During his first day on the job at a small 3D-modelling company, Griffith noticed that his new colleagues’ workstations were hopelessly out of date. So he took the initiative to suggest some automation upgrades to the higher-ups, who concurred. Two years later, 20 employees—some of them good friends—were out of work.
“I feel bad because I knew these people well,” Griffith, which is not his real name, tells me. He says feels responsible because he kicked off the project, and after it was underway, he quickly realized they’d lose their jobs. “But I was powerless to stop it.”
Automation is too often presented as a faceless, monolithic phenomenon—but it’s a human finger that ultimately pulls the trigger. Someone has to initiate the process that automates a task or mechanizes a production line. To write or procure the program that makes a department or a job redundant. And that’s not always an executive, or upper-, or even middle management—in fact, it’s very often not. Sometimes it’s a junior employee, or a developer, even an intern.
Inside many companies, automation doesn’t simply unfold as a top-down imperative. It can stem from random efficiency experiments or pilot programs initiated by employees who don’t always intend for their ideas to cascade into large-scale job loss. In some cases, management will ask a junior staff to spearhead an automation initiative (perhaps, some speculate, to help redirect blame for the job-eliminating policies). When either happens, it can lead to long-term guilt, confusion, and regret on the part of the automator—few people want to delete their friends’ or colleagues’ jobs—and embitterment and anger on the part of the automated.
In a series of interviews with coders, technicians, and engineers who’ve automated their colleagues out of work—or, in one case, been put in a position where they’d have to do so and decided to quit instead—I’ve attempted to produce a snapshot of life on the messy front lines of modern automation. (Some names have been changed to protect the identities of the automators.) We’ve heard plenty of forecasting about the many jobs slated to be erased, and we’ve seen the impacts on the communities that have lost livelihoods at the hands of automation, but we haven’t had many close up looks at how all this unfolds in the office or the factory floor.
So, I was hoping to examine the politics of automation within today’s workplaces, and how the phenomenon unfolds in smaller businesses and in circumstances beyond corporate fiat. To take stock of the personal impacts this strain of automation might unleash. Some automators carry regret and guilt for years; others say they’re only automating bad jobs, or are doing work assigned to them—and expected of them—by management.
So how does automation roll through a workplace in real-time? Who suffers the consequences, and who lives with the responsibility? Sometimes, those who are left feeling guiltiest are the ones made to pull the trigger—enterprising junior employees or staffers with innovative ideas—who see their colleagues get the ax, usually without even seeing any of the profit gains or material benefits the company subsequently enjoys.
Erin Winick was a sophomore mechanical engineering student when she took a summer internship at a southern California tech company. She was enthusiastic about 3D printing, so a manager there soon asked her to use the technology to streamline an older mold-making process. That meant studying up with the man then in charge of the process, whom I”ll call Gary. It soon dawned on both of them that if her project succeeded, Gary would be out of a job.
“I remember explicitly feeling my heart beat faster when we had the initial conversation,” Winick tells me in an email. “It was some nervousness and some guilt for sure. When I first got the project I didn’t realize what it would involve.” Winick would later write about the experience for MIT Technology Review, where she’s now an associate editor: “As he described the process and his role in it, I realized that making molds was Gary’s sole responsibility. He had spent over 30 years perfecting these tools and parts.” This was his life’s work.
At first, Gary was friendly, eager to show a new hand the ropes. After he realized what was happening—well, less so. “Each time we spoke, I was closer to making a working product—and more nervous about telling him how things were going,” Winick writes. “I felt that by doing so, I was letting him know how close he was to losing his job.” But the project moved ahead, and the company said it would retrain Gary to work on the new printers. It turned out he wasn’t much interested in learning a new job three decades into his career, however, and took the news as the latest in a long line of slights from management. “More of the feeling of guilt came about over the summer,” Winick tells me, “as I saw he was unhappy and didn’t want to work in another role.”
After Winick left, she says, the company implemented her system and reassigned Gary. He said the company took an “aggressive” line with him and others in his position. So he quit. Today, he works customer service in another state. Winick is clearly still conflicted over the whole thing, and seems to blame herself for killing Gary’s job. But it is deeply suspect to me that the company, which Winick prefers I don’t name, would put an intern in charge of a project that would lead directly to the elimination of someone’s job. Not because Winick wasn’t capable with the technology, certainly—but because that’s a hell of a burden to lay on a summer intern.
Another mechanical engineer, whom I’ll call Manny, was just one year into his first job when his boss approached him and asked him to automate his and his coworkers’ CAD jobs. It was made clear that his coworkers, some of whom he was friendly with, could lose their jobs upon completion of the project. For the obvious reasons, the ask rubbed him the wrong way.
“The project would likely get some of my colleagues fired,” Manny told me in a Reddit DM, “who were 1st-generation immigrants, who I know would have already left if they could find something better.”
So, he declined—and then quit altogether. “I chose not to do the project mostly in fear of this, yes, though I’m sure the next engineer in line would be happy to complete that project and get it on his resume.”
The looming threat of automation wasn’t merely an efficiency measure, either, but perhaps one of the tools deployed by management to keep employees in line. “It was often a negative place to work, many of my colleagues seemed to be kept in line through constant threat of being fired, even if they were clearly doing a great job and being underpaid,” Manny says.
Part of the psychological impact of the so-called rise of the robots is that work—especially technical work—now suddenly seems more precarious in general. We toil with the looming threat of automation at our backs, a fear management can exploit if it so chooses.
For good reason.
