Google Maps is making me stupid, about that I have no doubt. After getting out of the subway, I’ll frequently follow my phone’s blue arrow for a block in the wrong direction before the GPS catches up and turns me around. This happens embarrassingly often, despite having lived in New York for years and being surrounded by landmarks like the Empire State Building.
It’s lapses like these that Nicholas Carr warns about in his new book, The Glass Cage, which gives automation the same skeptical treatment that his 2010 book, The Shallows, gave the internet. He paints a scary picture. Planes are crashing as pilots are lulled into a stupor by autopilot. Financial markets flirt with disaster as traders place too much faith in algorithms they barely understand. Doctors are acting like robots themselves as they rotely click through prompts on diagnostic and billing software. And something more ineffable is taking place, Carr worries, as automation subtly cuts us off from the world.
When tasks get automated, people space out
Much of Glass Cage is concerned with flight, where the combination of risk, speed, and tedium made it an early target for automation. Today, autopilot technology has progressed to the point that humans hold the controls for only three minutes or so on a typical passenger flight, and this, the FAA acknowledges, poses some novel problems. When tasks become heavily automated, people space out — "automation complacency," Carr calls it — and their skills grow rusty. Several recent crashes have been attributed to pilots reacting badly when forced to take over the controls in an emergency.
Pilots were ahead of the curve. Now drivers, obeying turn-by-turn directions in self-parking (and soon-to-be self-driving) cars are in a similar position. Carr takes a broad approach to automation, so any technological abbreviation of a task would qualify. Google’s auto-completing searches automates inquiry, Carr says, while legal software automates research, discovery, and even the drafting of contracts. CAD automates architectural sketching. Thanks to an explosion in computing power, more and more things are getting automated, and Carr worries that it’s all combining to degrade our skills and insulate us from the world. "When automation distances us from our work," Carr writes, "when it gets between us and the world, it erases the artistry from our lives."
I think Carr is probably right about much of this, but I have a hard time mustering his concern. I too am nostalgic for the romance of early flight, when pilots were intuitively attuned to their surroundings through the shuddering of their plane’s throttles and levers; but as Carr himself notes, a great many of those early pilots died in crashes, and personally I’m glad my captain isn’t flying by unaided gut feeling. Pilots might be less manually skilled now, but flying is far safer. For my part, I’m probably less engaged with my surroundings because of Google Maps, but it also allows me to explore more new places without getting lost. Every tool, automated or not, opens new possibilities and closes others, fosters new skills and lets others lapse. Most of the problems Carr points to either seem like good trade-offs or fixable shortcomings. He even suggests some possible design solutions, including taking cues from game makers and designing tools that are always slightly challenging to use.
A great many of those early pilots died in plane crashes
What I found more troubling than the existential effects of automation is the specter of unemployment looming over the book. Carr doesn’t focus on it as much, but it’s there as a corollary. If a job is so automated that it’s alienating, it’s also so automated that it requires very little skill — that is, if it hasn’t been automated entirely.
Again, airplanes are a useful preview. Sixty years ago passenger flights were staffed by five well-paid professionals; now there are two, and the increasing automation of their jobs has coincided with a steady decline in wages. Legal discovery has been automated, eliminating many entry level-jobs, and now startups like Lex Machina are tackling more advanced decisions, like court selection and trial strategy. With computers like Watson, it’s only a matter of time before similar fates befall other professions, either automating them entirely or reducing the skill and training required to do them.
This has always been automation’s promise, that it would liberate us from drudgery and free us for higher pursuits. "So while humanity will be amusing itself," wrote Oscar Wilde, "or enjoying cultivated leisure — which, and not labour, is the aim of man — or making beautiful things, or reading beautiful things, or simply contemplating the world with admiration and delight, machinery will be doing all the necessary and unpleasant work." But it may turn out to be a cruel irony of automation that the talents that distinguish humans from machines aren't what we’d hoped. In many cases it’s easier to automate logistics — higher-level reasoning and planning — than seemingly basic roles like driving cars, picking things off shelves, and walking around.
The talents that distinguish humans from machines might not be what we’d hoped
In warehouses, Amazon’s Kiva robots carry shelves to humans, whose job is simply to pick products pointed out by a laser and drop them in a basket. Car services like Uber and Lyft could be seen as cases of stealth automation, with an algorithm directing a group of freelance drivers, the training of whom has been largely obviated by GPS navigation. (Although a London cabbie trained in "The Knowledge," the profession's book of canonical shortcuts and routes, can still beat a GPS-equipped Uber.) The same could be said of many other service startups whose main innovation has been automating the clearinghouses that match freelancers with customers. There are obviously many other forces at work in the slow recovery from the recession, but economists believe that automation is part of the reason why three-quarters of new jobs created have been in low-paying sectors, though most of the jobs lost in the crash were in well-paying industries.
These issues will become only more pressing, and I wish Carr had dealt more with them. Regarding the political implications of automation-driven inequality, he dismisses them quickly, preferring to dwell on the skills and ways of being that automation may be allowing to atrophy.
Maybe the most ominous thing about Carr’s history is how short its purview is — just a hundred years or so. During the Great Depression, John Maynard Keynes wrote that though industrialization would destroy some jobs, it would create new ones. It did — white-collar jobs managing the new logistical requirements of growing companies. But now, as we watch those roles get automated and service jobs spring up in their place, that history starts to look like a disconcertingly small sample. There’s no guarantee that mechanization has to create new jobs to replace the ones it destroyed. This time could be different.