Technology Doesn’t Drive Itself: The Future of Workers in the Automation Era
California seems to have settled on a common understanding of how automation will shape the future of workers’ lives. In this vision, hatched beneath the ever-present sunlight and palm trees, at companies whose workers are always within minutes of one of the most beautiful vistas in America, the future of workers is verdant with sun-streaked optimism: In the future, robots will do the hard work, and we’ll be free to pursue our best, creative selves.
Like many workers in Silicon Valley, I did not grow up here. I grew up in Maine, a state where technology created a very different outcome. Maine is most known as the source of the lobster industry, but already largely forgotten for its paper mills. In Maine, where the sky is streaked in glorious pines, rather than the Salesforce Tower, the shift from paper to screens did not lead to a creative renaissance among creative people; it lead to unemployment.
So, I’m skeptical of San Francisco ambition, which is guided by slogans like this one: “We ignite opportunity by setting the world in motion.”
If you haven’t seen it already, that’s the first line of Uber’s IPO filing in March. It’s spelled out in bright white text, taking up an entire black page, and it’s hard not to feel a little optimistic about all that stark possibility. But as you get deeper into the report, there’s plenty of gray: in particular, in how Uber relies on the classification of its drivers, and how it relates to their eventual replacement by self-driving vehicles.
With news of an upcoming driver’s strike on the horizon, I wanted to look at Uber as a case study for the future of work — to see what we might learn about an industry built on the premise that its workers will one day be replaced by machines.
Freelance vs Freedom
If you haven’t been paying attention, Uber’s IPO filing contains a very direct admission:
“If, as a result of legislation or judicial decisions, we are required to classify Drivers as employees (or as workers or quasi-employees...), we would incur significant additional expenses for compensating Drivers … any such reclassification would require us to fundamentally change our business model.”
Many met this statement with a raised eyebrow. Nitasha Tiku, an editor at Wired, suggested that this understanding of the company’s business model requires them to withhold health insurance, benefits, and pay protections from the drivers who serve as the backbone of the service. Not to pick on Uber: Lyft is, essentially, the same model; meanwhile delivery companies such as DoorDash rely on similar visions of their employees-who-aren’t.
These companies are eyeing a future where these workers are automated. Uber has said 1 million human drivers could be replaced with autonomous vehicles as its long-term strategy. It’s investing heavily in research and development in a race against driver dissatisfaction. This is something of an arms race: you need time to build out the robots, but a company spending billions of dollars on research and development can’t pay drivers proper wages forever.
“The rhetoric of [freelancers] being ‘free’ is false,” said Dr. Florence Chee, Assistant Professor of Digital Communication and Director of the Social & Interactive Media Lab (SIMLab) at Loyola University Chicago. She recently published a paper on the social and ethical issues of ride-share services such as Uber and Lyft, and I wanted to talk to her about what the future of automation looks like if we just roll it forward from what’s happening today.
“There is an enormous underclass emerging of people who have to piece together numerous side hustles,” she told me. “There’s no more ‘one full job’ that pays a living wage.”
This is especially dangerous in a country that ties crucial services, such as health care, to employment. In the case of drivers, who are categorized as “contractors,” health care is a burden placed on the driver.
“There are people who have the narrative that they have a side income, but sometimes people threw their back out, and the only option they have for any income at all is to drive. It’s a stopgap measure for inaccessible health care. It’s not just a side gig. That’s not an ideal system for leisure time, creative inspiration — we’re certainly not building a societal golden age or utopia.”
A recent MIT study by Stephen Zoepf found that the typical driver makes something between $8.55 and $10 (that was revised after a research error showed them making even less). More than half of drivers are earning less than minimum wage, and 8% of drivers actually lose money on the job. And that’s at a time when drivers are essential to the business model — before autonomous vehicles are on the road at all.
Which might not happen as soon as many of us might think. Technology today grants a certain level of autonomy to vehicles, but cities are weird spaces, full of unpredictable activity that can’t be predicted. Research has focused on vision and made great strides, but as Corey Levandowski, a pioneer of the LIDAR tech behind Google’s automated vehicles, said at a recent TechCrunch AI+Robotics conference, driving isn’t a vision problem, it’s a prediction problem. Deciding how to handle the behavior of other drivers requires an emotional intelligence that isn’t easy to code.
So we’re likely to have human drivers behind the wheel for quite some time — and as the pressure to automate meets the realities of automation, it’s likely that these freelancers will feel a squeeze as companies begin to hit a funding wall.
But there’s a deeper social question buried here, and it’s this: we’re looking at companies that are quite openly talking about the replacement of their workforce, which isn’t usually how we talk about automation. Automation has always been about helping workers get the job done — things like anti-lock brakes, power steering, GPS devices to help a driver navigate to a location.
That transition is something we should be paying attention to.
ROBOT, TAKE THE WHEEL
Daren Acemoglu of MIT and Pascual Restrepo of Boston University wrote a paper on the subject of automation’s benefits to the labor market. The gist is this: Traditionally, innovation in automation has been valued as a job creator in the long run. That’s because of an expectation that gains in productivity will ultimately create more tasks for humans.
I’ll use a silly example: if you have a factory where people are making shoes, and suddenly replace them with a robot that can cobble more shoes per minute than a human could in a day, you have such a surplus of shoes that you need even more humans to put them into boxes, ship them, and sell them. People who made the shoes can be taken off the shoe-making line and reassigned to these tasks. They stop working in factories and start working in shoe stores.
