimed, AI might free up workers for more meaningful work. This is a common argument. But given the history of U.S. employers driving costs down to offer cheaper products, many employers will probably cut the people they can. Because they share this hunch, some of the more humane Silicon Valley engineers embrace universal basic income proposals; they see the day coming when work itself is a luxury, but they cannot see how to change the systems — the “disruptions,” their innovations — that contribute to that end. After law firms incorporated AI in the “document review” process, for example, the army of first-year law grads that used to do that work was not redeployed, in full complement, to more inspiring tasks.
On the other hand, more plausible than the newly meaningful work that awaits us is the notion that AI and robotics will create new industries and new jobs that we cannot yet predict. “One-third of new jobs created in the United States in the past 25 years did not exist, or barely existed, 25 years ago,” the McKinsey report reminds us. It seems likely that at least some of the new jobs, and the remaining old ones, will involve connective labor. The radiologist’s “patient-facing” solution to the threats AI poses to his job suggests that connective labor is safer than other jobs, that its particularistic and emotional nature might even make caregiving the “last human job.”
A few analysts share the radiologist’s conclusion. As one Harvard Business Review contributor writes: “In the human economy, the most valuable workers will be hired hearts.” “When machines and software control more and more of our lives,” New York Times columnist Thomas Friedman argued in 2017, “people will seek out more human-to-human connections — all the things you can’t download but have to upload the old-fashioned way, one human to another.” Now during the coronavirus crisis, the collective experience of conducting life online has made more people aware of the value of face-to-face relationships in general.
There’s no small irony in the fact that this labor — once given lower pay and less respect precisely because of its emotional, relational component, at times even denigrated as “women’s work” — is now prized as perhaps the last human job. Radiologists are paid almost twice the salary of primary care physicians. Standing under the tent in the rainstorm, it was suddenly clear to me that primary care, whose claim on patient face-time had perhaps been the source of its deprecation compared to more highly paid medical fields, could soon enjoy some real bargaining power over those very same sub-specialties.
But the future of these jobs is by no means certain. In fact, recent history in the U.S. suggests instead that whether or not this job gets done by a person depends not just on what kind of person is slated to do it, but on the value ascribed to the population on the receiving end. One of the fastest-growing set of occupations in the U.S. in the past year has been what economists call “wealth work,” personal services that workers provide for rich people — personal trainers, personal chefs and the like. As the word “personal” tells us, this is work that is customized, often face-to-face, and involves some kind of intimate knowledge, where the worker knows the client, their troubles and their desires. In other words, connective labor. The growth of this kind of work is one of the outcomes of the meteoric rise of inequality and the hollowing out of mid-level jobs in the U.S. and other rich countries.
While the affluent hire humans to be their personal investment advisors or personal shoppers, at the very bottom of the income ladder, disadvantaged people increasingly get connective labor that is automated. Thousands of young American children, including nearly half of Utah’s four-year-olds, for example, are enrolled in “virtual preschool”: online programs that use animation and songs to teach pre-reading and other skills. The engineers who design these programs argue that they are “better than nothing,” and they are not necessarily wrong. But the question remains whether nothing is the best we can do in terms of human connective labor for poor people. Meanwhile, for the harried middle-class comes the proliferation of tech companies that provide online platforms for connective labor on the fly, such as Care.com and UrbanSitter, with disenfranchised workers conducting connective labor as a gig.
In the emerging landscape of this work, then, we are seeing the rise of a servant economy and a thinning out of connective labor by increasing precarity and automation, and these trends are likely to continue post-pandemic. There is inequality between who is on the receiving end of different kinds of connective labor, as well as whose labor is automated. While the well-off meet their personal needs with the help of hired humans, those increasingly underpaid humans will likely have to turn to machines (like virtual preschools) for their own personal needs. Yet the distribution and conduct of connective labor matter not just because of their documented impact on healing and learning or because they are frequently experienced as meaningful work. It can also provide people with a form of dignity that resists measurement and mechanization.
One nurse practitioner I spoke to, for example, carries a powerful memory with her. A few years back, she was working in a Philadelphia homeless shelter when a man came in, his back hunched over from years of osteoporosis, his feet rough and torn from miles of walking. “He had some wounds on his feet — just gnarly, calloused,” she recalled. “And I just sat and did wound care for his feet.”
The moment was not going to change much for him necessarily, she said. He was still going to be on his feet all the time, and he was still reluctant to go to a shelter. “It was just a Band-Aid over a really big problem.” Yet she reckoned that few people ever even looked him in the eyes.
“It was powerful for both of us,” she told me. “Just to give him that moment — I’m seeing you, I’m acknowledging you, this is me caring for you.” She paused. “I think we all just want acknowledgment of our sufferings, even if you can’t cure it or do anything about it.”
The capacity of humans to confer dignity on others just by virtue of seeing them is not an unvarnished good, however. Part of the power of the human audience rests in the fact that humans have the potential to judge others, and with that possibility of judgment comes risk — risk for the student or patient or client who might be judged negatively. Engineers remind us that people are more honest with computers about their struggles, and they argue that technology aids in therapy or teaching could be important tools for assessing needs and weaknesses in clients and students.
But the experience of someone witnessing us, of being seen by another human, is undeniably powerful, and we know this partly by the threat of its absence. The opposite of feeling seen is, of course, feeling invisible, which had its own cultural moment in films like “Parasite” and “Joker.” It is captured in the politics of white working-class rage and implicated in our recent social crises, such as the aptly termed “deaths of despair,” those deaths by suicide and drug and alcohol overuse that radically lowered the life expectancy for working-class whites. It may seem odd to point to invisibility as a problem while workers, from warehouse laborers to truck drivers to gig-economy nannies, were increasingly being monitored, rated and replaced, their work ever more measured and intensified. “We literally have to be careful about everything that we do, even in our personal life,” one careworker told researchers.
But to be surveilled or counted is not to be witnessed. Instead, accompanying these trends is a new stratification of human contact. We are only just beginning to know what connecting labor does and how it works. And we surely have a social interest in making certain kinds of connective labor more systematic, so that getting good teaching or healing is less reliant on being affluent or lucky.
But even before the pandemic, trends have been transforming the conditions of this work without much understanding about what it is and what enables it. The profound impact of seeing and being seen makes connective labor a way we confer dignity, an ongoing demonstration of another person’s worth. After the virus recedes and we start to pick up the pieces of our economy again, and as we embark on a massive campaign on many different fronts to measure, evaluate and scale up connective labor, we should remember that we are tampering with not just a peripheral task, nor one made modest because it occupies the realm of sentiment. Instead, we tinker with nothing less than the dignity that makes us fully human.