Some 5 million manufacturing jobs have been lost since the turn of the 21st century, and the bulk of those have been lost to automation. Some of those are from closed plants, dramatically announced in headlines and in figures of the tens of thousands. But just as many likely erode away by automation’s attrition, or are inflicted in the winding ways described above. And we’re just getting warmed up.
In perhaps the most cited study on the topic, researchers found that 47 percent of American jobs are still susceptible to automation. A more recent one estimated 800 million jobs worldwide will be wiped away. So there’s a fairly pressing imperative to understand how automation is instigated, how it flows through departments, how it’s coped with by employees—and even how it can be mitigated or better harnessed to yield more equitable returns.
As we watch the rise of the automation engineer—one of the hottest new jobs in the IT field, I’m told—these will be important considerations to foreground.
When engineers and coders are asked to automate their peers’ jobs, or realize that work at their company is ripe for automation, they do what coders do—turn to web forums. On Reddit, in particular, ethical questions about automating coworkers’ jobs away are not uncommon. Manny, for instance, posted his plight to Reddit, where his story elicited an outpouring of support and encouragement, should he have chosen to go through with it. (If he didn’t do it, someone else would, engineering is all about maximizing efficiency, the arguments tended to go.) Others do the same: “Just started learning programming this year, found a way to automate a coworker’s job, need advice,” reads another post.
Automators also share war stories of accidental automation, like this one: TIFU [“today I fucked up”] by getting 6 people laid off, in which a former temp worker tells the tale of automating a few library workstations, only to discover the IT department later took his program system-wide. There are plenty displays of thoughtful automation, too: “We have a lovely lady who does a lot of manual data entry of figures, but it involves essentially rekeying data from one place to another more than once,” one Redditor wrote. “She is coming up to retirement and is already gradually reducing her hours. Rather than put my employer in a position where they would have to make her redundant, I’ll hold back on optimising the workflow and implementing any automation. I did already remove the need for us to employ temp staff during our peak season this year, but that was an easy win.”
I’ve spoken with software engineers, help desk workers, drone technicians—they’ve all been in similar situations. One worked at a company that used manually operated probes to test sugar beets in a laboratory. But it was delicate work, and the probe would often crash.
“I used two ultrasonic transducers to get the exact trailer position and did some math to find the center which was the go-to point for the probe, added a camera as a safety measure too, and a button which the truck driver had to push to ensure that the truck was stationary during probing, and that’s pretty much it without going into too much technical details,” an automation process engineer I’ll call Bob told me. “It works pretty well, didn’t have a single problem with it since being automated.”
It did, however, put a number of seasonal workers out of work. But the probe-driving jobs were impermanent, had to be done in a freezing operating room, and Bob says people hated them anyway. “I don’t feel all that bad as those jobs are redundant and we still have a long way to go in eliminating redundant tasks and automating stuff,” he tells me. “I follow orders, but most of the time it’s up to me to do something, and bosses really appreciate when you remove the human factor from the process; more control, unmeasurably higher reliability, and it’s much cheaper in the long run, all of that saves money which then hopefully shows on my salary.”
And his employers seem to be of a feather with Manny’s. “There’s various techniques to make people want to quit themselves, and management here is known to use them to rid themselves from responsibility, it’s bad, really bad but it saves tons of money…”
These stories all have something in common—they all suggest that the most workplaces aren’t prepared to deal with automation. That it’s often a much more clumsy, awkward process than the ominous news reports and ruthless-sounding CEO dictum would have us believe. Even some offices that aren’t perpetually looking for maximal efficiency gains at any cost wind up face to face with automation. Interns wind up spending the summer side-by-side with the people who they are about to help become redundant. Bosses hide behind junior staffers, who are enlisted to push the big red button themselves.
It’s messy, chaotic, and often painful on just about everyone besides management—imagine that. Of course, it’s the worst for those put out of work. Either way, we may be witnessing the rise of a truly broken system where coworkers are forced to struggle with whether or not to automate each other out of work, competing for a thinning pool of decent jobs as profits flow upstream, where the pool of investment for future automation projects grows.
One important thing I gleaned from these interviews was that that’s true not only at massive corporations, but at small businesses, and the trendlines are apt to move forward in the absence of any strong intent thanks to sheer inertia.
In Griffith’s case, the employees were painting images in 2D and using custom software to translate them into 3D models—a lot of time was spent processing the results. The machines and the process were slow and clunky. Griffith, who’d just started as a junior help desk technician, took the initiative to point this out to a developer; together they installed upgraded software on new machines, along with an application management they had been waiting to implement that automated parts of the process.
In the beginning, everyone was thrilled with the faster machines and more efficient process—the company could start clearing out its pesky backlog. But it wasn’t long before the writing was on the wall. “Management seemed ... like they assumed they’d find work to fill the time and make us more money,” Griffith tells me. “They weren’t creative enough or successful enough in these endeavors, and employees went to per-diem and eventually were let go.” It’s been half a decade since he implemented the program, and the lost jobs still haunt him, he says.
“I honestly don’t know what this means for the workforce as this doesn’t always mean a job will be created while one is made obsolete,” says Griffith, “but I know that people will adapt to work beside new technology for as long as we can.”
But that can be small solace, both to him—the automator—and those he inadvertently drove out of work.
“I still feel bad five years later.”
This story is part of Automaton, an ongoing investigation into the impacts of AI and automation on the human landscape. For tips, feedback, or other ideas about living with the robots, I can be reached at firstname.lastname@example.org.