The key to this success relies on speeding up productivity so much that you create space for an army of shoe-box-packers, and if (a crucial if) the robots are so speedy that humans can’t keep up with the subsequent support of that product, the company may even hire more humans to match those demands.
But that’s the essential piece of that puzzle: you have to make enough stuff, faster enough, to create new work. If you replace workers with a machine that makes shoes at the same speed, there’s no gains for labor. Acemoglu and Restrepo call this “so-so automation” — you end up replacing workers without creating new human tasks, leading to a decline in labor demand.
Acemoglu and Restrepo looked at places where that was exactly the case, such as Detroit, and concluded that a rise in AI and automation as it has been implemented over the past 20 years “will not be the end of work anytime soon, but the trend towards lower labor share and anemic growth in labor demand will continue, with potentially disastrous consequences for income inequality and social cohesion.”
That’s a far cry from the sun-streaked, halcyon visions of factory workers liberated from rote tasks to master paintbrushes and violins.
“In the automotive industry, automation destroyed factory jobs more than any elements, jobs going off-shore included. That’s a fact,” said Chee. “This is why rhetoric can be so harmful; because it divorces us from the facts of the matter. The way automation is going, it’s to replace humans. With Uber, the aim is self-driving vehicles. The drivers who are giving up their careers to drive for Uber are only in this as a stop-gap measure while the AI sorts itself out.”
In the meantime, the rest of us are benefiting from cheaper rides at rates that can’t sustainably maintain the lives of the people driving us.
“The humans are assisting with troubleshooting of routes, gathering intelligence in terms of machine learning, and we’re all helping that process happen,” said Chee, “[But] our data is being paid for in what we think are discounts. The VC’s are subsidizing these operations, because if we actually paid what things cost, an Uber or Lyft ride would be much more expensive than it is.”
It’s not just having a toll on the drivers. A range of recent studies suggests that ride-share services are increasing traffic and moving people away from public transit. Chee said it’s a common story: as public transport loses funding, it grows less reliable; as it becomes less reliable, people become frustrated and lose interest in funding it. Instead, people turn to private models — and speculative tech — rather than looking at public models with stable, existing technology.
A Regulatory Solution?
The conversation around regulation has given rise to two competing ideas about taxation. Should we tax robots, or not?
“There’s a body of research and policy work beyond me, the conversation about these services becoming public utilities — even Facebook. That would subject them to regulations in the public interest,” Chee said.
In a column for the OECD, the University of Geneva’s Xavier Oberson describes an emerging idea of an “automation tax”, which he describes would be “based on the ratio of a company’s revenues (total sales) to their numbers of employees. The higher the ratio of robots to sales, the higher the tax.” Obserson has been an advocate for creating a specific tax profile for robots.
On the other hand, the Information Technology & Innovation Foundation argues that the productivity slump is precisely why policies such as driver surcharges or a “robot tax” are a problem: they suggest that taxation of automation would hinder innovation just when nations need it most to remain competitive in a global race for AI and automation supremacy. Their report actually calls for tax breaks for robots and automation — calling them “a shot in the arm” for productivity. That’s not at odds with Acemoglu & Restrepo’s perspective — perhaps massive amounts of automation will, in the end, spur productivity gains so broadly that they’ll create new opportunities.
Another challenge: from a legal standpoint, the distinctions between robots and workers are still vague. In one extreme case, Ford dressed employees up to look like empty car seats to resolve a regulatory paradox requiring humans to be behind the wheel of autonomous vehicles.
“The blurring of the lines between machines, robots, and humans means that regulations specifically targeting robots need to be pretty clear about exactly who or what they’re attempting to regulate,” write Mark A. Lemley & Bryan Casey, authors of a fascinating paper on the legal vagaries of defining robot workers, You Might be a Robot. “So too, for that matter, do regulations targeting humans but not robots.”
For Chee, the core of the issue for the short term is to treat drivers as employees, not contractors. But the trend to automation isn’t limited to just the drivers. As coders code robots that code (sorry about that sentence) the scale of displacement could rise.
“Highly educated Silicon Valley workers should know that they are no exception to this; they are writing their own obsolescence. Which is why it’s a problem that touches everyone, and we should make sure that these innovations serve all of us. It’s that whole ‘and then they came for me’ thing. There has to be a spectrum of jobs available for people of all sorts.”
In the meantime, we as consumers should be more critical of the rhetoric surrounding automation, and even the ideology embedded into unsustainable pricing schemes.
“We’re so careful about government surveillance and handing information to the government, but we give it, hand over fist, to these corporations with no mandate for public service or social justice, and we do it uncritically.”
For Chee, this suggests the need for a more empowered multi-stakeholder conversation around the future of work, workers, and automation.
“The only way I see going forward are comparably sized entities that look not only at legalities but a comprehensive sense of ethical good, justice, equity, society and history. I advocate for a triple-helix model — industry, academia and government weighing in with their various capacities,” Chee said.
Ideally, that leads to a radical reimagination of the way we think about “work” and “workers,” and the way we support them, rather than relying on misguided replacements of tasks solely because we can. The centrality of labor in society will need to be reimagined, somehow, if there’s going to be anyone left for the robots to serve.
Written by Eryk Salvaggio
Eryk Salvaggio is a researcher and content strategist at swissnex San Francisco. As a writer for swissnex San Francisco's publication, nextrends, Eryk is focused on insights that emerge at the intersections of science, art, and technology.
The opinions expressed in nextrends are those of the individual authors and interviewees; they do not reflect the position of nextrends or swissnex San Francisco.